Snapshots recorded before outcomes are known, never rewritten.
Every trading day after the close, the site records what the options market was pricing — per symbol, per expiration: spot, expected move, expected range, ATM IV and the straddle price, stamped with the date and methodology version. These snapshots are append-only: once written they are never modified, whatever happens next.
After an expiration passes, the outcome fields are filled in alongside (never over) the original estimate: the closing price on expiration day, the realized move from the snapshot's spot, and whether that close landed inside the recorded range.
Reading the statistics
- % inside range — how often the close landed within the expected range. Under the model's own assumptions a one-straddle range should contain the outcome roughly 60% of the time (it is about a 0.85σ band each way, not a 68% band) — so neither 55% nor 70% is a "failure"; the point is the record.
- Implied vs realized — average |actual| against average expected move shows whether options were pricing more or less movement than occurred over the sample.
- Mean absolute error — average | |actual| − expected |, a symmetric accuracy measure.
We deliberately avoid the phrase "win rate": nothing here wins or loses — these are calibration statistics for a market-implied estimate.
Comparisons use closing prices on the expiration date. Some index options settle to special opening quotations (SPX AM-settled series, VIX), which can differ from the close; the comparison shown is close-based and labeled as such.
Recording began August 31, 2026. History accumulates from that date — the site does not reconstruct or purchase backdated options data, so early pages show short samples and say so.