Best-executed hedges this week
FORECAST — expected outcome, not realised return. Performance policyGeneral and impersonal investment research. Not personalised investment advice. Forecasts, valuations and model outputs are estimates and may be wrong. See Investment Disclaimer.
What this list is. Ranked on confidence-adjusted expected value against the market's own price, the frequency of the downside, probability of profit and execution quality. Maximum payoff is reported and never ranked on. Structures are ordered by a single score built from confidence-adjusted expected value against the market’s own price, how often the downside is reached, probability of profit, and how cheaply the position can actually be opened.
Equity research as of 2026-09-09 (today) · Option chain as of 2026-09-08 · 1d old
What was thrown away first. Of 160,606 listed contracts across the names covered, just 25,985 (16.2%) were liquid enough to build a structure from — the rest were rejected on too little open interest (97,562), no usable bid (21,477) and a spread too wide to cross (15,582). That filter runs before any structure is assembled, so the ranking never sees the illiquid part of the chain.
What was withheld, and why. The option chain for AVB, NSA, TMHC, WBS was more than 3 trading days behind the rest of the book (21d behind 2026-09-08; 47d behind 2026-09-08; 20d behind 2026-09-08), so those names were dropped rather than priced off a frozen quote. 2 rows — LUV Protective Put (if held), WMT Long Strangle — priced at an expected value far outside anything else in the same table (robust z above 5, and more than 10pp from the table median). The option arithmetic is correct; the scenario set behind it puts too much probability above today’s price. The row is withheld, not adjusted, and the slot goes to the next-ranked structure.
Why it is mostly hedges. 5,346 structures were built and assessed in full; 1,745 rank across 654 names, and each table below shows its top structures. Once the cost of crossing the spread is charged, most option structures carry negative expected value, which is what a roughly fairly priced market looks like. What survives is cheap to put on and bounded in what it can lose — a hedge.
Income
Structures that collect premium. Ranked on what is left after the cost of opening them, not on the premium collected.
| # | Underlying | Score | Adj. EV | Total loss | |
|---|---|---|---|---|---|
| 1 | NBIX | 70.7 | -0.1% | 0.0% | 26.0% |
| 2 | WM | 70.7 | -0.1% | 0.0% | 6.0% |
| 3 | TSLA | 70.1 | -0.6% | 14.2% | 43.0% |
| 4 | TSLA | 69.9 | -0.5% | 14.2% | 48.0% |
| 5 | XYL | 69.8 | -0.2% | 0.0% | 4.0% |
| 6 | CI | 69.7 | -0.2% | 0.0% | 9.0% |
| 7 | PG | 69.6 | -0.1% | 0.0% | 6.0% |
| 8 | TMUS | 69.5 | -0.1% | 0.0% | 11.0% |
| 9 | KMI | 69.5 | -0.2% | 0.0% | 2.0% |
| 10 | AMZN | 69.5 | -0.2% | 0.0% | 21.0% |
| 11 | GOOGL | 69.5 | -0.1% | 0.0% | 26.0% |
| 12 | CB | 69.4 | -0.5% | 0.0% | 15.0% |
| 13 | SO | 69.4 | -0.3% | 0.0% | 4.0% |
| 14 | AMZN | 69.3 | -0.3% | 0.0% | 28.0% |
| 15 | CB | 69.3 | -0.5% | 0.0% | 10.0% |
| 16 | KMB | 69.3 | -0.2% | 0.0% | 7.0% |
| 17 | WST | 69.3 | -0.6% | 0.0% | 8.0% |
| 18 | GOOGL | 69.3 | +0.2% | 0.0% | 19.0% |
| 19 | BRK-B | 69.3 | -0.3% | 0.0% | 20.0% |
| 20 | LHX | 69.2 | -0.3% | 0.0% | 8.0% |
Leverage
Directional structures with defined risk. Expected shortfall sits at its floor for essentially all of these, so the downside is described by how often total loss occurs rather than how deep it goes.
