Reader question: A repo finances a security and normally returns collateral later. A TBA dollar roll also creates financing-like economics, but the investor can sell agency MBS for one settlement month and buy back substantially similar, not necessarily identical MBS for a later month. How does the algorithm turn those two TBA prices into an implied financing signal?
The core structure is a paired forward trade:
sell front-month TBA + buy back-month TBA
with the same broad TBA characteristics—agency/program, maturity/term and coupon—but different settlement months. The front price and back price produce a drop. The economic return from that drop must then be compared with the coupon and principal cash flows the roll seller gives up, the exact number of financing days, expected prepayments/paydowns, settlement conventions and the value of being able to deliver or receive different eligible pools.
That final step is why a dollar roll is not just “price difference divided by days.” A proper implied-financing calculation is a dated cash-flow internal-rate-of-return problem.
What this page owns — and what it does not
This page owns:
front-month TBA + back-month TBA + expected intervening MBS cash flows + settlement timing → drop → roll economics → implied-financing/specialness diagnostics.
It does not replace agency-MBS cash-flow modelling, mortgage OAS, repo pricing, or mortgage-pipeline hedging. Those are adjacent owners.
This is public mortgage-market mathematics, not a recommendation to enter a dollar roll or trade agency MBS.
First: what is a TBA trade?
The To-Be-Announced market is a forward market for agency mortgage-backed securities. At trade time, the parties agree standardized characteristics such as agency/program, coupon, maturity class and settlement month rather than identifying every specific mortgage pool immediately.
Specific pools are allocated later under market notification and good-delivery rules.
SIFMA publishes the industry’s TBA governance, notification dates and settlement calendars. The standardized contract makes large volumes of agency MBS economically fungible enough to trade before exact pool identities are known.
Why this creates a delivery option
Not all eligible pools are economically identical.
Two 30-year 5% agency pools can differ in:
- loan ages;
- borrower rates;
- geography;
- loan size;
- prepayment incentives;
- pool factors;
- expected cash-flow timing.
The seller can satisfy a TBA obligation with pools meeting good-delivery requirements, which creates a cheapest-to-deliver-style delivery option within the standardized contract.
That option is central to dollar-roll economics because the pool delivered in the back month need not be the exact pool sold in the front month.
The dollar roll
The New York Fed describes an agency-MBS dollar roll as a transaction that generally involves a purchase or sale of agency MBS for delivery in one month together with a simultaneous agreement to resell or repurchase substantially similar securities on a specified future date.
For a common sale-and-repurchase roll:
- sell front-month TBA;
- simultaneously buy the same broad TBA contract for the next settlement month.
The roll seller temporarily gives up MBS exposure to specific pools and later receives eligible replacement pools.
The drop
Define:
Drop = Front-Month Price − Back-Month Price.
If front month trades at 100-16 and back month at 100-08, the drop is 8/32 of a price point, or 0.25 points per 100 of face under ordinary Treasury-style fraction conversion.
A positive drop means the roll seller sells at a higher price and buys back at a lower price.
That looks like a financing benefit—but it is not free money because the seller gives up the MBS remittance economics during the interval.
Why the drop exists
Several effects can contribute:
- coupon interest that accrues to the holder of the security during the roll period;
- scheduled principal;
- prepayments and resulting paydowns;
- time value of money / short-term financing;
- expected changes in pool factors;
- delivery optionality;
- scarcity or abundance of good-delivery pools;
- balance-sheet and settlement demand.
The back-month TBA therefore does not need to equal the front price minus an ordinary money-market financing charge.
Cash-flow view: the safest way to compute implied financing
Instead of memorising one fragile shortcut, build the exact roll cash flows.
For a roll seller, a stylised timeline contains:
- front settlement: receive proceeds from selling the front-month TBA;
- between settlements: forego the coupon/principal remittance associated with holding the MBS;
- back settlement: pay to purchase back-month eligible MBS;
- pool/factor effects: account for the contractual settlement amounts and expected principal factors under the relevant market conventions.
The implied financing rate is the rate r that makes those dated cash flows economically equivalent to borrowing/lending over the roll interval.
Conceptually:
NPV(roll cash flows discounted at r) = 0.
This formulation is robust because it forces the analyst to account for the actual settlement and remittance dates.
A simplified financing illustration
Suppose, purely for teaching:
- front sale proceeds = 100.50;
- back repurchase cost = 100.25;
- the seller gives up 0.20 of expected net coupon/principal economic value during the roll interval;
- roll interval = 30 days.
Net financing-like benefit before detailed settlement adjustments:
100.50 − 100.25 − 0.20 = 0.05.
Relative to roughly 100 of proceeds, a simple annualized approximation is:
0.05 / 100 × 360/30 = 0.60%.
A production calculation must use the exact MBS remittance, factor, settlement and day-count conventions rather than this simplified illustration.
