Do same-day, next-day and weekly option straddles look rich or cheap against the recent realized move?
At 10:00 ET, when the at-the-money straddle's implied move is at least 25% above the recent realized move (sell) or at most 80% of it (buy), does the trade pay after costs at same-day, next-day and weekly expiries on SPY, QQQ and IWM?
No trial passed, so no Buy or Sell label may be shown for any expiry. None of the six pre-registered trials was profitable, and all six are themselves reported not shown: 41.0% to 79.0% of each trial's trades had at least one leg with no usable exit quote, above the 20% limit fixed before the run. The measured totals, one straddle each and net of $0.65 per contract per side, were: same-day sell +$238.80 over 307 trades; same-day buy -$2,094.40 over 134; next-day sell -$12,228.00 over 280; next-day buy -$11,546.20 over 162; weekly sell -$2,727.20 over 62; weekly buy -$1,804.00 over 125. The only positive total, the same-day sell, is the sum of IWM +$777.80, SPY -$102.80 and QQQ -$436.20, and 58.8% of IWM's trades have an imputed exit. No trial's 5% lower bound was above zero, even at zero commissions. Because every trial is not shown, the study does not show that these rules lose money; it shows they could not be measured cleanly with the quotes available. The 120-session holdout was not read and stays sealed.
The study verdict is fail because no trial passed the pre-registered profitability bar. All six trials are themselves reported not shown: 41.0% to 79.0% of each trial's trades had at least one leg whose exit price was taken from the underlying because no usable quote existed, above the 20% limit, so their profit and loss is not a clean measurement.
Backtested by QS Research
Pre-registration
Six trials were fixed before any quote for the counted window was read, on SPY, QQQ and IWM. At 10:00 ET the study measures the at-the-money straddle's implied move, the straddle mid divided by spot, against the mean absolute move over the last 20 completed instances of the same horizon. It sells when implied is at least 1.25 times realized and buys when at most 0.80 times, for same-day, next-day and weekly expiries. Sells enter at the bid and buys at the ask; exits are at 15:55 on expiry day; costs are $0.65 per contract per side. A trial is profitable only if total profit is above zero and a 5% lower bound from resampling by week is above zero; a trial with more than 20% of its trades with an exit priced from the underlying, because no usable quote existed, is reported not shown. The counted window is 2023-01-03 to 2024-08-30; the later stretch from September 2024 to October 2026 is descriptive only and was not run; a 120-session holdout stays sealed. Four amendments were made before any quote: the fetch order, the listing start dates, the two pinned snapshots, and the rulings that IWM enters from its first listing and the three half days are left out of the counted read.
Registered 2026-11-06Amendments
- 2026-11-12 Schedule only: this study's data fetch was moved ahead of another study's. No rule, window, symbol, threshold, trial, fill, cost or verdict changed; N stays 6. approved: Approved 2026-11-12; schedule only, no rule changed
- 2026-11-12 Record only, before any quote: the first session each symbol's same-day, next-day and weekly expiries were listed, and the contracts to fetch. No rule changed; N stays 6. approved: Recorded 2026-11-12, before any quote was read
- 2026-11-12 Record only, before any quote: the two pinned data snapshots, one for 1-minute prices and one for 1-minute option quotes and listings. No rule changed; N stays 6. approved: Recorded 2026-11-12, before any quote was read
- 2026-11-12 The rulings: IWM enters the same-day and next-day trials from its first listing, with 122 of the 418 sessions having no such expiry and counting as skips, and trades with a half day on any session from entry to expiry are left out of the counted read. The variant with them left in is shown beside it. No other rule changed; N stays 6. approved: Approved 2026-11-12; rulings recorded before the counted run
A holdout sample is sealed: its results stay unread until the study is final.
Trials · 6 trials · 0 passed · judged at N = 6 trials; 2965 trials run in total
| Trial | Verdict | |
|---|---|---|
| S0: 0DTE, sell when implied/realized ≥ 1.25 | RESULT | counted |
| B0: 0DTE, buy when implied/realized ≤ 0.80 | RESULT | counted |
| S1: 1DTE, sell when implied/realized ≥ 1.25 | RESULT | counted |
| B1: 1DTE, buy when implied/realized ≤ 0.80 | RESULT | counted |
| SW: weekly, sell when implied/realized ≥ 1.25 | RESULT | counted |
| BW: weekly, buy when implied/realized ≤ 0.80 | RESULT | counted |
Limitations
- All six trials are reported not shown: 41.0% to 79.0% of each trial's trades had at least one leg with no usable exit quote, above the 20% limit fixed before the run. Their profit and loss is therefore not a clean measurement, and the study does not show that the rules lose money.
- The only positive trial total, the same-day sell at +$238.80, is the sum of IWM +$777.80, SPY -$102.80 and QQQ -$436.20; 58.8% of IWM's trades have an imputed exit, so that positive total is not reliable.
- Quoted one-minute bid and ask, not actual fills: sells enter at the bid and buys at the ask, one straddle at a time, with no size or market-impact model. Costs are $0.65 per contract per side.
- IWM's same-day and next-day expiries were listed only on some sessions until April 2024: 122 of the 418 counted sessions for each are skips with no listed expiry, and IWM enters those trials only on sessions where an expiry was listed.
- The three half days in the window (2023-07-03, 2023-11-24 and 2024-07-03) are left out of the counted read. With them left in, every trial is still not shown and the totals differ by up to $1,112.80.
- The later stretch from 2024-09-03 to 2026-10-30 was not fetched or run; only the counted window is tested.
- One design only: the at-the-money straddle entered at 10:00 and held to 15:55 on the expiry. It does not test SPX, single names, defined-risk spreads or intraday exits.
- The weekly control sells or buys every session, so its positions overlap; the weekly trials enter once per expiry week.
- The exits priced from the underlying are a rule, not a market price: a sell pays the in-the-money value at the close plus $0.05 a leg and a buy receives the in-the-money value only, so profit and loss on those trades is not a fair-value estimate.
- Each of the six trials was judged at its own one-sided 5% lower bound; a correction for the number of trials run (the deflation line against all QS Research trials) is shown with each verdict but does not decide it. Six chances at 5% is not a family-wise 5% test.
The numbers: every trial's checks, metrics, curve and yearly results, and the provenance
1 published version · current 65a7abf5d7ce38f9, published 2026-11-21