| # | Underlying | Score | Adj. EV | Total loss | |
|---|---|---|---|---|---|
| 1 | PVH | 77.4 | +32.6% | 10.4% | — |
| 2 | CAVA | 75.0 | +15.0% | 13.9% | 86.0% |
| 3 | HON | 74.9 | +15.0% | 8.3% | — |
| 4 | KO | 72.6 | +31.5% | 42.0% | 449.0% |
| 5 | PVH | 70.9 | +9.5% | 4.7% | — |
| 6 | LUV | 70.1 | +39.3% | 51.9% | — |
| 7 | BAC | 69.8 | +18.6% | 40.2% | 293.0% |
| 8 | FIX | 69.7 | -0.8% | 6.4% | — |
| 9 | MA | 69.5 | +8.5% | 33.3% | 143.0% |
| 10 | AXP | 69.2 | -1.0% | 7.5% | — |
| 11 | LIN | 69.2 | -0.5% | 7.5% | — |
| 12 | HOOD | 68.9 | +12.5% | 29.5% | 127.0% |
| 13 | GD | 68.0 | -2.8% | 8.7% | — |
| 14 | HOOD | 68.0 | +7.2% | 29.4% | 100.0% |
| 15 | CVX | 68.0 | +9.8% | 39.2% | 155.0% |
| 16 | MGM | 67.9 | +43.4% | 47.6% | — |
| 17 | WEX | 67.8 | -4.8% | 6.3% | — |
| 18 | EME | 67.8 | -1.2% | 11.7% | — |
| 19 | M | 67.5 | +48.3% | 54.0% | — |
| 20 | CMCSA | 67.1 | -5.6% | 5.7% | — |
Hedge
Protection for an existing holding. Capital at risk is the underlying position, so unlike the defined-risk structures above the tail genuinely varies and can be ordered on.
| # | Underlying | Score | Adj. EV | Total loss | |
|---|---|---|---|---|---|
| 1 | SO | 70.0 | +0.1% | 0.0% | — |
| 2 | CLX | 69.5 | -0.1% | 0.0% | — |
| 3 | ROL | 69.5 | +1.3% | 0.0% | — |
| 4 | DECK | 69.4 | +0.0% | 0.0% | 11.0% |
| 5 | LHX | 69.4 | +0.3% | 0.0% | — |
| 6 | UPS | 69.4 | +0.0% | 0.0% | — |
| 7 | NOC | 69.3 | -0.6% | 0.0% | 3.0% |
| 8 | HON | 69.3 | +7.7% | 0.0% | — |
| 9 | LUV | 69.2 | +1.1% | 0.0% | — |
| 10 | AA | 69.2 | -0.1% | 0.0% | — |
| 11 | LHX | 69.1 | 0.0% | 0.0% | 16.0% |
| 12 | PG | 69.0 | -0.1% | 0.0% | — |
| 13 | CME | 69.0 | -0.1% | 0.0% | — |
| 14 | PWR | 69.0 | +0.0% | 0.0% | — |
| 15 | XOM | 69.0 | -0.1% | 0.0% | — |
| 16 | DECK | 69.0 | +0.6% | 0.0% | — |
| 17 | UPS | 68.9 | -0.0% | 0.0% | — |
| 18 | BLK | 68.9 | -0.3% | 0.0% | — |
| 19 | VFC | 68.9 | +1.4% | 0.0% | — |
| 20 | ECL | 68.9 | -0.1% | 0.0% | — |
Volatility
Structures whose thesis is the SIZE of the move rather than its direction, admitted under AM-017. They are only built for a name whose IV/RV sits in the top or bottom quartile of the estate on the same date — roughly half the book has no volatility view and carries none of these. Long-vol and short-vol are never built for the same name: they lose money in opposite states of the world, so they are not ranked against each other on one scale.
