MOO→11:00 Full Audit · May 02
Window: 2021-04-01 .. 2026-04-24 (5y) | Train/OOS split: 2025-11-01 | Generated: 2026-05-03
1. Executive Summary
ALL DEAD — NO DEPLOY
5 strategies audited. 15 hypotheses screened. 0 HEALTHY strategies and 0 VALIDATED hypotheses survived the three-persona verdict gate on the full 5y window.
OOS permutation p-values cluster near 1.0 for most strategies, meaning within-ticker shuffle gives equal-or-better Sharpe — consistent with Apr 26 audit finding "0/12 stacks robust".
Production state preserved per Apr 29 manual-only directive. No flag changes.
2. Phase 1 — Strategy Audit (5 strategies)
| Strategy | N Train | N OOS | Sh Train | Sh OOS | Perm p (OOS) | Verdict |
| 17_sub5_pump_short |
1225 |
220 |
0.26 |
0.40 |
0.982 |
DEAD |
| 18_organic_pump_short |
17640 |
636 |
-0.10 |
1.59 |
1.000 |
DEAD |
| 19_organic_dump_long |
5780 |
899 |
-0.70 |
0.31 |
0.690 |
DEAD |
| 21_moo_long_baseline |
694486 |
139146 |
-0.10 |
0.41 |
0.999 |
DEAD |
| 22_moo_short_baseline |
694486 |
139146 |
0.32 |
0.08 |
0.997 |
DEAD |
| smoke |
150 |
50 |
0.92 |
2.16 |
0.930 |
DEAD |
3. Phase 2 — Hypothesis Triage (15 candidates)
| Hyp | N Total | Sh Train | Sh OOS | Perm p | Mean PnL% | Verdict |
| H1 |
0 |
— |
— |
— |
— |
INSUFFICIENT_DATA |
| H2 |
1326 |
0.33 |
1.25 |
0.314 |
0.0548 |
REJECTED |
| H3 |
0 |
0.00 |
0.00 |
— |
— |
REJECTED |
| H4 |
1413 |
-0.05 |
-0.51 |
1.000 |
-0.0056 |
REJECTED |
| H5 |
0 |
— |
— |
— |
— |
INSUFFICIENT_DATA |
| H6 |
333 |
-2.10 |
3.17 |
1.000 |
-0.2299 |
REJECTED |
| H7 |
84 |
2.60 |
0.00 |
— |
0.3529 |
REJECTED |
| H8 |
367 |
-1.79 |
1.75 |
1.000 |
-0.2094 |
REJECTED |
| H9 |
536 |
-0.90 |
-0.84 |
1.000 |
-0.1047 |
REJECTED |
| H10 |
280 |
0.07 |
0.63 |
1.000 |
0.0095 |
REJECTED |
| H11 |
0 |
— |
— |
— |
— |
INSUFFICIENT_DATA |
| H12 |
0 |
— |
— |
— |
— |
INSUFFICIENT_DATA |
| H13 |
542 |
0.55 |
2.37 |
1.000 |
0.0682 |
REJECTED |
| H14 |
0 |
— |
— |
— |
— |
INSUFFICIENT_DATA |
| H15 |
1967 |
-0.38 |
-0.25 |
0.999 |
-0.0413 |
REJECTED |
4. Phase 3 — Allocation Synthesis
# Allocation Recommendation — Phase 1+2 Synthesis
**Generated:** 2026-05-03
**Audit window:** 2024-2026 partial (post-open backfill incomplete)
**Phase 1 strategies:** 6 evaluated
**Phase 2 hypotheses:** 15 screened
---
## 1. Surviving Strategies (HEALTHY)
_None_
**Dead strategies (6):** 17_sub5_pump_short, 18_organic_pump_short, 19_organic_dump_long, 21_moo_long_baseline, 22_moo_short_baseline, smoke
---
## 2. Joint Portfolio
No joint backtest possible — 0 HEALTHY strategies.
---
## 3. Validated Hypotheses - Plugin Roadmap
_None validated._
---
## 4. Promising Hypotheses (paper-only candidates)
_None promising._
---
## 5. Rejected + Insufficient Data
### Rejected (10)
| hyp_id | n_total | oos_sh | perm_p | mean_pnl_pct |
| --- | --- | --- | --- | --- |
| H2 | 851 | 1.2535368520488246 | 0.3136863136863136 | -0.1178641032759733 |
| H3 | 0 | 0.0 | nan | nan |
| H4 | 470 | -0.5116291966371407 | 1.0 | -0.1162314244373363 |
| H6 | 154 | 3.173281206329874 | 1.0 | -0.025017128430915 |
| H7 | 36 | 0.0 | nan | 0.642911442208864 |
| H8 | 170 | 1.7471317304838565 | 1.0 | -0.1879443898808548 |
| H9 | 232 | -0.8406858428814901 | 1.0 | -0.0553623680673909 |
| H10 | 125 | 0.6343669653432151 | 1.0 | -0.1071047935065588 |
| H13 | 158 | 2.367234878093825 | 1.0 | 0.0253317487640101 |
| H15 | 626 | -0.2502570240248232 | 0.999000999000999 | -0.183853065012351 |
### Insufficient Data (5)
| hyp_id | n_total | verdict |
| --- | --- | --- |
| H1 | 0 | INSUFFICIENT_DATA |
| H5 | 0 | INSUFFICIENT_DATA |
| H11 | 0 | INSUFFICIENT_DATA |
| H12 | 0 | INSUFFICIENT_DATA |
| H14 | 0 | INSUFFICIENT_DATA |
---
## 6. Recommendation
Phase 1+2 partial-window (2024-2026) found **0 HEALTHY strategies** and **0 VALIDATED hypotheses**.
**Recommendation: no live deploys from this audit cycle on partial window. Re-run on full 2021-2026 window after post-open backfill completes.**
Rationale:
- All 6 strategies scored DEAD under OOS perm-test (perm_p all ≥ 0.69, most ≥ 0.99)
- All 15 hypotheses REJECTED or INSUFFICIENT_DATA on 2024-2026 slice
- 2024-2026 window is structurally different (post-tariff, high-VIX 2025 regime shifts)
- Post-open intraday data backfill is incomplete — true edges likely masked by data gaps
- Re-audit after full 2021-2026 OHLCV+PM backfill, targeting N≥500 OOS trades per strategy
5. Recommendations & Follow-Up
No live-deploy changes from this audit cycle.
- Backfill features for 5 INSUFFICIENT_DATA hypotheses: first5min ORB break (H1), halt-yesterday (H5), sector divergence (H11), RSI14 (H12), first5min low broken (H14).
- Expand OOS window: currently 5 months (2025-11 .. 2026-04); need 12+ months for stable N≥500 OOS trades per strategy.
- 2026 tariff/macro regime shift may be masking signal — re-run full audit Q3 2026 (next: 2026-08-02).
- Explore joint-edge stacks (multi-feature) instead of single-factor screens before declaring strategy dead.
- Production config frozen per Apr 29 directive — manual decision required for any live changes.