Signal performance
Do the consensus sweeps actually beat the market? Here’s the honest answer — method first, numbers with their sample sizes, nothing cherry-picked.
Past results on a small, growing sample — descriptive, not a forecast. Not financial advice.
How this is computed
- We don’t keep a signal archive, so we reconstruct historical sweeps from kept trade history using the exact live definition (≥2 tracked traders entering the same outcome within 15 minutes).
- For each reconstructed sweep we check the market’s resolved outcome — did the swept side actually win?
- We count only resolved sweeps with a usable entry price. Still-open and unpriced sweeps are excluded — no survivorship, no cherry-picking.
- “Following” = buy one share at the sweep’s average entry price; a win pays +(1−p), a loss costs −p.
- Edge = hit rate − the market-implied baseline (the average entry price, i.e. the market’s own win probability at signal time). ROI = total PnL ÷ total cost deployed.
By category
as of 2026-11-02 · categories with fewer than 30 resolved sweeps are greyed — too thin to read into.
| Category | Hit rate | Implied | Edge | ROI | N (resolved) |
|---|---|---|---|---|---|
| Crypto | 47.2% | 49.8% | −2.7pp | −5.3% | 53 |
| Sportsinsufficient sample | 33.3% | 49.1% | −15.8pp | −32.1% | 9 |
| Overall | 45.2% | 49.7% | −4.6pp | −9.2% | 62 |
This is preliminary. The overall number is one small, growing sample (N = 62) and it swings as more sweeps resolve — don’t read a single figure as a verdict. We show every category’s sample size so you can weight it yourself. Information only, not financial advice.