SPX 0DTE — Entry-Geometry Screen v1

Research · pre-pipeline · descriptive priors, not a tradeable signal

What this is. A descriptive screen of intraday path geometry for a candidate SPX 0DTE iron condor — how often (touch-rate) and how far (excursion) price breaches the short strikes, conditioned on entry-day state. It is neither a Setup nor a Discovery: there is no trade and no P&L here, so no tail-aware gate is constructible from it. It emits priors that feed the S026 forward trade simulation.

It does not name a best position. Geometry-best and P&L-best can be opposite: a tight short stays in less but earns more credit; a calm day has small strike distances. Credit and the IV-spike on a breach are invisible here. The "which pays best" verdict is made forward, where credit and the tail are real.

Built 2026-06-18 · SPY 5-min (Tiingo/IEX) as the SPX proxy · VIX1D prior-close (Cboe via Yahoo, EOD) · 797 regular full sessions, 2023-04-01 → 2026-06-18 (777 with a warmed-up gap σ).

Method

Entry time is an axis, not a constant: we sweep entry anchors on a 5-min grid from 09:30 to 11:00. At each anchor t the short strikes are placed at a vol-scaled distance d(Δ,t) = z(Δ) · (VIX1D⁄√252) · √τ · P₀ (τ = time left to the 16:00 PM-settlement), for three deltas {0.16, 0.25, 0.35} and two wings {$10, $20}. Over the hold window t→15:30 we record whether either short is touched, the breach excursion normalized by the wing (E/w — where ≈1 means through the wing, near max loss), and the time to first breach. Distances use % of SPY (SPY ≠ SPX). Conditioners are frozen a-priori and evaluated marginally.

Entry-time axis — the headline curve

Touch-rate (any short touched before 15:30) and the p95 breach depth, by entry time. Stay-in falls as the short moves closer to spot (0.35Δ) and as entry moves later (less time, but the distance shrinks too). This is descriptive. Reading the best time off the curve and trading it carries a selection penalty — a forward entry-time choice is a single pre-registered pick. The two shaded rows are the reference anchors profiled by tranche below.

Entry	Touch 0.16Δ	Touch 0.25Δ	Touch 0.35Δ	E/w p95 (0.16Δ/$20)
09:30	27%	62%	93%	0.16
09:35	27%	60%	92%	0.16
09:40	27%	61%	92%	0.15
09:45	26%	59%	93%	0.15
09:50	25%	56%	91%	0.16
09:55	25%	57%	91%	0.15
10:00	24%	55%	91%	0.15
10:05	23%	56%	90%	0.14
10:10	23%	54%	90%	0.13
10:15	22%	54%	89%	0.12
10:20	22%	52%	88%	0.11
10:25	22%	51%	89%	0.12
10:30	21%	52%	89%	0.12
10:35	21%	51%	87%	0.11
10:40	21%	50%	88%	0.11
10:45	20%	49%	87%	0.11
10:50	21%	48%	87%	0.10
10:55	20%	46%	86%	0.11
11:00	19%	46%	86%	0.10

Touch-rate by VIX1D tercile (0.16Δ short) — the realized-vol gradient at the reference entries:

VIX1D tercile	Touch @ 10:10	Touch @ 11:00
Low VIX1D	22%	17%
Mid VIX1D	25%	21%
High VIX1D	23%	20%

Tranche breach profiles — at the reference entries {10:10, 11:00}

For each conditioner bucket (a tranche) and each short delta: the stay-in rate, the p95 breach depth (E/w, on the $20-wing condor), and the median time to first breach. These are breach profiles shown side by side — they are not ranked or scored. The owner reads them and forms candidate tranche→structure hypotheses, tested forward under S026.

C2 · Overnight gap (σ-standardized) entry 10:10

A-priori: Asymmetric: a gap down is worse than a gap up of equal size.

Tranche	n	Stay 0.16Δ	Stay 0.25Δ	Stay 0.35Δ	E/w p95	Median to breach
Large gap ↑	56	80%	59%	20%	0.23	195m
Small ↑	190	76%	45%	11%	0.10	162m
Flat	361	79%	47%	10%	0.11	162m
Small ↓	118	78%	43%	8%	0.14	217m
Large gap ↓	52	63%	33%	4%	0.18	170m

C2 · Overnight gap (σ-standardized) entry 11:00

A-priori: Asymmetric: a gap down is worse than a gap up of equal size.

Tranche	n	Stay 0.16Δ	Stay 0.25Δ	Stay 0.35Δ	E/w p95	Median to breach
Large gap ↑	56	80%	61%	23%	0.21	180m
Small ↑	190	77%	56%	19%	0.10	162m
Flat	361	84%	57%	13%	0.09	175m
Small ↓	118	80%	44%	13%	0.10	172m
Large gap ↓	52	77%	42%	4%	0.10	145m

C1 · VIX1D level (prior close) entry 10:10

A-priori: Higher implied vol → lower stay-in (monotonic).

Tranche	n	Stay 0.16Δ	Stay 0.25Δ	Stay 0.35Δ	E/w p95	Median to breach
Low VIX1D	236	78%	47%	10%	0.10	197m
Mid	221	75%	44%	10%	0.14	155m
High	265	77%	47%	11%	0.18	190m

C1 · VIX1D level (prior close) entry 11:00

A-priori: Higher implied vol → lower stay-in (monotonic).

Tranche	n	Stay 0.16Δ	Stay 0.25Δ	Stay 0.35Δ	E/w p95	Median to breach
Low VIX1D	236	83%	59%	14%	0.07	190m
Mid	221	79%	53%	14%	0.11	137m
High	265	80%	49%	16%	0.16	165m

C4 · Scheduled-flow day entry 10:10

A-priori: FOMC lands ~14:00 ET, inside the hold window → strongly lower stay-in.

Tranche	n	Stay 0.16Δ	Stay 0.25Δ	Stay 0.35Δ	E/w p95	Median to breach
FOMC day	26	69%	27%	4%	0.13	285m
08:30 release (NFP)	36	72%	42%	8%	0.20	125m
Ordinary day	720	78%	46%	10%	0.13	182m

C4 · Scheduled-flow day entry 11:00

A-priori: FOMC lands ~14:00 ET, inside the hold window → strongly lower stay-in.

Tranche	n	Stay 0.16Δ	Stay 0.25Δ	Stay 0.35Δ	E/w p95	Median to breach
FOMC day	26	65%	31%	8%	0.11	220m
08:30 release (NFP)	36	75%	64%	11%	0.07	175m
Ordinary day	720	82%	54%	15%	0.10	162m

Hard fences. Stay-in and E/w are priors on breach frequency and depth — not P&L: they cannot see credit, nor the IV-spike on a put-side breach (real loss is worse), nor that calm tranches carry small strike distances. Geometry-best ≠ P&L-best. The screen names no best position and the entry-time curve is descriptive — any forward choice is a single pre-registered pick under S026's tail-aware gate, not an argmax read off these tables.

Coverage caveat. ~797 sessions, and the high-vol regime that matters most is the least sampled (~33% of sessions in the top VIX1D tercile, mostly clustered in a few episodes). Tail buckets (large gaps, FOMC) hold tens of sessions — read them as directional, not precise.

Descriptive screen. No gate, no trial budget, no vault touch. Output is priors for S026 tranche & entry-filter candidates; any verdict is forward, under S026's tail-aware gate. The underlying raw feeds and the forward 0DTE chain capture live under Data.