Updated monthly · free to cite
Crypto Overfitting Index
Every month our engine re-tests 10 classic strategy templates on 50 coins and splits each history 70/30. The index is the share of configurations whose out-of-sample Sharpe turned negative — how much of what backtests promise fails on data it never saw.
July 2026 reading
54.6
0 = every strategy held up out-of-sample · 100 = every strategy failed
History
| Month | Index | OOS-positive share | Median Sharpe haircut |
|---|---|---|---|
| July 2026 | 54.6 | 45.4% | 0.21 |
Methodology
Each month, every template × coin combination is backtested on daily bars since 2021 with default parameters, $10,000 and standard fees. The history is split chronologically 70/30; a configuration counts as failed when its out-of-sample Sharpe is ≤ 0. The index is the failed share, in percent. Same engine, same rules, every month — the level matters less than the trend.
Embed the badge
The badge always shows the latest reading. Paste this HTML anywhere — please keep the attribution link.
<a href="https://noonbarbari.xyz/en/overfitting-index" title="Crypto Overfitting Index — Noon Barbari"> <img src="https://noonbarbari.xyz/api/public/v1/overfitting-index/badge" alt="Crypto Overfitting Index" height="20"> </a>
Cite it
Free to quote and reprint with attribution (CC BY 4.0). Suggested wording:
Crypto Overfitting Index, Noon Barbari — https://noonbarbari.xyz/en/overfitting-index
Press: monthly readings are published automatically around the 1st. For questions or the underlying per-run data, see the open dataset or contact us.
Auto-generated from hypothetical backtests of the stated past periods. Educational statistics, not financial advice; past performance does not guarantee future results.