← all research
REJECTED

"Holdout Validation — 20 random names (not in flagship universe)"

2026-08-13

Seed: 42 | 20y daily bars | initial $100,000 | region-A config Universe: ARES, ATO, AVY, BLK, CMG, GNRC, GOOGL, HLT, IVZ, JBL, MRNA, NVDA, PAYX, PGR, PNW, PRU, RMD, STX, SWK, WDC

Portfolio results

Year by year

| 2006 | -0.25% | | 2007 | -1.36% | | 2008 | -1.36% | | 2009 | +16.49% | | 2010 | -0.91% | | 2011 | -18.42% | | 2012 | -0.69% | | 2013 | +35.31% | | 2014 | +14.10% | | 2015 | -9.24% | | 2016 | +17.37% | | 2017 | +28.49% | | 2018 | -25.30% | | 2019 | +16.81% | | 2020 | -15.23% | | 2021 | +11.97% | | 2022 | -37.17% | | 2023 | +14.21% | | 2024 | +20.98% | | 2025 | -5.95% | | 2026 | -3.05% |

35% lock comparison (pre-fix max_drawdown_total 0.35)

Verdict

MIXED / INCONCLUSIVE — random unseen names returned +21.30% vs SPY +753.62% (P/L 1.06).

Breaker effect: the 35% lock retired the strategy for 2023-26 (-5.20%, 1301 total trades) vs the live 0.50 setting (+26.51%, 2170 trades).