ALPHA ALGOS — VALIDATION EVIDENCE PACKAGE
OG (Alpha Dog Elite, MNQ 15m, gate-off lineage)
Generated 2026-08-22T18:04:11+00:00 ·
attempt universe N = 20,734 (strict, incl. screening; N-LOCK-1.0) ·
no statistic in this document is shown un-deflated where deflation applies.
This package is evidence, not a recommendation.
It reports what the validation battery measured, including regions where the
data is insufficient to conclude anything. It does not certify the strategy
and does not recommend deployment. Interpretation and any deployment decision
belong to the reader.
Pre-registration record
item value
Run pre-registration OG-RUN_v1.0 · locked 2026-08-02T00:08:21+00:00 · sha256 0a3472be6fad029f27424d8ab8467d061df43bfdf732010fb7e8d35448a3d4a9
House standard HS-1.1 · sha256 1b17933c50d9be224455343c3554c54d610dd154a362286372218d9764d36a6a
Timing discipline All thresholds locked and hashed BEFORE results were computed; any change is a new versioned lock — originals stay on record.
Per-layer results
layer status sample n deflated stat as measured
OOS FAIL 309 0.087 @ N=20,734 DSR=0.087 at N=20,000 (bar ≥0.95), OOS net=$+22,431 (>0 required); SR0(luck benchmark)=0.2180 vs raw SR=0.1443/trade WalkForward FAIL 309 EDGE IS REAL BUT EVENT-CONCENTRATED OUT-OF-SAMPLE; falsifier(s) fired: F_D_eligible_worse MonteCarlo PASS 411 P(DD≥30%)=2.1% (≤5%), P(DD≥50%)=0.0% (≤1%) at $20,000 CostStress PASS 309 OOS stays $+22,761 at 2× honest costs (breakeven ≈ 10.5× — not cost-fragile) PBO INSUFFICIENT DATA 0 per-trial return streams were not recorded by the search engine; PBO requires the T×M trial matrix. Not a pass.
Headline index (secondary summary — never stands alone)
50.0/100 — secondary summary over 4 gating layers (2 PASS) per HS-1.1; characterization rows inform edge type, never the score.
Uncertainty bands
metric low high sign check total_oos_net_usd +$22,431 +$25,449 both ends same sign
Walk-forward folds (as measured)
OOS window n net PF
max DD IS PF deploy-eligible 2022-01..06 46 +$3,314 1.43 16.2% 1.28 yes 2022-07..12 30 −$189 0.95 15.1% 1.32 yes 2023-01..06 26 −$1,973 0.49 21.7% 1.21 yes 2023-07..12 25 +$95 1.03 10.2% 1.18 yes 2024-01..06 25 −$1,182 0.68 24.9% 1.07 no 2024-07..12 34 +$2,839 1.85 7.3% 0.78 no 2025-01..06 51 +$14,014 2.82 6.3% 0.98 no 2025-07..12 37 +$3,552 1.60 20.5% 1.9 yes 2026-01..06 35 +$1,960 1.23 18.2% 1.93 yes
Headline (pre-registered first): ex-2025 OOS aggregate = +$4,865 .
Concentration: top-fold share 54.4% —
EDGE IS REAL BUT EVENT-CONCENTRATED OUT-OF-SAMPLE.
Monte Carlo cell (at the manifest account)
metric value
P(DD ≥ 30%) 2.08%
P(DD ≥ 50%) 0.00%
ruin probability 0.00%
median max drawdown 13.8%
median ending equity $49,404
Insufficient-data regions (always surfaced, never rounded away)
PBO (n=0, non-load-bearing): per-trial return streams were not recorded by the search engine; PBO requires the T×M trial matrix. Not a pass.
Alpha Algos Research Platform · evidence package v1 ·
per-layer detail and raw arrays in structured_verdict.json alongside this file.
Package SHA-256 (of this HTML, excluding this line's hash placeholder): 196d50dd76a71ee47227be4561c8b156ef80ae19aff7b41e21bbfb3b632b4ce0