Banking And Finance Mathematics | Yield Curves, Spot Rates, Forward Rates and the Term Structure of Interest Rates

A yield curve is the mathematical map that connects interest rates to time. It is the bridge from a single bond yield to the full term structure of interest rates: spot rates, zero-coupon rates, discount factors, par yields, forward rates, curve bootstrapping, interpolation, curve shape, steepening, flattening, inversion, roll-down and interest-rate risk. This flagship guide develops that system from first principles and then connects it to Singapore Government Securities, SORA-linked finance, bond valuation, bank asset-liability management and modern fixed-income risk.

For readers searching for yield curve mathematics, term structure of interest rates, spot rate curve, zero curve, discount factor curve, par curve, forward curve, bootstrapping, implied forward rates, bond yield curves, Singapore SGS yield curve, SORA, curve steepening, curve flattening or inverted yield curve, the main idea is simple: one interest rate cannot describe every future date. Financial contracts pay at different times, so a coherent valuation system needs a discount factor or rate for each relevant maturity.

Current professional fixed-income teaching makes the same transition. CFA Institute’s 2026 term-structure material explicitly moves from valuing all cash flows with one yield to recognising that interest rates vary by maturity, with spot, par and forward curves carrying different information. In Singapore, MAS publishes benchmark SGS prices and yields across maturities, giving a real local curve to inspect. The mathematics here is educational and analytical; it is not financial advice.

50-Second Router

• Core idea: a yield curve assigns rates, yields or discount factors across maturities rather than using one rate for every cash flow.
• Spot curve: the set of zero-coupon rates used to discount single payments at specific maturities.
• Discount-factor curve: the most direct valuation representation; each maturity has a factor that converts future cash into present value.
• Par curve: coupon rates that would make hypothetical bonds trade at par for each maturity.
• Forward curve: rates implied today for future intervals by current discount factors.
• Bootstrapping: solve discount factors sequentially from instruments of increasing maturity.
• Curve shape: level, slope and curvature matter; curves can steepen, flatten, invert or twist.
• Risk: one duration number is not enough when different maturities move differently; key-rate and bucketed sensitivities become useful.
• Singapore route: MAS SGS benchmark yields provide observable SGD government-market points; SORA is an overnight benchmark, not the same object as a multi-maturity government yield curve.
• Verification: rebuild instrument prices from the curve and confirm that the curve reprices its calibration instruments within tolerance.

The Central Proposition: Finance Needs a Rate for Every Date

A single interest rate is a useful classroom abstraction. A real cash-flow system is dated. A payment next month, one year from now and thirty years from now do not necessarily share the same market discount rate. The yield curve is the structure that restores the missing time dimension.

This is why a yield curve is more than a chart. It is a valuation engine. Once discount factors are known for future dates, any deterministic cash-flow stream can be valued by multiplying each cash flow by the factor for its date and summing. Bonds, swaps, forwards, loans and many derivatives can then be understood as different arrangements of dated cash flows sitting on top of one term-structure backbone.

Adrian’s useful question is therefore not “what is the interest rate?” but “what is the rate for which date, under which convention, and for which risk class?” That one question prevents a huge class of fixed-income errors.

1. Term structure of interest rates

Term structure of interest rates is the relationship between market interest rates and time to maturity for otherwise comparable claims. It replaces one-rate thinking with a maturity-indexed system. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. A term structure can be written as a function r(T), a set of discount factors D(T), or a family of zero prices P(0,T). In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Treating a ten-year yield as though it were the correct discount rate for a six-month cash flow destroys maturity information. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into bond pricing, swaps, forward rates and bank ALM. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

2. Discount factor

Discount factor is the present value today of one unit of currency received at a specified future date under the chosen curve. Discount factors are the cleanest valuation objects because cash-flow value is simply amount multiplied by factor. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. With annual effective spot rate s_T, D(T)=1/(1+s_T)^T in a simplified annual model. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Confusing a discount rate with a discount factor is a units error: one is a rate, the other is a present-value multiplier. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into zero-coupon bonds, NPV and derivative discounting. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

3. Zero-coupon bond

Zero-coupon bond is a claim that pays one amount at maturity and no intermediate coupon. Its price reveals a maturity-specific discount factor directly when credit and convention are matched. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. If a zero pays F at T and price is P, D(T)=P/F. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Using a coupon-bearing bond yield as a zero rate without decomposition mixes cash flows from several maturities. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into spot rates and bootstrapping. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

4. Spot rate

Spot rate is the zero-coupon interest rate for a cash flow from today to one specified maturity. It provides the maturity-specific rate needed to value a single future payment. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Under annual compounding, D(T)=(1+s_T)^(-T). In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. A spot rate is not the same as a bond YTM; YTM is one internal rate across many cash flows. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into zero curve, forward curve and discounting. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

5. Par yield

Par yield is the coupon rate that makes a hypothetical standard bond price at par for a given maturity under the curve. It is observable-like and intuitive because market bonds are often discussed in coupon/yield terms. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. For annual coupons and par 100, c=(1-D_n)/(D_1+…+D_n) when coupon periods align. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Using a par yield to discount every cash flow as if it were a spot rate causes repricing error when the curve is not flat. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into benchmark curves and swap par rates. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

6. Forward rate

Forward rate is a rate implied today for an interest period that begins in the future. It converts the cross-sectional spot curve into a sequence of future-period break-even rates under no-arbitrage compounding. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. For annual periods, 1+f_{1,1}=(1+s_2)^2/(1+s_1). In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Calling an implied forward rate a forecast confuses a no-arbitrage relation with an expectation about realised future rates. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into FRAs, swaps, curve expectations and carry. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

7. Forward discount factor

Forward discount factor is the ratio of two discount factors that values one future date relative to another. It is often more stable conceptually than manipulating multiple quoted annualised rates. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. The discount factor from T1 to T2 is D(T2)/D(T1). In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Mixing compounding conventions while taking rate ratios can create false forward rates. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into forward valuation and collateralised pricing. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

8. Bootstrapping

Bootstrapping is a recursive procedure that solves discount factors or zero rates from instruments of increasing maturity. It turns observable instrument prices into an internally consistent zero curve. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Known early discount factors are used to strip early cash flows from the next instrument, leaving one new unknown. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Trying to solve all maturities independently can lose the cash-flow relationships that make the curve coherent. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into government bonds, swaps and OIS curves. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

9. Interpolation

Interpolation is a rule for estimating curve values between observed or calibrated maturity points. Markets do not provide a liquid instrument for every possible future date, so a continuous valuation curve must bridge gaps. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Interpolation may be performed on zero rates, log discount factors, forwards or another representation. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. A visually smooth zero-rate curve can imply unstable forward rates; interpolation choice affects derived quantities. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into curve construction and risk. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

10. Extrapolation

Extrapolation is a rule for extending the curve beyond the last reliable market point. Long-dated liabilities and instruments can extend beyond liquid benchmarks. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Methods range from flat forwards to convergence toward long-run anchors depending on use case. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Extrapolated values may look precise while being driven mainly by modelling assumptions. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into insurance, pensions and long-dated valuation. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

