signal-platform/docs/research/execution-recovery.md

Execution recovery (2026-07-18) — when you fill is the edge you leave on the table

Report: reports/execution_recovery_matrix.json / .md
Follows Phase A A4 (phase-a-matrix.md).
Arms: close_control · next_open · stale_close · next_open_gap2.

Mechanics verified before sign-off: stale_close re-anchors stop to fill − 1.5×ATR(signal day); no same-day stop after a close fill; each t−1 candidate keeps its own gate/rank cross-section (no lookahead). Numbers trusted.

---

Results

Arm	Full Sharpe	Full CAGR	Full DD	Train Sharpe	Val Sharpe
close_control	1.77	48.3%	21.6%	1.75	1.68
stale_close	1.57	40.6%	21.8%	1.36	1.74
next_open	1.20	30.0%	28.2%	0.94	1.44
next_open_gap2	1.06	25.2%	27.8%	0.72	1.41

Monotonicity (the strongest evidence)

In every window, Sharpe recovers as the fill moves toward the signal:

Window	next_open → stale_close → close
Train	0.94 → 1.36 → 1.75
Validation	1.44 → 1.74 → 1.68*
Full	1.20 → 1.57 → 1.77

*Val close 1.68 is within SE of stale 1.74 — not a break of the story.

A dead edge does not produce a monotone gradient in fill timing. A live edge that is progressively surrendered to execution delay does. Combined with full-period DD recovery (28.2% → 21.8% ≈ close’s 21.6%), this confirms the diagnosis: when you fill, not decaying alpha.

Auto-flag recover: False is not a null

Pre-registered recovery applied the “≥ close − 0.5 SE” bar to stale_close, the lower-bound arm (one full session of lag). Flags:

Flag	Result
near_close_control (val)	True (Δ +0.06)
beats_next_open	True (val Δ +0.30)
train_ok	False (1.36 vs need ≥ ~1.39 — miss by ~0.03)

The floor missed “≥ close − 0.5 SE” by 0.03 while the live design is expected to sit above the floor. The flag worked correctly on the wrong object.

Log sentence: Partial recovery proven; full recovery needs same-day fill.

Live outcome bracket

Bound	Arm	Meaning
Floor	stale_close ~1.57 full	One-session-stale close fill (conservative)
Ceiling	close_control ~1.77 full	Same-day close; optimistic only by final ~15 min of signal info

Real near-close execution (scan ~15:30–15:40 ET on a ~99% complete bar, MOC by 15:50/15:55) is signal-at-partial-bar filled at the same close the control uses. Live truth is bracketed [1.57, 1.77] with residual uncertainty of ~15 minutes of staleness, not 24 hours — expect near the top of the bracket.

stale_close alone already justifies the schedule change. No further fill-timing simulation on this snapshot — the bracket is the result. You cannot simulate 15:45 partial bars from daily data, and you do not need to.

Gap-cap — dead (third tail-trim instance)

next_open_gap2 worse than plain next_open on every window (full Sharpe 1.06 vs 1.20). 262 full-period gap-ups skipped — they were continuations.

This is the third independent instance of the same lesson:

1. Take-profit exits (gate target as TP)
2. FIP as an in-book filter
3. Gap-up entry caps

Any rule that trims the right tail improves cosmetic quality metrics and destroys P&L. Standing method rule: momentum filters must be presumed guilty of tail-trimming until shown otherwise.

---

Decisions locked

1. Biggest lever is execution scheduling, not a strategy rewrite.
2. Until near-close execution is live: grade strategy promotion under fill_mode=next_open; keep close-fill as historical control.
3. After near-close ships: grade under a close-like / near-close fill mode (actual live path).
4. No more sim arms on fill timing for this snapshot.
5. Gap-cap: do not ship.
6. Strategy work (nasdaq_all, fip_id, sector) waits until the execution path is decided — those experiments must be graded under the fill mode you will trade.

---

Ops design — implementation plan (code-checked)

Assumptions verified against current code before ship:

• Intraday pipeline already fetches/upserts the in-progress day-t bar all session (fetch_ohlcv end_date defaults to today). Near-close job = OHLCV fetch → R:R scan (no new snapshot synthesizer).
• One global schedule_timezone (default Europe/Berlin); stored SystemSetting values shadow code defaults — defaults alone do not migrate prod.
• observe_reentry_gate_transitions stamps timestamps with no same-day guard today — dual scans would accelerate fail→requalify unless fixed in trade_policy.

Semantic guard (ship step 1 — precondition)

In trade_policy (not the scheduler):

reentry_gate_requalified_at may only be set when reentry_gate_failed_at falls on an earlier America/New_York trading date than the current observation.

Manual mid-day scans stay allowed; same-day fail+qualify cannot unlock. Unit test: fail 10:00 / qualify 15:35 same day → still locked; qualify next day → unlocked.

Schedule split

Slot (America/New_York)	Jobs
Morning (~02:00)	OHLCV backfill, benchmark, sentiment, fundamentals — no qualifying R:R scan
Near-close (~15:30 Mon–Fri)	OHLCV fetch (refresh day-t bar) → R:R scan (only daily qualifying observation)
After close (~16:30–17:00)	Outcome eval on its own slot (not chained to the partial-bar scan)
Intraday hourly	Unchanged in NY terms; last ~16:00 still mid-session under 15m feed

• Near-close scan 1–5 only; US-holiday no-ops are fine (stale identical data can’t flip gates) — comment only, no exchange calendar.
• Do not run morning + near-close qualifying scans; move the scan, don’t add a second.

Behavior change to document (not an accident)

With scan at ~15:35 ET, stops closed by earlier same-day intraday outcome evals can get a same-day fail observation — closer to the promoted gate_reset arm (stop-day close may establish failure) than today’s strict_gate_reset analogue (scan always before any eval). Stops after the bell still wait a day. Rewrite README “Live timing matters” / post-stop sections and a line here when shipping.

Feed / paper honesty

• Document 15-minute delayed SIP: 15:35 scan may see ~15:20 prices; OK for 12-1.
• Paper entry price ≈ scan entry (near close) is nearly automatic; add fill_mode=near_close era tag so Track Record can separate morning-scan / near-close / future broker-routed eras.
• Morning sentiment staleness is display-only; gate is price-only (GTL parity / neutral-sentiment backtest). One doc line closes that.

Stored settings migration (ship step 4)

Flip global default TZ to America/New_York and re-express crons in NY time. Also migration (or documented Admin rewrite) of stored schedule_* keys so prod does not keep 07:00 Berlin silently.

Ship order

1. trade_policy distinct-day requalify guard + unit test
2. Near-close job = existing fetch → scan; outcome eval own after-close slot
3. Paper fill_mode=near_close era tag; verify entry marking
4. Defaults + stored settings migration + README/research timing rewrite

Out of scope

Broker MOC routing, more fill-timing sim, nasdaq_all / fip / sector (grade later under the fill mode you trade).

---

Implementation status

Item	Status
Research evidence	Closed — this doc + matrix report
Distinct-day gate-reset guard	Shipped — trade_policy + unit test
Schedule split + near-close scan→alert	Shipped — morning / near-close / after-close (fetch→outcome)
Paper era tag	Shipped — fill_mode=near_close on new paper trades
Settings migration + docs	Shipped — alembic 023 rewrites schedule_*; README updated

Shipped behavior change (not accidental)

Near-close scan at ~15:30 ET lets same-day fail observations after earlier intraday stop closes — closer to promoted gate_reset than the old morning-scan strict_gate_reset analogue. Documented in README.