EXP-000105

NEXUS — two builds on the identical dataset

Archived Version Research source: Human Date not recorded

System: NEXUS
Source reports: Engish-fullbt - コピー.html · englishi-02fullbt - コピー.html

Historical experiment metadata unavailable. The hypothesis and decision behind this work were never written down; the metrics below are reconstructed from surviving tester reports and the narrative is honestly absent.

Hypothesis

Not recorded

Dataset & conditions

Two runs over the identical window (2018.01.01–2026.03.28, BTCUSD M5) with the identical trade count (19,968) but different profit factor (1.52 vs 1.54) and equity drawdown (2.60% vs 5.56%) — a genuine A/B of two builds on the same data. Which build came first is not recorded.

Before / After

Before

Period: 2018.01.01 – 2026.03.28
Symbol: BTCUSD
Profit factor: 1.52
Max drawdown: 2.6% (equity)
Trades: 19,968
Win rate: 66.52%

After

Period: 2018.01.01 – 2026.03.28
Symbol: BTCUSD
Profit factor: 1.54
Max drawdown: 5.56% (equity)
Trades: 19,968
Win rate: 66.52%

Difference

Profit factor +0.02 Max drawdown +2.96pp Trades 0 Win rate 0pp

Validation

Out-of-sample, walk-forward, Monte Carlo and forward results are shown only where they exist as data. None exist for this entry.

Decision

Archived

Not recorded

Backtest improvement does not imply future improvement. The more experiments run, the more likely some succeed by chance — this log exists partly so that multiple-testing risk stays visible instead of hidden.