11. Curve level

Curve level is the broad height of yields across maturities. Parallel or near-parallel shifts are the simplest first factor of term-structure movement. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. A level factor adds roughly similar basis points to many maturities. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Assuming every real curve move is parallel hides slope and curvature risk. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into duration and macro rate shocks. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

12. Curve slope

Curve slope is the difference between longer- and shorter-maturity yields. Slope describes whether long rates are above, near or below short rates. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. A simple slope measure is y_10-y_2, though many tenor pairs are possible. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. One spread does not capture the entire curve and can miss changes elsewhere. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into steepening, flattening and bank maturity transformation. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

13. Curve curvature

Curve curvature is the bending of middle maturities relative to short and long ends. Many real curve changes are not explained by level and slope alone. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. A stylised curvature factor compares a middle tenor with an average of short and long tenors. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Ignoring curvature can leave a supposedly duration-neutral portfolio exposed. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into PCA, butterfly trades and key-rate risk. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

14. Normal upward-sloping curve

Normal upward-sloping curve is a shape in which longer maturities yield more than shorter maturities over much of the curve. It may reflect expected future short rates, term premia and risk compensation. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. No single formula determines the shape; it emerges from prices across maturities. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Calling an upward curve a guaranteed growth signal overstates what the curve can establish. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into macro interpretation and term premia. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

15. Flat curve

Flat curve is a shape in which yields across selected maturities are similar. It can arise during transitions when short and long rates converge. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Slope spreads approach zero. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. A flat par curve does not mean all spot and forward rates are identical. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into curve decomposition. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

16. Inverted curve

Inverted curve is a shape in which shorter yields exceed longer yields over selected maturities. It is an observable price relationship often discussed in macroeconomic analysis. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Selected long-minus-short spreads become negative. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Treating inversion as a deterministic recession timer converts an empirical indicator into a certainty claim. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into macro finance and expectations. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

17. Steepening

Steepening is an increase in the difference between long and short yields under the chosen slope definition. It tells us the curve changed shape, not merely level. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. A 2s10s spread moving from 20bp to 80bp is a 60bp steepening under that metric. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Steepening can occur with both rates rising, both falling, or opposite-direction moves; name the move precisely. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into bear steepeners and bull steepeners. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

18. Flattening

Flattening is a decrease in the long-short yield spread. It can change portfolio and bank exposures even when average yield changes little. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. A slope spread falling from 80bp to 20bp is a 60bp flattening. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Saying ‘rates fell’ is not enough; one end may have risen while the other fell. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into ALM and curve trades. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

19. Twist

Twist is a non-parallel move in which different maturity regions change in different directions or magnitudes. Twists expose the limits of single-duration hedging. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Key-rate shocks represent one practical decomposition. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. A portfolio can have near-zero aggregate duration yet large twist exposure. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into key-rate duration and bucket hedging. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

20. Carry

Carry is the income and financing effect of holding a position over time under stated assumptions. Carry separates passage-of-time economics from changes in market rates. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. For a bond, coupon accrual and financing costs contribute to carry. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Positive carry is not guaranteed profit because curve moves and spread changes can dominate. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into fixed-income return decomposition. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

21. Roll-down

Roll-down is the valuation effect as an instrument ages and moves to a shorter point on an unchanged curve. A bond can gain or lose value even if the curve is assumed unchanged because it is now priced off a different maturity point. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Reprice the same cash-flow structure one period later on the unchanged curve. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Roll-down assumes a stable curve shape; realised returns differ when the curve moves. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into bond strategy and scenario analysis. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

22. Term premium

Term premium is compensation investors may require for bearing duration and interest-rate uncertainty over longer horizons. It helps explain why forward rates need not equal expected future short rates. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Observed long yields can be thought of as expected path components plus term-premium components in many models. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Treating the decomposition as directly observable ignores model dependence. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into macro-finance models. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

23. Expectations hypothesis

Expectations hypothesis is a family of ideas relating long rates to expected future short rates. It provides intuition for the connection between spot and forward rates. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Under a pure expectations simplification, forwards proxy expected future short rates. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Risk premia and market frictions mean real-world forwards need not be unbiased forecasts. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into term-structure theory. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

24. Liquidity preference

Liquidity preference is the idea that investors may require compensation to hold longer or less liquid claims. It introduces a risk-premium channel into curve shape. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. No single universal equation captures the premium without a model. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Using it as a complete explanation for every curve shape is too strong. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into term premia and market microstructure. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

25. Market segmentation

Market segmentation is the idea that supply and demand can differ by maturity bucket because investor clienteles have preferred horizons. It explains why maturity-specific flows can affect curve points. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Pension demand at long maturities can differ from money-market demand at the short end. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Segmentation is one influence among many, not a standalone predictive law. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into institutional portfolio demand. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

26. Nelson-Siegel family

Nelson-Siegel family is a parametric curve representation using level, slope and curvature-shaped loadings. It compresses many observed maturities into a small number of interpretable parameters. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. The exact form uses exponential-decay basis functions controlled by a decay parameter. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. A good statistical fit does not guarantee arbitrage-free dynamics or correct extrapolation. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into curve fitting and central-bank term-structure analysis. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

27. Svensson extension

Svensson extension is an extension of Nelson-Siegel with additional curvature flexibility. It can fit more complex yield-curve shapes with a second hump component. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. It adds another exponential-decay term and associated parameters. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. More parameters can improve in-sample fit while increasing instability and overfitting risk. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into curve estimation. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

28. Spline curve

Spline curve is a piecewise polynomial representation joined smoothly at knots. Splines can fit local curve structure flexibly without one global functional form. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Cubic splines enforce continuity conditions across segments. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Poor knot choice or unconstrained fitting can produce implausible forward-rate oscillations. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into market curve construction. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

29. Monotone-convex interpolation

Monotone-convex interpolation is an interpolation approach designed to produce well-behaved forward curves from discrete inputs. It addresses the fact that smooth zero rates can still imply problematic forwards. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. The method constrains local behaviour between nodes. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. No interpolation method eliminates data-quality and liquidity problems at the nodes. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into production curve systems. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

30. OIS curve

OIS curve is a curve built from overnight-indexed swap instruments and related products. In collateralised markets it is commonly associated with discounting under specified collateral conventions. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Discount factors are inferred from OIS cash flows across maturities. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. An OIS curve is not automatically interchangeable with a government-bond curve; credit, liquidity and instrument structure differ. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into derivatives discounting. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

31. Government yield curve

Government yield curve is a maturity structure built from sovereign bills and bonds. It is often used as a benchmark but reflects the sovereign market’s own liquidity, supply and tax/convention features. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Benchmark yields are observed from selected securities and can be fitted into a continuous curve. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. On-the-run and off-the-run bonds can trade differently, so raw yields are not a frictionless zero curve. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into SGS and global government markets. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

32. Swap curve

Swap curve is the maturity structure of par fixed rates on interest-rate swaps. It is central to derivatives and bank balance-sheet pricing. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Par swap rates are linked to discount factors and projected floating cash flows. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Modern multi-curve frameworks mean one curve may discount while another projects a reference rate. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into swap valuation. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