| # | Underlying | Score | Adj. EV | Total loss | |
|---|---|---|---|---|---|
| 1 | ABT | 71.5 | +7.0% | 10.3% | — |
| 2 | PYPL | 69.2 | +0.5% | 7.4% | — |
| 3 | AAPL | 68.9 | -1.7% | 4.6% | — |
| 4 | GEV | 68.5 | -2.3% | 5.3% | — |
| 5 | LRCX | 68.5 | -2.3% | 5.9% | — |
| 6 | TMUS | 68.3 | -2.1% | 5.9% | — |
| 7 | VRT | 67.9 | -2.7% | 5.8% | — |
| 8 | NVDA | 67.5 | +1.4% | 4.6% | — |
| 9 | IFF | 67.4 | +7.1% | 9.2% | — |
| 10 | BAC | 67.2 | -9.1% | 12.1% | 40.0% |
| 11 | ACM | 66.8 | +5.2% | 12.8% | — |
| 12 | CMG | 66.7 | -1.7% | 9.3% | — |
| 13 | MSFT | 65.8 | -1.9% | 5.5% | — |
| 14 | ONTO | 65.0 | -4.1% | 4.2% | — |
| 15 | AMAT | 65.0 | -2.7% | 5.5% | — |
| 16 | COIN | 65.0 | -7.4% | 4.4% | — |
| 17 | DDOG | 64.5 | -4.7% | 5.1% | — |
| 18 | WDAY | 64.4 | -4.5% | 7.1% | — |
| 19 | LITE | 63.0 | -2.6% | 13.5% | — |
| 20 | OLN | 62.4 | -9.5% | 9.2% | — |
Track record
Nothing has settled yet, so there is no performance to report. The record itself exists: 79 structures across 60 names and 9 families, each recorded at entry (legs, price, capital at risk) in a hash-chained ledger from 2026-08-05. The first expires 2026-08-21 and the last 2027-06-17; they settle as they expire and are never backfilled. Families are separate cohorts, never pooled.
1 withdrawn. A structure taken off this page stays in the record: the registration stands and the withdrawal is recorded beside it with its date and reason. A withdrawn trade is carried as neither live nor settled.
Withheld by analyst review: ACI. These names are excluded from the ranking entirely — the same hold that stops their research report being published now stops them appearing here.
This ranking was published from 2026-07-31 and its forward record opens 2026-08-05: for those five days no structure was pre-registered, because the path from a ranked row to the ledger did not exist. That week is NOT reconstructed — backfilling pre-registrations after the outcomes have begun is not a record of anything.
The previous ranking’s record, reported separately. 60,078 structures pre-registered, 55,557 still open since 2026-07-28, 4,521 settled. 1,429 duplicate registrations are excluded from that count rather than netted off in silence. Those structures were selected under different rules, so the two records are kept apart and never pooled.
Where the return comes from. Our own backtest attributes essentially all of this sleeve’s historical return to the volatility risk premium — implied volatility persistently above realised — not to any ranking. The edge tested as systematic, not selective, and the sample is optimistic on frictions: a screen over a known premium, not evidence of selection.
What this cannot tell you
- Volatility structures compete for the same per-name candidate budget as everything else (the most tradeable few per name are carried forward). Their introduction therefore displaced structures from the other tables — 243 admitted, 171 displaced — so rows in Income, Leverage and Hedge can change without anything about those structures changing.
- Held to expiry — early exercise and assignment are not modelled.
- The option chain is an end-of-day snapshot; there is no per-contract quote timestamp, so quote freshness cannot be measured and is not claimed.
- The house distribution is the engine's own persisted quantile grid, sampled exactly where that grid exists and reconstructed from stored moments where it does not. Each row says which it got, and a reconstruction carries its own error.
- The house view is a valuation distribution at a one-year horizon, re-horizoned to each option's expiry — drift scaled linearly in time, dispersion as its square root. It is not a forecast of the path in between, and for short expiries most of the house's conviction is scaled away, by design.
- No forward track record exists for this ranking. The v1 record continues separately and the two are never pooled.
- MCH is not a registered financial adviser. Nothing here is advice or an executable quote.