When SPY first crosses down into the multiday put trigger (-23.6% of weekly ATR off the prior Friday close), the question is whether it reclaims pivot before the downside Golden Gate opens. Across 694 clean events from 2000-2026, pivot reclaim happens 38.0% of the time. The downside GG opens first 58.5% of the time. Shorting the break is the base-rate play; the long reversion only earns its keep inside specific regimes.
Weekly ATR levels are computed from the prior completed weekly bar using Wilder's 14-period ATR — the Saty period_index = 1 convention. Every ISO week since 2000 is classified into one of five cohorts based on whether and how price interacted with the multiday put trigger. The race runs with 1-minute RTH bars, wick-clipped against phantom prints.
Path rules: the downside stop is counted from the tap bar itself (the bar that prints the trigger). Upside targets only count from bars after the tap bar, to avoid same-minute ambiguity. All outcomes must resolve within the same trading week.
The base case is that the trigger tap fails. The downside Golden Gate opens first in 58.5% of events, and only ~38% of taps reclaim pivot before losing -38.2%. Stacking the runner legs on top — pivot, then call trigger, then upside GG — collapses the cumulative odds to 14.8% for a full weekly rotation.
| Outcome in-week (after clean tap) | Rate | N |
|---|---|---|
| Breakdown (-38.2% dgg opens before pivot) | 58.5% | 406 |
| Reversion only (pivot reclaimed, no call trig) | 16.6% | 115 |
| Continuation (pivot + call trig, no ugg) | 6.6% | 46 |
| Full rotation (pivot + call trig + ugg open) | 14.8% | 103 |
| No resolution by Friday close | 3.5% | 24 |
Runners compound, but they also shed events fast. After pivot reclaims, just over half continue to the call trigger. Of those, two-thirds make it to the upside GG.
Where does the weekly put trigger sit relative to the quarterly Saty levels? This context matters. When the weekly put trigger falls inside the q_pivot_to_call band (between prior-quarter close and quarterly call trigger), pivot reclaim jumps to 45.3% and breakdown drops to 50.9% — the best edge in the table. When it sits above the quarterly upside GG (an extended market), reversion is weaker.
| Quarterly band at tap | N | P(pivot) | P(call) | P(ugg) | P(breakdown) |
|---|---|---|---|---|---|
| Above quarterly ugg | 160 | 35.0% | 21.2% | 13.1% | 61.9% |
| q_call to q_ugg | 79 | 40.5% | 24.1% | 15.2% | 58.2% |
| q_pivot to q_call | 159 | 45.3% | 22.0% | 15.1% | 50.9% |
| q_pivot to q_put | 129 | 36.4% | 24.8% | 14.7% | 58.9% |
| q_put to q_dgg | 36 | 36.1% | 25.0% | 22.2% | 63.9% |
| Below quarterly dgg | 83 | 37.3% | 21.7% | 20.5% | 57.8% |
Classic Bilbo states — weekly PO above +61.8 rising, or below -61.8 falling — are rare at the moment the trigger taps. The bulk of events happen with weekly PO mid-zone. mid_rising is the largest and best single cell: n=321, 40.2% pivot reclaim, 55.1% breakdown.
| Weekly PO at start of week | N | P(pivot) | P(call) | P(ugg) | P(breakdown) |
|---|---|---|---|---|---|
| Bull bilbo (PO high & rising) | 97 | 35.1% | 17.5% | 11.3% | 62.9% |
| Bear bilbo (PO low & falling) | 14* | 28.6% | 7.1% | 7.1% | 71.4% |
| Mid & rising | 321 | 40.2% | 23.1% | 14.3% | 55.1% |
| Mid & falling | 267 | 34.1% | 19.9% | 15.4% | 64.0% |
| High & falling | 41 | 39.0% | 26.8% | 19.5% | 58.5% |
| Low & rising | 9* | 44.4% | 22.2% | 22.2% | 55.6% |
* n < 20. Sample too small for a confident read.
Not every tap is tradable. Of 1,352 weeks, only 694 produce a clean intraweek cross where the 1m bar actually prints at the trigger price. The other tap-type cohorts — gap-throughs, Monday open-belows — are materially worse and must be treated separately.
| Cohort | N | P(pivot) | P(breakdown) | P(full rotation) |
|---|---|---|---|---|
| Intraweek clean cross | 694 | 38.0% | 58.5% | 14.8% |
| Gap-through cross (bar opened below trigger) | 63 | 14.3% | 85.7% | 3.2% |
| Opened below put trigger (Monday gap) | 60 | 25.0% | 75.0% | 11.7% |
| Opened below dgg (gap-through-stop) | 50 | — | — | — |
| Non-tap weeks | 485 | — | — | — |
Gap-through warning: when the tap bar itself opens below the trigger (not a clean cross), breakdown rate is 85.7%. Entering long on a gap-through is structurally different from entering on a clean trigger print. Don't conflate.
The reversion rate is remarkably stable across the 2000s, 2010s, and 2020s — all 36-38%. The edge (or lack of it) is not an artifact of a single market regime.
| Decade | N | P(pivot) | P(call) | P(ugg) | P(breakdown) |
|---|---|---|---|---|---|
| 2000s | 313 | 36.1% | 21.1% | 13.7% | 60.7% |
| 2010s | 279 | 37.6% | 21.5% | 16.5% | 58.4% |
| 2020s (through 2026-Q1) | 162 | 37.7% | 20.4% | 13.0% | 60.5% |
Long at the tap, stop at downside GG open (-38.2%). That's a 0.146-ATR risk unit. Each upside target produces a different R:R and a different win rate. Unresolved events exit at Friday close — two views shown: conservative (unresolved = -1R) and time-stop (unresolved = 0R).
Read: the "first tap of multiday put trigger is a good long" thesis earns barely positive expectancy on the pivot-only target, and goes negative if you hold for runners. The edge you really want is regime-filtered — the q_pivot_to_call quarterly band at 45.3% pivot reclaim is where this trade earns its keep.
Levels: weekly ATR anchored to the prior Friday close, Wilder 14-period RMA (SMA-seeded). Quarterly context built from calendar-quarter daily OHLC. No look-ahead: week W's levels come strictly from week W−1.
Data: 2.56M 1-minute RTH SPY bars, 1,367 weekly bars, 6,588 daily bars. 2000-01-03 through 2026-04-09. Wick-clipped at 2% of body to filter phantom prints.
Path resolution: downside stop counted from the tap bar; upside targets only counted from bars after the tap bar. All outcomes must resolve in-week. Unresolved events are labeled explicitly and reported under both conservative and time-stop expectancy views.
Cohort hygiene: intraweek cross requires the tap bar's range to contain the trigger price. Gap-through taps, Monday opens below trigger, and gap-through-stop weeks are separated.
Control: non-tap week base rates are reported but labeled — they are a bullish-regime cohort, not a clean counterfactual. Near-tap weeks (within 0.10 ATR of trigger) sit in between.