33. Multi-curve framework

Multi-curve framework is a system using distinct curves for discounting and for projecting different reference-rate cash flows. It emerged from the recognition that funding, collateral and tenor basis matter. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Each cash flow is projected from an appropriate forward curve and discounted with the relevant discount curve. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Collapsing all curves into one can misprice basis-sensitive instruments. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into post-crisis derivatives valuation. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

34. Basis spread

Basis spread is the spread between rates associated with different tenors, indexes or funding structures. It measures a relative price, not a standalone level. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Basis swaps exchange one floating index for another plus a spread. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. A basis can move because of liquidity, credit, demand or balance-sheet constraints; no single cause is universal. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into multi-curve and cross-currency markets. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

35. SORA

SORA is Singapore Overnight Rate Average, an SGD overnight interest-rate benchmark administered by MAS. It anchors SGD overnight-rate products and compounding conventions but is not itself a long-maturity yield curve. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Term-equivalent cash flows can be built by compounding overnight observations according to contract rules. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Equating today’s overnight rate with a five-year government yield confuses horizon and instrument. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into Singapore loans, swaps and derivatives. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

36. SGS benchmark curve

SGS benchmark curve is the maturity pattern visible across MAS benchmark Singapore Government Securities. It gives Singapore readers a concrete sovereign-market term structure to inspect. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. MAS publishes benchmark maturities from bills through long-dated bonds with closing yields and prices. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Raw benchmark yields are not identical to zero-coupon spot rates because coupon bonds contain multiple cash flows. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into Singapore fixed-income education. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

37. Curve calibration

Curve calibration is the process of choosing curve parameters or nodes so model prices match observed instrument prices. A useful curve should reproduce the market instruments from which it was built. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Calibration minimises or eliminates pricing residuals subject to conventions and model constraints. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. A curve that looks smooth but cannot reprice its instruments is not calibrated for valuation. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into quality control. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

38. Repricing residual

Repricing residual is the difference between observed market price and model price after calibration. Residuals reveal convention errors, interpolation problems, bad data or deliberate smoothing choices. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Residual=ModelPrice−MarketPrice. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Tiny residuals do not prove the model is economically correct outside calibration instruments. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into curve validation. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

39. Bid-ask uncertainty

Bid-ask uncertainty is the fact that market instruments are quoted with spreads rather than one exact frictionless price. Curve inputs therefore contain a feasible region, not perfect truth. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Calibration may target mid-market or fit within bid-ask bounds. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Treating every last decimal as economically meaningful overstates data precision. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into liquidity and model governance. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

40. Stale price

Stale price is a quote or transaction that no longer represents current executable market conditions. Illiquid maturities can contaminate a fitted curve. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Diagnostics compare timestamps, neighbouring nodes and alternative sources. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Smooth interpolation cannot repair fundamentally stale data without judgment. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into market data quality. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

41. On-the-run effect

On-the-run effect is the tendency for the newest benchmark government security to trade differently from older issues. Liquidity and benchmark demand can create yield differences unrelated to pure maturity. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Curve builders may select, adjust or model benchmark issues carefully. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Blindly choosing the lowest-yield bond at a maturity can import liquidity specialness into the curve. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into government-curve construction. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

42. Day-count convention

Day-count convention is the rule converting calendar dates into year fractions. Curve equations use time, and time itself is convention-dependent in finance. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Accrual factor τ determines interest over a period. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Using Actual/365 in one leg and 30/360 in another without recognising it can break parity. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into swaps, bonds and money markets. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

43. Business-day adjustment

Business-day adjustment is the rule shifting contractual dates that fall on weekends or holidays. Exact cash-flow dates affect discounting and accrual. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Following, modified following and preceding are common rule families. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Ignoring date adjustment can cause production systems to disagree with contract cash flows. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into operational fixed income. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

44. Compounding convention

Compounding convention is the rule translating quoted annual rates into accumulation or discount factors. Spot and forward comparisons are meaningful only after putting rates on compatible bases. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Periodic, effective and continuous compounding can represent equivalent economics with different quoted numbers. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Subtracting rates across incompatible conventions produces false spreads. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into curve conversion. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

45. Continuous compounding

Continuous compounding is a representation in which D(T)=exp(-r_c T) for a continuously compounded zero rate. It simplifies some calculus and additive relations across time. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. r_c=−ln D(T)/T. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Continuous rates are a quotation convention, not evidence that money literally compounds every instant. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into quantitative finance. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

46. Instantaneous forward rate

Instantaneous forward rate is the limiting forward rate over an infinitesimally short future interval. It links the derivative of the log discount curve to local future discounting. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. f(0,T)=−d ln D(T)/dT under smooth continuous-time assumptions. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Noisy curve interpolation can make instantaneous forwards extremely unstable. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into term-structure models. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

47. Curve arbitrage check

Curve arbitrage check is a test that fitted discount factors and forwards obey required economic constraints under the chosen market assumptions. Curve construction is not merely statistical smoothing. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Positive discount factors and appropriate monotonicity conditions may be required depending on rate regime and setup. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Overly rigid constraints can be inappropriate in negative-rate environments; assumptions must match the market. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into validation. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

48. Scenario curve

Scenario curve is a shocked yield or discount curve used to revalue a portfolio under hypothetical conditions. It turns qualitative rate views into cash-value consequences. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Apply parallel, slope, curvature or custom tenor shocks and reprice. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Adding the same percentage to every quoted instrument may not create a coherent shocked zero curve. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into stress testing. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

49. Key-rate shock

Key-rate shock is a localised movement centred on a selected maturity while other nodes are held or interpolated. It identifies where curve risk lives. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Reprice after bumping the 2y, 5y, 10y or other node. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Shock interpolation rules influence the resulting sensitivity. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into DV01 ladders and hedging. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

50. Principal components

Principal components is statistical factors that summarise common historical yield-curve movements. Empirical studies often find level, slope and curvature-like factors explain much variation. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. PCA diagonalises the covariance matrix of yield changes. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Historical factors can change across regimes and are not structural laws. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into risk compression. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

51. Curve covariance

Curve covariance is the covariance structure of yield changes across maturity points. It determines whether curve risks diversify or reinforce each other. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. A covariance matrix across tenors supports scenario generation and VaR. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Assuming perfect correlation across the curve exaggerates some risks and hides others. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into portfolio risk. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

52. Curve hedge

Curve hedge is a position designed to offset sensitivity to specified yield-curve movements. Hedging is matching risk-factor sensitivities, not merely matching maturity labels. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. A hedge ratio may use DV01 or key-rate DV01 vectors. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. A single bond hedge can neutralise total DV01 while leaving large shape risk. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into fixed-income risk management. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

53. Curve carry and roll attribution

Curve carry and roll attribution is a decomposition separating income, financing, roll-down and market-move effects. It explains why a bond return can differ from its starting yield. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Compute expected passage-of-time value on an unchanged curve, then isolate actual curve changes. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Attribution is convention-dependent; define the counterfactual curve clearly. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into performance analysis. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