Why coupon carry cannot be ignored
If the dollar roll seller no longer owns the MBS across the relevant record/remittance period, it gives up the associated coupon economics.
A large positive drop can therefore merely compensate for foregone carry.
Comparing the drop alone with a repo rate is incomplete.
Paydowns are unusually important in MBS
Agency MBS principal changes because mortgages amortise and borrowers prepay.
Between front and back settlement, the pool factor of a specific MBS can decline. The TBA back-month trade is for eligible securities at the later settlement state, not a promise that the original front-month pool comes back with unchanged principal.
This creates a key difference from simple financing of a fixed-principal Treasury security.
Prepayment assumptions change implied financing
Suppose mortgage rates fall sharply and prepayments are expected to accelerate.
The amount and timing of principal remitted during the roll interval can change materially.
The exact dollar-roll economics therefore depend on a prepayment/cash-flow model. This connects directly to the existing CPR/SMM and pool-factor page.
Dollar rolls versus repo
A repo is financing against collateral, with contractual repurchase mechanics and collateral treatment.
A dollar roll is two TBA transactions. The back-month securities are substantially similar but need not be the identical pools.
This matters because:
- pool identity can change;
- prepayment quality can change;
- the delivery option has value;
- principal factors can evolve;
- TBA settlement conventions determine delivery.
A dollar roll can behave like secured financing economically while remaining legally and operationally different from repo.
Specialness: compare implied financing with ordinary short-term funding
One useful diagnostic compares the roll’s implied financing rate with a general short-term financing rate.
If:
Implied Roll Financing < General Funding Rate,
the front-month TBA can be described as trading “special” in a financing sense: market participants are willing to accept unusually low implied financing returns to obtain or retain settlement exposure to that contract/pool supply.
The New York Fed has explicitly used low implied financing rates as a signal of scarcity in agency-MBS settlement supply.
New York Fed operational logic makes the signal concrete
In its agency-MBS operation FAQs, the New York Fed has explained that selling dollar rolls can postpone settlement of outstanding purchases, while buying rolls can bring settlement forward.
It has also stated that notably low implied financing rates can signal a shortage of securities available for settlement, while higher rates can signal more abundant supply.
This shows that the roll rate is not only a financing statistic. It is also a market-functioning signal.
Why scarcity affects the back month
Suppose many investors need front-month 30-year agency 5% TBA securities for settlement but few eligible pools are available.
The front-month contract can become expensive relative to the next month.
The drop widens, reducing the implied financing rate after carry is considered.
The roll becomes “special” because the immediate settlement asset is scarce.
Good-delivery rules are part of the algorithm
A back-month TBA purchase is valuable only if delivered pools satisfy the agreed market standards.
SIFMA’s Uniform Practices specify good-delivery requirements across agency MBS and publish notification/settlement schedules by class.
An implied-financing engine must therefore use the correct:
- agency/program;
- coupon;
- maturity class;
- settlement class/date;
- good-delivery eligibility.
Otherwise it is comparing two instruments that are not truly the same TBA contract across months.
The notification date matters
Specific pools are identified shortly before settlement under the TBA notification process.
Until pool allocation, the holder knows the standardized contract characteristics but not the final pool identities.
That uncertainty is part of the delivery option.
Current 2026 settlement calendars are explicit data
SIFMA publishes annual MBS notification and settlement dates. For 2026, the calendar specifies separate Class A, B, C and D notification and settlement dates by month.
A production roll system should never generate these dates by “add one month” arithmetic alone. The official settlement class and calendar matter.
Dollar-roll seller faces redelivery quality risk
The seller of the front month may later receive pools in the back month with less desirable prepayment characteristics than the pools sold.
This is sometimes described as adverse-selection or redelivery risk.
The roll drop can partly compensate for giving the counterparty this delivery option.
Counterexample: a bigger drop is not automatically a better financing return
Suppose Roll A has a 12/32 drop but also gives up a large coupon/principal remittance. Roll B has an 8/32 drop but gives up almost no remittance.
Roll A can have a lower implied financing rate despite the bigger headline drop.
The correct unit of comparison is the full cash-flow IRR.
Counterexample: same coupon does not mean same pool value
Two pools can both be 30-year agency 5% securities and satisfy TBA delivery rules while having different expected prepayment speeds.
The TBA contract intentionally abstracts from some pool-specific differences. Those differences reappear economically through specified-pool pay-ups and dollar-roll delivery value.
Counterexample: cheap implied financing can be caused by scarcity, not cheap credit
If the roll implied rate is extremely low, it does not mean banks suddenly obtain universally cheap unsecured funding.
It can mean one specific TBA settlement asset is scarce and investors are willing to sacrifice financing return to secure it.
Counterexample: a dollar roll can change mortgage exposure quality
Because the back-month delivery can contain different pools, the investor can emerge with the same broad TBA coupon exposure but different prepayment characteristics.
