Key numbers
| Measure | Value |
|---|---|
| Tests | 88 |
| Idea variants | 22 |
| Beat benchmark | 43.2% |
| Median Sharpe vs benchmark | −0.02 |
| Looked good before 2015 | 11 |
| … and held after | 45.5% |
| Verdict | No consistent edge |
By asset class
| Asset class | Tests | Beat benchmark | Median Sharpe vs bm | Verdict |
|---|---|---|---|---|
| US equities | 49 | 37% | −0.04 | No consistent edge |
| international equities | 29 | 55% | +0.01 | No consistent edge |
Caveats
The 2015–2026 test window has been looked at many times during research, so we label it and do not call it a clean holdout. "Beat benchmark" compares each test with a fixed asset mix or buy-and-hold in the same window; a weak benchmark flatters the result. Many tests use US-listed ETFs.
Terms used: base rate, out-of-sample test, placebo test, dead end.
Let your AI check how ideas like yours did before it runs a single test. In our pilot, agents that checked first spent 1% of their backtests on known dead ends instead of 15%, and every backtest not wasted is time and tokens you keep. Cite this entry as ER-BEH-ALL, as of 2026-11-01.
Related evidence
Questions
Do behavioural signals work?
Behavioural signals were tested 88 times across 22 idea variants. 43% beat their benchmark in 2015–2026 after costs, with a median Sharpe difference of −0.02. Verdict as of 2026-11-01: no consistent edge.
Is this investment advice?
No. These are research statistics, not recommendations.