54. Curve inversion and lending economics

Curve inversion and lending economics is the interaction between short funding costs and longer asset yields. Banks transform maturities, so curve shape can affect margins, but actual outcomes depend on deposit betas, hedges and asset repricing. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. A simple spread proxy is long asset yield minus short funding rate. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Using 2s10s alone as a bank-profit formula ignores balance-sheet structure. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into bank ALM. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

55. Curve shape and mortgages

Curve shape and mortgages is the link between benchmark rates, funding curves and mortgage pricing. Mortgage rates reflect more than one government yield point. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Pricing can combine reference rates, funding costs, credit, capital and margin. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Treating a mortgage rate as a direct mechanical function of one SGS yield oversimplifies bank pricing. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into Singapore household finance. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

56. Curve shape and corporate finance

Curve shape and corporate finance is the effect of maturity-specific discount rates on financing and investment valuation. A long project and a short project face different benchmark horizons. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Term-specific discounting can alter present values and hurdle rates. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. One WACC for every horizon can hide term-structure effects. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into DCF and debt issuance. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

57. Curve shape and derivatives

Curve shape and derivatives is the role of discount and forward curves in valuing swaps, forwards and options. Derivatives make curve consistency especially visible because replication relationships span dates. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Forward prices and swap rates are functions of discount factors and projected reference rates. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Mixing curves across collateral conventions can create artificial arbitrage. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into derivatives mathematics. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

58. Curve governance

Curve governance is the controls around market data, curve methodology, approvals, testing and change management. A production curve is shared infrastructure: a small error can propagate into thousands of valuations. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Governance includes data-source hierarchy, calibration tolerance, independent validation and versioning. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. A mathematically elegant curve without reproducibility and audit trails is operationally weak. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into bank model risk. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

59. Versioned curve snapshot

Versioned curve snapshot is a stored set of market inputs, conventions and calibrated outputs for a specific timestamp. It makes historical valuation reproducible. The strongest way to learn it is to keep the economic object and the mathematical representation side by side rather than memorising the representation alone.

Mathematical form. Store timestamp, instruments, quotes, conventions, parameters and resulting nodes. In every calculation, label the time unit, compounding convention, currency and instrument class. A curve quantity is meaningful only when those conventions travel with the number. If the problem uses six-month periods, the rate or discount factor must refer to the same interval or be converted consistently.

Failure mode. Saving only final zero rates can make later diagnosis impossible. Jo’s diagnostic is to ask whether the mistake changes the cash flow, the date, the discount factor or merely the quote. That classification quickly separates contract errors from rate-convention errors and from market-data errors.

Connection. The concept feeds directly into audit and P&L explain. Ryan would verify it by rebuilding at least one cash flow from first principles and checking direction under a small rate change. The objective is not simply to obtain a number but to create a valuation chain another reader can audit.

Worked Lab 1: Turn Discount Factors Into Spot Rates

Suppose the one-, two- and three-year discount factors are 0.980392, 0.951814 and 0.915142. Under annual effective compounding, solve s_1=1/D_1−1, s_2=D_2^(-1/2)−1 and s_3=D_3^(-1/3)−1. The exponents matter because each spot rate applies from today all the way to its maturity.

Now reverse the calculation. Rebuild each discount factor from the recovered rate. If the reconstructed values do not match the original factors within rounding tolerance, the rate conversion is wrong. This two-way test is one of the simplest curve validation tools.

Clara then asks a qualitative question: are discount factors decreasing with maturity in this positive-rate example? They should be. A later factor larger than an earlier one would require explanation, such as a negative-rate environment or inconsistent inputs.

Worked Lab 2: Bootstrap a Two-Year Zero Rate

Assume a one-year zero-coupon instrument implies D_1=0.970874. A two-year annual-pay bond with face 100, coupon 4 and price 100 has cash flows 4 at year 1 and 104 at year 2. The par pricing equation is 100=4D_1+104D_2. Solve D_2=(100−4D_1)/104.

Once D_2 is known, convert it to a two-year spot rate using s_2=D_2^(-1/2)−1. The logic is recursive: the one-year discount factor prices the first coupon, leaving one unknown factor attached to the final cash flow. That is the essence of bootstrapping.

Aisha verifies the answer by substituting both factors back into the bond equation. If the model does not return 100, do not continue bootstrapping the curve. An early residual will contaminate every later node.

Worked Lab 3: Derive a Forward Rate

Let the one-year spot rate be 2.0% effective and the two-year spot rate be 2.8% effective. The one-year forward rate beginning one year from now satisfies (1.028)^2=(1.02)(1+f_1,1). Therefore f_1,1=(1.028)^2/1.02−1.

The calculation is a no-arbitrage compounding identity in the simplified model. Investing two years directly must match the compounded value of investing one year and locking the second-year rate today, assuming comparable credit, liquidity and transaction conditions.

Do not rename the result “the market forecast for next year’s one-year rate”. The forward contains current price information and potentially risk premia. It is an implied break-even rate under the valuation framework.

Worked Lab 4: Price a Coupon Bond With the Spot Curve

A three-year bond pays 3 at years 1 and 2, then 103 at year 3. Suppose the annual effective spot rates are 2.0%, 2.5% and 3.0%. Price = 3/1.02 + 3/(1.025)^2 + 103/(1.03)^3.

Compare this with discounting all cash flows at a single 2.7% YTM. The two prices may be close, but they answer different modelling questions. The spot curve preserves maturity-specific discounting; the YTM compresses the resulting price into one internal rate.

Ben’s test is to alter only the three-year spot rate. Only the final cash flow’s discount factor should change in the direct spot-curve valuation. If year-one present value moves, the spreadsheet references are wrong.

Worked Lab 5: Build a Par Rate From Discount Factors

Suppose annual discount factors are D_1=0.98, D_2=0.95 and D_3=0.91. A three-year bond with face 100 trades at par when 100=cD_1+cD_2+(100+c)D_3. Rearranging gives c=100(1−D_3)/(D_1+D_2+D_3).

The coupon rate c/100 is the three-year par yield under these simplified annual conventions. Notice that it is built from the entire set of discount factors up to maturity, not just the three-year spot rate.

This explains why par, spot and forward curves are related but numerically different. They are transformations of the same underlying discount structure, each answering a different question.

Worked Lab 6: Measure Steepening Precisely

Suppose the 2-year yield moves from 2.20% to 2.60% while the 10-year yield moves from 2.80% to 3.60%. The 2s10s spread changes from 60 basis points to 100 basis points. The curve steepened by 40 basis points even though both yields rose.

Because both ends rose, this would commonly be described as a bear steepening in market shorthand. If both yields had fallen while the long end fell less, the curve could steepen in a bull move. The label should never replace the actual arithmetic.

Mira writes the start spread, end spread and change before using any verbal label. That keeps interpretation anchored to measurable quantities.

