The bet: buy the dip / mean-reversion.
All eight coins came back with a negative out-of-sample Sharpe and a trade-level t-statistic below -2 — a statistically significant loss, not a bad run. The in-sample and out-of-sample results were about equal, which means it never had an edge to lose: this is "never had it," not decay. The BTC and ETH controls were the worst of the set.
A faithful port of superbot’s dominant live strategy: buy deeper as RSI falls (dip-buying / mean-reversion), long-only. On this universe the dips kept going, so averaging into weakness just accumulated losers — and a 0.30% round-trip cost across ~400 trades per coin ground away whatever was left.
| Pair | Sharpe | PF | trade t | n |
|---|---|---|---|---|
| ADA | -0.28 | 0.57 | -3.53 | 381 |
| XRP | -0.32 | 0.56 | -3.76 | 403 |
| TON | -0.19 | 0.69 | -2.24 | 404 |
| SOL | -0.22 | 0.69 | -2.48 | 415 |
| AVAX | -0.25 | 0.67 | -2.75 | 442 |
| LINK | -0.17 | 0.72 | -2.12 | 393 |
| BTC | -0.35 | 0.49 | -4.19 | 407 |
| ETH | -0.23 | 0.67 | -2.53 | 424 |
⚠ = underpowered (fewer than 30 out-of-sample trades). Negative values in red.
This was our live flagship. A 14-day live run lost -$0.37 over 110 trades at a 32.7% win rate — the backtest verdict reproduced in production. The runner was stopped 2026-06-02 and is now parked. We keep a small version running and fully visible on the track record on purpose, specifically because we graded it negative — rather than quietly retiring the inconvenient one.
Losing and statistically significant on all 8 coins. This was our live flagship.
Every trade we place — wins and losses, from real money — is on the live track record. This page is the research that decides what we will and won't run.
See the live track record →