resets, unwinds, novations
What happens to a swap between the day it's traded and the day it ends, and one finding from the public trade tape.
concept noteA reset is settling the monthly rent and checking the meter. An unwind is ending the lease early, possibly with a break fee. A novation is handing the lease to a new tenant. The landlord has to agree, because they're now relying on someone else to pay.
what happens after the trade
| Event | What happens | What gets calculated | Why it matters |
|---|---|---|---|
| Reset | the period's equity move, dividends and funding are paid; the notional is re-struck | compounded rate over the period, equity return since the last reset | keeps credit exposure small between the two sides |
| Dividend | the dividend (often net of withholding) is paid to the long | gross or net amount, and on which date it counts | tax treatment differs by market and investor |
| Corporate action | splits, rights issues and mergers adjust the share count or basket | an adjustment that keeps the position economically unchanged | gets it wrong and the hedge no longer matches the swap |
| Partial / full unwind | the client closes some or all of the position early | accrued value, plus the off-market spread value if locked, plus hedge unwind costs | the dealer sells its hedge shares, with impact and stamp duty |
| Novation | one party steps out and a new one steps in; the trade continues | changes in credit and funding adjustments (CVA, FVA) between the old and new party | saves re-trading, but needs everyone's consent |
the sign mistake that's easy to make
As the long, you pay the spread. So a contract spread below today's market is an asset to you (you're paying less than you would now), and giving it up early deserves compensation. A contract spread above market is a liability you pay to exit. Flip the sign and you quote the unwind with money flowing the wrong way.
The textbook case for novation is sound: unwinding and re-trading with a new dealer crosses the bid-offer twice, while novating crosses it once. So I expected novations to be common.
When I counted lifecycle events on the public DTCC tape for single-name equity swaps, novations were well under 1% of terminations in every sample. In practice getting three parties to consent costs more than crossing the spread twice. Finding them also meant reading two DTCC fields, not one: a novation is reported as a termination with a separate novation event flag, so counting on the first field alone shows none at all.
- Resets settle the P&L each period and keep credit exposure small.
- An unwind is priced on the accrued value, any locked off-market spread, and the hedge's exit cost.
- Novation is elegant in theory but rare in single-name equity swaps: consent is expensive.