Worked Lab 7: Key-Rate Sensitivity

Imagine a portfolio whose value changes by −S$1,500 when the 2-year node is bumped up 1bp, −S$4,000 for a 5-year node bump and −S$7,500 for a 10-year node bump under the chosen interpolation rule. Its bucketed risk vector is therefore concentrated toward the 10-year region.

A single total DV01 of roughly S$13,000 per basis point hides that shape. Two portfolios can share the same total DV01 but have very different key-rate vectors. They will behave differently under twists and butterfly moves.

This is the mathematical reason curve risk management becomes vector-valued rather than scalar.

Worked Lab 8: Carry and Roll on an Unchanged Curve

Take a bond today and value it from the current curve. Advance the valuation date by one month while assuming the entire curve is otherwise unchanged. Remove any cash flow paid during the month, accrue coupon according to convention, and reprice the remaining cash flows at their now-shorter maturities.

The difference between the new model value plus received cash and the old value reflects carry and roll under that counterfactual. Actual realised P&L then differs because the curve, spreads and other factors moved.

This decomposition is valuable because it separates the return you might expect from passage of time under a frozen market from the return generated by market changes.

Singapore Lab: Read the MAS SGS Benchmark Table

MAS publishes daily SGS benchmark data across Treasury-bill and bond maturities, including 6-month, 1-year, 2-year, 5-year, 10-year, 15-year, 20-year, 30-year and longer benchmark points when available. Each row contains instrument details and closing yield information. This is a real Singapore sovereign term-structure surface, not a textbook illustration.

A reader can take a single observation date, record the benchmark yields, order them by maturity and draw the curve. Then calculate simple slope spreads such as 10-year minus 2-year and 30-year minus 5-year. Next, compare the raw benchmark-yield curve with the idea of a zero curve: the government benchmark bonds are coupon-bearing, so their yields are not automatically zero rates.

That distinction is the doorway to proper curve construction. Market benchmark yields are inputs and summaries. Discount factors are the valuation engine.

SORA and the Yield Curve Are Related but Different

SORA is an overnight benchmark. A multi-year SGD yield curve contains information across future horizons. Contracts can use compounded overnight rates over an accrual period, and derivatives markets can imply expectations and risk premia about future overnight-rate paths, but today’s SORA fixing is not itself a five-year or ten-year rate.

This is the same time-indexing discipline used throughout this lane. An overnight rate answers an overnight question. A five-year discount factor answers a five-year present-value question. A five-year par swap rate summarises a sequence of projected and discounted cash flows. These quantities interact but should not be collapsed into one number.

For Singapore mortgages and corporate funding, the reference-rate mechanics, lender funding curve, credit, capital, liquidity and margin can all matter. The reader should resist single-rate explanations of multi-component prices.

Curve Shapes Do Not Have One Universal Economic Meaning

A curve shape is an observed price relationship. Explanations involve expectations for future short rates, term premia, central-bank policy, inflation, supply, demand, liquidity, hedging flows and institutional balance sheets. Different combinations can produce similar shapes.

This is why a yield curve should be used as evidence, not as an oracle. A historical association between inversion and later slowdowns can be investigated statistically, but the mathematical curve itself does not contain a deterministic economic future. It contains prices and implied break-even relations under current market conditions.

Good financial writing distinguishes what is observed from what is inferred.

Curve Construction Is an Inverse Problem

Market instruments produce prices. Curve builders seek the hidden discount function that explains those prices. This is an inverse problem because the object we want—the continuous discount curve—is not directly observed at every date. We infer it from a finite and noisy collection of instrument prices.

Inverse problems need regularisation, interpolation and judgment. Fit too loosely and the curve fails to reproduce the market. Fit too tightly and it may chase noise, producing unstable forward rates. The best construction depends on the use case: trading, valuation, risk, accounting or long-horizon liability work may prioritise different properties.

This is one reason “the yield curve” is not a single universally unique object. Methodology matters.

The Curve as Shared Financial Infrastructure

Inside a bank, the same family of curves can feed valuation, transfer pricing, treasury, derivatives, collateral, risk, product pricing and P&L explain. An error in one node or convention can therefore propagate across many systems. Curve quality is infrastructure quality.

That is why professional curve systems need timestamped market data, convention libraries, calibration controls, fallback hierarchies, audit trails and independent validation. The mathematics is necessary but insufficient without operational discipline.

Ethan’s rule is useful here: if a valuation cannot be reproduced from a stored curve snapshot and its input data, the organisation has a control problem even if today’s number looks plausible.

A Curve-Building Workflow

1. Choose the economic curve purpose: sovereign benchmark, OIS discounting, projection, funding, credit or another role.
2. Define eligible instruments and market-data hierarchy.
3. Normalise calendars, day counts, compounding and quote conventions.
4. Convert observed quotes into instrument pricing equations.
5. Solve the shortest reliable discount factor or curve node.
6. Bootstrap or calibrate later nodes in maturity order.
7. Interpolate between nodes using a method appropriate to the chosen representation.
8. Apply any required extrapolation beyond the liquid horizon.
9. Reprice every calibration instrument and inspect residuals.
10. Inspect discount factors, zero rates and forward rates for economic and numerical pathologies.
11. Store the curve snapshot, inputs, methodology version and diagnostics.
12. Run sensitivity and scenario tests before the curve is admitted to valuation or risk.

A Curve-Reading Workflow for Students and Parents

1. Read maturity before yield.
2. Check whether the instrument is a bill, coupon bond, swap or another product.
3. Identify whether the number is a spot rate, YTM, par rate, forward rate or overnight reference rate.
4. Convert percentages to consistent decimal and compounding conventions before calculation.
5. Draw the curve instead of memorising labels.
6. Calculate at least one slope spread explicitly.
7. Translate one pair of spot rates into an implied forward rate.
8. Price one cash flow using the appropriate maturity-specific discount factor.
9. Explain one way the model can fail or become incomplete.
10. Verify the result by reversing the calculation.

Common Failure Modes

1. Using YTM as a zero rate

A coupon bond YTM summarises many cash flows. A zero rate belongs to one maturity. Strip or bootstrap rather than relabel. The repair is to return to the underlying cash-flow equation, identify which curve object belongs to which date, and verify the result independently.

2. Treating forward rates as forecasts

Forwards are implied break-even rates under current prices and conventions; expected future rates can differ because of premia and shocks. The repair is to return to the underlying cash-flow equation, identify which curve object belongs to which date, and verify the result independently.

3. Mixing government and swap curves

They contain different instrument, credit, liquidity and collateral features. Name the benchmark before using it. The repair is to return to the underlying cash-flow equation, identify which curve object belongs to which date, and verify the result independently.

4. Ignoring quote conventions

A yield, discount rate, money-market quote and continuously compounded zero rate can describe different conventions. The repair is to return to the underlying cash-flow equation, identify which curve object belongs to which date, and verify the result independently.

5. Overfitting sparse data

A curve can pass through every noisy quote and still produce unusable forwards. The repair is to return to the underlying cash-flow equation, identify which curve object belongs to which date, and verify the result independently.

6. Underfitting real structure

Excessive smoothing can erase economically meaningful changes and create repricing residuals. The repair is to return to the underlying cash-flow equation, identify which curve object belongs to which date, and verify the result independently.

