Correction (2026-10-06): The +92% trend-filter figure below left out trading fees and was measured on the same data used to choose the 50-day setting. Our own out-of-sample check, after fees, found the filter lost less than simply holding (−36% vs −52% median per coin), but it still lost. Lines like "'X' leads on auc" show the best of many coin-flip results, not a real edge. The grid-suitability result was also misread: the score ranks coins backwards, so the coins it rated best lost the most.
The Lab ran 5 experiments across 5 research threads this cycle (2026-08-17 → 2026-08-24) on the idle GPUs — every verdict below is published as the data said it, wins, losses and coin-flips alike. No cherry-picking.
Which crypto coins can a simple ML model actually predict?
- run of 2026-08-17 — 'SOL' leads on auc: +0.519
We Audited Our Own "Possible Edge." It Died in an Hour.
- run of 2026-08-18 — We flagged a 109bps/trade edge last week. A methodology audit killed it: leaky early-stopping + overlapping trades. −37 bps under honest accounting.
Which coins are actually good for grids?
- run of 2026-08-23 — 'bottom (cull?)' leads on mean_grid_pnl: -7.16
Is a simple trend filter better than the grid?
- run of 2026-08-23 — 'trend_follow_50' leads on mean_return_pct: +91.50
Would pausing grids in a BTC downtrend stop the bleed?
- run of 2026-08-24 — 'BTC-downtrend' leads on total_pnl: -16.17
Auto-published weekly by Gordon Lab. Each entry is a real backtest or model run on real market data; methods and full result tables live in the lab archive. This is research, not financial advice.