Sales & Trading · Delta One Financing

Cross-Asset Financing

Swap unwinds, bilateral novations, and the calculation and expectation of borrow cost — across Interest Rates, FX, Commodities, Credit and Equities.

Generated 2026-07-28 from the EquitySwap warehouse.

Contents

Section 1

The carry identity

One equation. Five asset classes. The same trade in five costumes.

F = S · exp[ ( r + u − y − b ) · T ]

r funding · u storage · y yield accruing to the holder · b cost of borrowing the asset

The borrow column is the point

Equity borrow fee, repo specialness, commodity convenience yield, cross-currency basis and CDS-cash basis are not five phenomena. Each is the market price of a scarce balance sheet standing between a synthetic and a physical position. Read the borrow column down: when one widens, the others usually are too.

A desk that hedges its equity financing without watching the cross-currency basis is hedging one leg of the same trade.

Section 2

The curve stack — OIS discounting underneath everything

Post-LIBOR, every collateralised derivative discounts on the OIS curve of its collateral currency. Unwind PV, novation fee, cross-currency basis and CVA all consume discount factors from the same object, so an error here biases all four the same way — the hardest kind to notice.

Bootstrapped by solving S · Σ τᵢ · DFᵢ = 1 − DFₙ at each pillar. The test suite asserts the curve reprices its own input quotes to within 1e-6 — if it does not, every PV downstream is wrong.

Reference rates

SOFR − EFFR: 1.0bp — secured minus unsecured. The cleanest free read on whether funding stress is about collateral scarcity or about credit.

Section 3

Cross-asset rate interplay

Covered interest parity says a fully hedged foreign investment must earn the domestic rate. It does not hold, and the gap is a direct charge on a cross-currency financing book.

F = S · (1 + r_dom·τ) / (1 + r_for·τ)

How the five markets connect

Section 4

Swap unwinds — an unwind is not the mark

Early termination of an equity TRS has four components, and the third is the one naive implementations omit: the present value of the off-market spread over the remaining life.

Termination value, decomposed

Why the sign matters

As the equity-leg receiver you pay the spread. A contracted spread below market is therefore an asset: terminating surrenders 36bp a year for the residual life, and you must be compensated. Getting this backwards quotes the unwind in the wrong direction.

Model notes

• 26.38% withholding applied (DE); 244,487 lost to tax
• contracted spread 42bp is 36bp below market (78bp); annuity 1.470
• hedge is 30.1% of ADV → 36.8bp impact

Observed unwind activity

DTCC records the lifecycle in two independent fields — Action type (NEWT/MODI/TERM) and Event type (TRAD/NOVA/ETRM). A novation prints as Action=TERM, Event=NOVA, so classifying on Action alone collapses every novation into the unwind bucket and reports zero.

Capped notionals are right-censored (DTCC reports large block sizes as “N+”), so size distributions must not be averaged naively.

Section 5

Bilateral novations

Three roles: the Transferor leaves, the Transferee steps in, and the Remaining Party stays — and must consent, because their credit exposure just changed. The fee is the mark plus everything that changes when the counterparty does.

Fee decomposition

Novate, or unwind and rebook?

The commercial question. Unwinding and rebooking crosses the bid-offer twice and prints two trades. A novation crosses it once.

• Novation crosses the bid-offer once: 18bp
• Unwind + rebook crosses it twice: 36bp
• Replacement spread 78bp over 1.50y, annuity 1.432

Pricing notes

• CVA Transferor — Fund A 2,327 (65bp) → Transferee — Fund B 430,701 (280bp)
• IM funding differs by -43,542 (uncleared transferee)
• EUR is not eligible collateral under the transferee CSA (eligible: USD) — FX gap charge applied

On liquid names the gap is small. On wide, illiquid or hard-to-borrow underliers it dominates — which is precisely when a client asks.

Section 6

Borrow cost — four estimators, reconciled

The two quoting conventions

A negative rebate — the fee exceeding the benchmark, so the borrower pays rather than earns on their cash — is the definitional marker of a hard-to-borrow special.

Estimator reconciliation

Options put-call parity carries the most weight because the borrow is embedded in listed option prices whether anyone is looking or not: a market maker short the call and long the put is synthetically short the stock and must borrow it. C − P = S·e^−(q+b)T − K·e^−rT, solved for b.

What it costs on a real ticket

Section 7

Borrow expectation

Borrow behaves like none of the standard processes on its own: it mean-reverts as supply responds to price, jumps on dateable events, and has a fat right tail when shorts cannot exit.

Current fee: 520bp

Event calendar — where the jumps come from

Index rebalances, dividend record dates (dividend-arbitrage demand), M&A, lockup expiries and convertible issuance all produce predictable, dateable spikes. This is what makes borrow forecastable in a way rates are not.

The squeeze condition

High utilisation and high days-to-cover — both, multiplicatively. A crowded short that can exit in a day is not a squeeze; a moderate short that needs two weeks to cover is. The tail is modelled asymmetrically because fees can multiply in a squeeze but cannot fall below the GC floor.

Forward borrow

Where a term borrow market exists, the forward rate follows the same no-arbitrage logic as a forward interest rate — the market's own expectation, and the cleanest possible input to this section.

Section 8 · Required

Data provenance and known limits

What is observed, what is inferred, and what you should not claim.

Warehouse coverage

Limits you must state when presenting this

Generated by equityswap deck on 2026-07-28. Methodology: docs/METHODOLOGY.md. Sources and quotas: docs/DATA_SOURCES.md. Redistribution terms: docs/DATA_LICENSING.md.