7. Ignoring bid-ask spreads

A single mid price hides the feasible market range. The repair is to return to the underlying cash-flow equation, identify which curve object belongs to which date, and verify the result independently.

8. Using stale nodes

One old quote can bend a curve and contaminate many valuations. The repair is to return to the underlying cash-flow equation, identify which curve object belongs to which date, and verify the result independently.

9. Assuming parallel moves

Real curves change level, slope and curvature. The repair is to return to the underlying cash-flow equation, identify which curve object belongs to which date, and verify the result independently.

10. Ignoring cross-tenor covariance

Tenor risks are related but not perfectly correlated. The repair is to return to the underlying cash-flow equation, identify which curve object belongs to which date, and verify the result independently.

11. Forgetting calendars

Exact payment dates alter accrual and discounting. The repair is to return to the underlying cash-flow equation, identify which curve object belongs to which date, and verify the result independently.

12. No read-back verification

A curve that cannot reproduce its source instruments is not ready. The repair is to return to the underlying cash-flow equation, identify which curve object belongs to which date, and verify the result independently.

Formula Map

Object	Simplified formula	Use
Discount factor	D(T)=1/(1+s_T)^T	Present value of one unit paid at T under annual effective compounding.
Spot rate from factor	s_T=D(T)^(-1/T)-1	Convert a discount factor into a zero rate.
Forward factor	D(T2)/D(T1)	Discount between two future dates.
1y forward	1+f=(1+s_2)^2/(1+s_1)	Implied year-2 rate in a simple annual model.
Par coupon	c=F(1-D_n)/(ΣD_t)	Coupon amount that prices a standard bond at par.
PV from curve	PV=Σ CF_t D(t)	General deterministic cash-flow valuation.
Continuous zero	r_c(T)=−ln D(T)/T	Continuously compounded zero rate.
Instantaneous forward	f(0,T)=−d ln D(T)/dT	Local forward rate under smooth continuous-time assumptions.

Authoritative Reference Map

• CFA Institute 2026 — The Term Structure of Interest Rates: Spot, Par, and Forward Curves
• CFA Institute 2026 — The Term Structure and Interest Rate Dynamics
• Monetary Authority of Singapore — Daily SGS Prices and Benchmark Yields
• Bukit Timah Tutor — How Yield-Curve Algorithms Build the Term Structure

Connected Banking And Finance Mathematics Route

• Complete Banking And Finance Mathematics System
• Time Value of Money
• Interest Rates
• Annuities and Cash-Flow Mathematics
• Bonds and Bond Pricing
• Finance & Banking Algorithms

Final Principle

A yield curve is not a decorative graph of rates. It is a dated valuation system. Spot rates convert specific future cash flows into present value. Par rates summarise coupon structures. Forward rates encode break-even relations between future intervals. Discount factors are the common language beneath them.

Once that architecture is clear, curve shape becomes easier to reason about. Steepening and flattening are measurable changes in maturity relationships. Key-rate sensitivity becomes a natural response to non-parallel movement. Bootstrapping becomes a recursive cash-flow exercise rather than a mysterious market technique.

The durable workflow is: identify dates, identify instruments, normalise conventions, solve discount factors, derive rate representations, reprice the market, inspect risk, and preserve the assumptions. That is how the term structure becomes usable mathematics.

Applied Case Study 1: Funding a five-year school campus lease

Situation. A five-year stream of contractual rent payments should be valued against date-specific discount factors rather than one undifferentiated rate. The exercise teaches why a liability is a vector of dated cash flows and why a change in the two-year curve point can matter differently from a change in the five-year point. This is a deliberately bounded educational case: the purpose is to expose the mathematical dependency rather than imitate a complete production pricing system.

Method. Map each annual payment, apply D(1) through D(5), sum PV, then bump one node at a time and record which payments respond. Adrian would put the dates first; Jo would normalise the quote conventions; Aisha would ask which assumption is doing the most work. Those three checks make the model easier to audit.

Verification. Recalculate the case after changing one curve input by a small amount. The cash flows attached to unaffected maturities should behave consistently with the interpolation rule. Reverse any rate conversion to recover the original discount factor. Then compare the sign of the value change with economic intuition.

Boundary. Using one average rate is acceptable only as an explicit approximation; hiding the approximation prevents meaningful sensitivity analysis. The reader should be able to state what the model knows, what it assumes and what lies outside its scope.

Applied Case Study 2: Comparing two SGD fixed deposits

Situation. A six-month and a two-year deposit live at different horizons. Their quoted rates should be converted to comparable effective-return conventions before interpretation, but a yield curve still tells us that horizon itself is part of the price. This is a deliberately bounded educational case: the purpose is to expose the mathematical dependency rather than imitate a complete production pricing system.

Method. Convert each quote to an effective period return, annualise consistently where appropriate, then place it beside comparable market horizons without claiming identical credit or liquidity. Adrian would put the dates first; Jo would normalise the quote conventions; Aisha would ask which assumption is doing the most work. Those three checks make the model easier to audit.

Verification. Recalculate the case after changing one curve input by a small amount. The cash flows attached to unaffected maturities should behave consistently with the interpolation rule. Reverse any rate conversion to recover the original discount factor. Then compare the sign of the value change with economic intuition.

Boundary. The higher quoted rate is not automatically the better choice; tenor, withdrawal terms, credit and personal liquidity needs differ. The reader should be able to state what the model knows, what it assumes and what lies outside its scope.

Applied Case Study 3: Pricing a three-year corporate note

Situation. Begin with the government or chosen benchmark term structure, then recognise that the corporate note may require additional spread for credit and liquidity. This is a deliberately bounded educational case: the purpose is to expose the mathematical dependency rather than imitate a complete production pricing system.

Method. Discount each cash flow using benchmark zero rates plus a spread model, then solve the spread that reconciles model price with market price. Adrian would put the dates first; Jo would normalise the quote conventions; Aisha would ask which assumption is doing the most work. Those three checks make the model easier to audit.

Verification. Recalculate the case after changing one curve input by a small amount. The cash flows attached to unaffected maturities should behave consistently with the interpolation rule. Reverse any rate conversion to recover the original discount factor. Then compare the sign of the value change with economic intuition.

Boundary. A constant spread is a model convenience, not proof that risk compensation is uniform across dates. The reader should be able to state what the model knows, what it assumes and what lies outside its scope.

Applied Case Study 4: A bank treasury duration gap

Situation. A bank owns assets and funds them with liabilities that reprice at different horizons. One scalar duration can hide which parts of the curve drive the mismatch. This is a deliberately bounded educational case: the purpose is to expose the mathematical dependency rather than imitate a complete production pricing system.

Method. Build asset and liability key-rate vectors, net them by tenor, and examine parallel, steepener and flattener scenarios. Adrian would put the dates first; Jo would normalise the quote conventions; Aisha would ask which assumption is doing the most work. Those three checks make the model easier to audit.

