The Back Door of an Exit

Unwinding a book on CFFEX index options — strike by strike, then the futures leg

“Better is the end of a thing than its beginning.”
— Ecclesiastes 7:8

Textbooks treat unwinding as bookkeeping: long a contract, sell it back; short one, buy it back. On paper the book closes. In trading, closing each line exactly as it sits is the expensive way through — especially the in-the-money lines, whose spreads are widest precisely where the position is largest.

This note is about that craft on China’s CFFEX index options — IO on CSI 300, MO on CSI 1000, HO on SSE 50. The setting makes the problem concrete. The options are European and cash-settled, so there is no early-exercise lever: before expiry, the only exit is trading. And the exchange lists IF/IM/IH index futures on the same underlyings, sharing the same third-Friday expiry, with deep books and narrow spreads. (The ETF options of the stock exchanges play a different endgame — physical delivery, exercise, next-day shares — catalogued in Note 03; this note keeps to the index products.)

The unit of risk is the strike, not the contract

At any strike, a call and a put carry the same non-linear profile: identical vega, identical gamma, identical skew and kurtosis exposure. The only difference between them is pure linear delta — and linear delta is exactly what index futures absorb at near-zero cost.

So define the strike position: calls plus puts held at a strike. When the strike position is zero, the non-linear risk at that strike is dead, regardless of which mix of calls and puts realizes the zero. The craft of unwinding is to drive every strike position to zero while executing only in liquid quotes — and to hand whatever linear risk the route leaves behind to the futures market.

A worked example

Consider a three-strike book on CSI 1000 index options, October expiry:

• short 3 × 7000 puts
• long 2 × 7200 puts
• long 1 × 7300 call

The aggregates read CallPos +1, PutPos −1. The strike positions read 7000 → −3, 7200 → +2, 7300 → +1. The goal is zeros across the board — strike positions and aggregates alike (they are linked: strike positions always sum to CallPos + PutPos). And any other contract mix carrying these same numbers is this book, up to synthetic futures — the strike and aggregate figures, not the contract list, are the book’s identity.

Here is the risk ladder as the desk sees it — each side’s model value (theo) against its executable quote, index at ≈ 7,126 by parity:

Step 1 — kill the deepest strike. The 7000 line needs +3. Parity says three calls and three puts do the same non-linear job; they differ only by a synthetic future’s worth of delta. So the choice of side is free — spend it on execution quality: compare each side’s executable price against its model value. Buying the 7000 call pays the offer at 172.4 against a theo of 170.92 — 1.48 over. Buying the 7000 put pays 45.8 against 44.84 — only 0.96 over. Buy 3 × 7000 puts: the same risk extinguished for two-thirds of the crossing cost, and the deep in-the-money call spread is never touched. (The put is the cheap side because the index sits above the strike — the out-of-the-money side is where the tight quotes live.)

Step 2 — sweep the aggregates with the liquid legs. The book now reads strikes 0 / +2 / +1, CallPos +1, PutPos +2. What remains are three near-the-money lines, and the bids are close enough to theo to take: the 7300 call at 29.6 against 30.50 (0.90 under), the 7200 puts at 132.2 against 133.32 (1.12 under each). Sell the 7300 call (1 lot), sell the 7200 puts (2 lots).

Three clips, every one executed on the side of its market that quotes nearest the model value. The deep in-the-money exposure was never closed directly — it was converted away by parity first, then closed where quotes are tight. Total crossing cost: about six index points across the whole unwind, without a single touch on the ladder’s widest spread. That is the whole trick: do not flatten the position list; flatten the risk structure, and the position list takes care of itself.

The futures leg

What about delta? Choosing puts over calls at 7000 shifted the route by three synthetic futures, and every strike’s side-selection contributes its own linear residue. Aggregate those across the book and the required futures offset each other: long deltas here cancel short deltas there, and what survives is one net clip in IF/IM/IH — a fraction of the gross hedge, executed where spreads are thinnest. In the worked example above the leg is a no-op: the three clips carry net delta of about +0.2 into a book that started essentially flat, so nothing survives to hedge. Because the futures share the options’ expiry, the leg converges into the same third-Friday settlement that closes the options themselves.

Expiry closes what you let reach it

At expiry, in-the-money index options settle in cash against the delivery settlement price — an average of the underlying over a window on delivery day, per the exchange’s convention — so the final mark does not hinge on one closing print. Letting settlement close a book is legitimate, but it means carrying spread and pin exposure to the last session. The back door is cheaper when used early: extinguish the non-linear risk while the OTM quotes are still tight, and let expiry be a formality rather than a trade.

The market offers two ways out. The front door crosses every spread you happen to be standing on. The back door uses parity to choose which spreads you stand on at all: extinguish risk strike by strike, execute only where quotes are narrow, and hand the linear remainder to the futures. On CFFEX index options the back door is structural — same underlyings, same calendar, cash settlement at the end of the road. Experienced traders take it every single time.