A dollar roll preserves standardized TBA exposure, not exact pool identity.
Inputs and outputs
A TBA dollar-roll engine can require:
- agency/program;
- TBA term/maturity class;
- coupon;
- front settlement month/date;
- back settlement month/date;
- front and back TBA prices;
- trade face amount;
- pool-factor/settlement conventions;
- expected coupon and principal remittances;
- prepayment assumptions;
- good-delivery rules;
- reference short-term funding rate.
Outputs can include:
- drop in price points/32nds;
- expected foregone remittance;
- dated roll cash flows;
- implied financing rate;
- spread versus general financing;
- specialness/abundance diagnostic;
- delivery-quality sensitivity;
- settlement exception flags.
Evidence polarity: what supports confidence?
Evidence for a reliable roll calculation includes front/back contracts matching on required TBA characteristics, official SIFMA settlement dates, correctly parsed prices, cash flows reconciled to agency-MBS factor/remittance data, implied financing reproduced by an independent cash-flow IRR and plausible sensitivity to prepayment assumptions.
Evidence against confidence includes one month using a different coupon or agency, a drop calculated from mismatched settlement classes, repo-style assumptions that force return of the identical pool, paydowns ignored, or an implied rate that does not change when the intervening MBS cash-flow assumption changes materially.
Weak links in implementation
price-format error. 32nds are converted incorrectly.
settlement-date error. Generic month arithmetic replaces the SIFMA class calendar.
cash-flow omission. Coupon or principal remittance is ignored.
factor error. Settlement amounts use stale pool factors.
repo analogy overreach. The engine assumes identical collateral returns.
prepayment-model drift. Roll economics use outdated CPR assumptions after mortgage-rate changes.
delivery-option omission. Redelivery quality is treated as irrelevant.
funding benchmark mismatch. Implied roll financing is compared with an inconsistent tenor or day-count rate.
Diagnostics: how to test the engine
- same-contract test: front and back must match agency/program, term and coupon.
- drop sign test: front 100-16 and back 100-08 should produce +8/32.
- cash-flow IRR test: recompute implied financing from the complete dated cash flows independently.
- zero-remittance teaching test: with no intervening MBS cash flow, the financing return should reduce toward the annualized price-difference economics.
- prepayment sensitivity test: raise CPR and observe the impact of expected paydowns/remittances.
- settlement-calendar test: use SIFMA’s published 2026 Class A/B/C/D dates.
- delivery-quality test: change expected back-month pool characteristics and measure roll value impact.
- specialness test: compare implied financing to a consistent short-term reference rate.
- factor test: update principal factors and require settlement amounts to change consistently.
- NY Fed logic test: extremely low implied financing should be capable of flagging settlement scarcity rather than being interpreted as generic cheap funding.
What would falsify confidence?
Confidence should be withdrawn if the implied financing rate cannot be reproduced from dated cash flows; if the calculation ignores MBS principal/coupon remittances; if the two TBA legs are not contractually comparable; if settlement dates do not match industry calendars; or if roll economics remain unchanged under materially different prepayment assumptions.
Alternatives and limits
Repo is a clearer instrument when the objective is financing a specific security or collateral position and receiving the contractual collateral back. Dollar rolls are native to the TBA agency-MBS market and combine financing-like economics with delivery optionality and mortgage cash-flow risk.
The implied rate is a modelled comparison metric, not a guaranteed future return. Pool delivery, factors, prepayments, transaction costs and settlement outcomes can differ from assumptions.
How this connects to the surrounding knowledge estate
Agency-MBS cash-flow algorithms supply the remittance and paydown layer. Mortgage OAS values the broader prepayment option. Repo pricing provides the financing benchmark and specialness analogy. Mortgage-pipeline hedging explains why TBA contracts matter upstream to mortgage originators.
Verification and update triggers
Preserve SIFMA settlement calendar/version, TBA contract identity, front/back prices, pool-factor source, expected remittance model, prepayment assumptions, reference funding rate and cash-flow dates. Revalidate after material mortgage-rate moves, SIFMA good-delivery changes, agency-program changes, large factor revisions or unexplained roll-price dislocations.
Primary and high-quality references
- Federal Reserve Bank of New York, FAQs: Agency MBS Operations.
- Federal Reserve Bank of New York, Agency Mortgage-Backed Securities.
- Federal Reserve Bank of New York, Agency MBS Reinvestment Purchases and Treasury Rollovers FAQ, including the link between implied financing rates and settlement scarcity.
- SIFMA, TBA Market Governance and Uniform Practices.
- SIFMA, 2026 MBS Notification and Settlement Dates.
- Federal Reserve Bank of New York, Agency MBS Operations.
Educational boundary: This article explains agency-MBS roll mechanics and implied-financing mathematics. It does not recommend a dollar roll, TBA position, repo trade or mortgage investment and does not provide personalized financial advice.