Verification. Recalculate the case after changing one curve input by a small amount. The cash flows attached to unaffected maturities should behave consistently with the interpolation rule. Reverse any rate conversion to recover the original discount factor. Then compare the sign of the value change with economic intuition.

Boundary. A position can be close to DV01-neutral while still having material slope or curvature exposure. The reader should be able to state what the model knows, what it assumes and what lies outside its scope.

Applied Case Study 5: A Singapore mortgage reference-rate discussion

Situation. A household may hear SORA, fixed rates, board rates and government yields in the same conversation. They belong to different contracts and horizons. This is a deliberately bounded educational case: the purpose is to expose the mathematical dependency rather than imitate a complete production pricing system.

Method. Separate the overnight reference, the bank’s funding and hedging curve, contractual spread and reset rule before comparing rates. Adrian would put the dates first; Jo would normalise the quote conventions; Aisha would ask which assumption is doing the most work. Those three checks make the model easier to audit.

Verification. Recalculate the case after changing one curve input by a small amount. The cash flows attached to unaffected maturities should behave consistently with the interpolation rule. Reverse any rate conversion to recover the original discount factor. Then compare the sign of the value change with economic intuition.

Boundary. Treating one public benchmark as the complete mortgage-pricing formula is an explanatory shortcut, not a full model. The reader should be able to state what the model knows, what it assumes and what lies outside its scope.

Applied Case Study 6: A long-dated endowment liability

Situation. An institution promising payments decades ahead may need discount factors beyond the most liquid market points. This is a deliberately bounded educational case: the purpose is to expose the mathematical dependency rather than imitate a complete production pricing system.

Method. Use liquid nodes where available, then state an extrapolation method and test how alternative long-end assumptions change present value. Adrian would put the dates first; Jo would normalise the quote conventions; Aisha would ask which assumption is doing the most work. Those three checks make the model easier to audit.

Verification. Recalculate the case after changing one curve input by a small amount. The cash flows attached to unaffected maturities should behave consistently with the interpolation rule. Reverse any rate conversion to recover the original discount factor. Then compare the sign of the value change with economic intuition.

Boundary. Long-horizon precision can be model-driven; report sensitivity rather than false certainty. The reader should be able to state what the model knows, what it assumes and what lies outside its scope.

Applied Case Study 7: A swap fixed rate

Situation. A par swap rate is the fixed rate that makes the swap’s present value zero at inception under the chosen projection and discount framework. This is a deliberately bounded educational case: the purpose is to expose the mathematical dependency rather than imitate a complete production pricing system.

Method. Project floating cash flows, discount both legs, and solve the fixed coupon that equalises present values. Adrian would put the dates first; Jo would normalise the quote conventions; Aisha would ask which assumption is doing the most work. Those three checks make the model easier to audit.

Verification. Recalculate the case after changing one curve input by a small amount. The cash flows attached to unaffected maturities should behave consistently with the interpolation rule. Reverse any rate conversion to recover the original discount factor. Then compare the sign of the value change with economic intuition.

Boundary. In a multi-curve system, the projection curve and discount curve may differ; one-curve formulas are approximations. The reader should be able to state what the model knows, what it assumes and what lies outside its scope.

Applied Case Study 8: A forward-starting loan

Situation. A borrower wants to understand a rate fixed today for borrowing beginning next year. This is a deliberately bounded educational case: the purpose is to expose the mathematical dependency rather than imitate a complete production pricing system.

Method. Derive the break-even forward rate from discount factors spanning today-to-start and today-to-end. Adrian would put the dates first; Jo would normalise the quote conventions; Aisha would ask which assumption is doing the most work. Those three checks make the model easier to audit.

Verification. Recalculate the case after changing one curve input by a small amount. The cash flows attached to unaffected maturities should behave consistently with the interpolation rule. Reverse any rate conversion to recover the original discount factor. Then compare the sign of the value change with economic intuition.

Boundary. The implied forward rate is not a promise that the future spot borrowing rate will equal it. The reader should be able to state what the model knows, what it assumes and what lies outside its scope.

Applied Case Study 9: A curve stress test

Situation. Risk managers want the P&L effect of a 100bp parallel rise, a 50bp steepening and a butterfly shock. This is a deliberately bounded educational case: the purpose is to expose the mathematical dependency rather than imitate a complete production pricing system.

Method. Construct coherent shocked curves, revalue every cash flow or instrument, and compare full repricing with sensitivity approximations. Adrian would put the dates first; Jo would normalise the quote conventions; Aisha would ask which assumption is doing the most work. Those three checks make the model easier to audit.

Verification. Recalculate the case after changing one curve input by a small amount. The cash flows attached to unaffected maturities should behave consistently with the interpolation rule. Reverse any rate conversion to recover the original discount factor. Then compare the sign of the value change with economic intuition.

Boundary. Shocking quoted par yields directly without rebuilding the zero curve can create inconsistent scenarios. The reader should be able to state what the model knows, what it assumes and what lies outside its scope.

Applied Case Study 10: A bond roll-down study

Situation. A bond with four years remaining will have three years remaining one year later if it survives and no optional event occurs. This is a deliberately bounded educational case: the purpose is to expose the mathematical dependency rather than imitate a complete production pricing system.

Method. Reprice the aged bond on an unchanged curve and separate coupon, carry, roll and curve-move effects. Adrian would put the dates first; Jo would normalise the quote conventions; Aisha would ask which assumption is doing the most work. Those three checks make the model easier to audit.

Verification. Recalculate the case after changing one curve input by a small amount. The cash flows attached to unaffected maturities should behave consistently with the interpolation rule. Reverse any rate conversion to recover the original discount factor. Then compare the sign of the value change with economic intuition.

Boundary. The unchanged-curve counterfactual is a model for attribution, not a forecast. The reader should be able to state what the model knows, what it assumes and what lies outside its scope.

Applied Case Study 11: Curve data quality incident

Situation. One benchmark quote is stale after a market closure or feed interruption. This is a deliberately bounded educational case: the purpose is to expose the mathematical dependency rather than imitate a complete production pricing system.

Method. Compare timestamps and neighbouring nodes, apply the data-source fallback hierarchy, and record the substitution. Adrian would put the dates first; Jo would normalise the quote conventions; Aisha would ask which assumption is doing the most work. Those three checks make the model easier to audit.

Verification. Recalculate the case after changing one curve input by a small amount. The cash flows attached to unaffected maturities should behave consistently with the interpolation rule. Reverse any rate conversion to recover the original discount factor. Then compare the sign of the value change with economic intuition.

Boundary. Silently interpolating over bad data may keep systems running but destroys auditability. The reader should be able to state what the model knows, what it assumes and what lies outside its scope.

Applied Case Study 12: SGS benchmark comparison

Situation. MAS lists benchmark prices and yields at several maturities. This is a deliberately bounded educational case: the purpose is to expose the mathematical dependency rather than imitate a complete production pricing system.

Method. Plot the observed benchmark yield points, calculate selected spreads, then explain why a bootstrapped zero curve would not numerically equal the raw coupon-bond yield curve. Adrian would put the dates first; Jo would normalise the quote conventions; Aisha would ask which assumption is doing the most work. Those three checks make the model easier to audit.

Verification. Recalculate the case after changing one curve input by a small amount. The cash flows attached to unaffected maturities should behave consistently with the interpolation rule. Reverse any rate conversion to recover the original discount factor. Then compare the sign of the value change with economic intuition.

Boundary. The exercise teaches the difference between market display data and valuation state variables. The reader should be able to state what the model knows, what it assumes and what lies outside its scope.

Deep Practice Lab 1: Rebuild the curve from prices

Choose three maturity points and write the exact cash-flow equations that convert quoted instruments into discount factors. Do not begin with a chart. Begin with the contracts. Solve the earliest factor first, carry it forward as known information, and show where the next unknown enters. Then reconstruct every input price from the solved factors.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 2: Change the representation, not the economics

Take one discount factor and express it as an annual effective zero rate, a continuously compounded zero rate and a zero-coupon price per 100. Confirm that all three representations reproduce the same present value. This exercise teaches that quotations can change while the underlying economics remains invariant.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 3: Test curve shape sensitivity

Hold the average level roughly constant while raising a long-end node and lowering a short-end node. Revalue a short cash flow, a long cash flow and a mixed bond. Record which values react most. The goal is to feel the difference between level risk and shape risk numerically.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 4: Audit an interpolation choice

Create two neighbouring curve nodes and interpolate a midpoint using a simple zero-rate interpolation and a log-discount-factor interpolation. Compare the resulting midpoint forward behaviour. Neither miniature example proves one method universally superior; it shows why the representation being interpolated matters.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 5: Separate observation from interpretation

Write two columns. In the first, record facts visible in a yield-curve snapshot: maturities, yields, slopes and changes. In the second, list possible explanations such as policy expectations, inflation, term premia or supply. This prevents an observed price pattern from silently turning into a causal claim.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 6: Rebuild the curve from prices

Choose three maturity points and write the exact cash-flow equations that convert quoted instruments into discount factors. Do not begin with a chart. Begin with the contracts. Solve the earliest factor first, carry it forward as known information, and show where the next unknown enters. Then reconstruct every input price from the solved factors.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 7: Change the representation, not the economics

Take one discount factor and express it as an annual effective zero rate, a continuously compounded zero rate and a zero-coupon price per 100. Confirm that all three representations reproduce the same present value. This exercise teaches that quotations can change while the underlying economics remains invariant.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 8: Test curve shape sensitivity

Hold the average level roughly constant while raising a long-end node and lowering a short-end node. Revalue a short cash flow, a long cash flow and a mixed bond. Record which values react most. The goal is to feel the difference between level risk and shape risk numerically.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 9: Audit an interpolation choice

Create two neighbouring curve nodes and interpolate a midpoint using a simple zero-rate interpolation and a log-discount-factor interpolation. Compare the resulting midpoint forward behaviour. Neither miniature example proves one method universally superior; it shows why the representation being interpolated matters.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 10: Separate observation from interpretation

Write two columns. In the first, record facts visible in a yield-curve snapshot: maturities, yields, slopes and changes. In the second, list possible explanations such as policy expectations, inflation, term premia or supply. This prevents an observed price pattern from silently turning into a causal claim.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 11: Rebuild the curve from prices

Choose three maturity points and write the exact cash-flow equations that convert quoted instruments into discount factors. Do not begin with a chart. Begin with the contracts. Solve the earliest factor first, carry it forward as known information, and show where the next unknown enters. Then reconstruct every input price from the solved factors.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 12: Change the representation, not the economics

Take one discount factor and express it as an annual effective zero rate, a continuously compounded zero rate and a zero-coupon price per 100. Confirm that all three representations reproduce the same present value. This exercise teaches that quotations can change while the underlying economics remains invariant.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 13: Test curve shape sensitivity

Hold the average level roughly constant while raising a long-end node and lowering a short-end node. Revalue a short cash flow, a long cash flow and a mixed bond. Record which values react most. The goal is to feel the difference between level risk and shape risk numerically.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 14: Audit an interpolation choice

Create two neighbouring curve nodes and interpolate a midpoint using a simple zero-rate interpolation and a log-discount-factor interpolation. Compare the resulting midpoint forward behaviour. Neither miniature example proves one method universally superior; it shows why the representation being interpolated matters.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 15: Separate observation from interpretation

Write two columns. In the first, record facts visible in a yield-curve snapshot: maturities, yields, slopes and changes. In the second, list possible explanations such as policy expectations, inflation, term premia or supply. This prevents an observed price pattern from silently turning into a causal claim.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 16: Rebuild the curve from prices

Choose three maturity points and write the exact cash-flow equations that convert quoted instruments into discount factors. Do not begin with a chart. Begin with the contracts. Solve the earliest factor first, carry it forward as known information, and show where the next unknown enters. Then reconstruct every input price from the solved factors.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 17: Change the representation, not the economics

Take one discount factor and express it as an annual effective zero rate, a continuously compounded zero rate and a zero-coupon price per 100. Confirm that all three representations reproduce the same present value. This exercise teaches that quotations can change while the underlying economics remains invariant.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 18: Test curve shape sensitivity

Hold the average level roughly constant while raising a long-end node and lowering a short-end node. Revalue a short cash flow, a long cash flow and a mixed bond. Record which values react most. The goal is to feel the difference between level risk and shape risk numerically.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 19: Audit an interpolation choice

Create two neighbouring curve nodes and interpolate a midpoint using a simple zero-rate interpolation and a log-discount-factor interpolation. Compare the resulting midpoint forward behaviour. Neither miniature example proves one method universally superior; it shows why the representation being interpolated matters.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 20: Separate observation from interpretation

Write two columns. In the first, record facts visible in a yield-curve snapshot: maturities, yields, slopes and changes. In the second, list possible explanations such as policy expectations, inflation, term premia or supply. This prevents an observed price pattern from silently turning into a causal claim.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 21: Rebuild the curve from prices

Choose three maturity points and write the exact cash-flow equations that convert quoted instruments into discount factors. Do not begin with a chart. Begin with the contracts. Solve the earliest factor first, carry it forward as known information, and show where the next unknown enters. Then reconstruct every input price from the solved factors.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.

Deep Practice Lab 22: Change the representation, not the economics

Take one discount factor and express it as an annual effective zero rate, a continuously compounded zero rate and a zero-coupon price per 100. Confirm that all three representations reproduce the same present value. This exercise teaches that quotations can change while the underlying economics remains invariant.

Complete the lab twice: once with a deliberately simplified annual-compounding model and once with the actual quote convention of a real instrument you can document. Compare the answers and identify exactly which convention caused the difference. Ben should be able to reproduce the arithmetic; Mira should be able to explain the economic meaning; Clara should be able to identify the data source and timestamp.

Finish with a falsification check. State one result that would tell you the spreadsheet, curve or interpretation is wrong. Examples include failure to reprice a calibration bond, a discount factor inconsistent with the stated rate, or a forward rate that cannot be reconstructed from the adjacent factors under the same convention.