Leaderboard
Methodology · Public
How we score. In public.
One formula, five components, capped 0–100. Every score reproducible from the published pipeline. If a number looks wrong, audit me.
- components
- 5
- candles
- 18.7M
- creators tracked
- 123
- extraction floor
- 70%
01
The pipeline.
Scrape → Extract → Match → Score → Rank
- step 01ScrapeAuto-generated subtitles, daily
- step 02ExtractAI identifies actionable predictions
- step 03MatchEach call ↔ Binance candles
- step 04Score5 components → 0–100
- step 05RankAvg Alpha across scored calls
02
The score.
5 components · max 100 · floor 70%
| Component | Max | How it's earned |
|---|---|---|
| Direction correct | 40 | Bullish call + price up at 30d (or vice versa). |
| Alpha over BTC | 25 | Each 1% excess return = 2.5pt, capped at 25. |
| Specificity | 15 | Entry, target, stop-loss, timeframe (¼ each). |
| Regime difficulty | 10 | Bullish in bear / bearish in bull = max. |
| Target hit | 10 | Stated target reached within 90d. |
| Total | 100 | Sum of components, no rescaling. |
03
Tracked coins.
18 symbols · Binance OHLCV
Calls on these tickers are matched against minute-grained Binance candles. Anything outside this universe is logged but not scored.
BTCETHSOLBNBXRPDOGEADAAVAXDOTLINKTAORENDERFETNEARARINJSUIPENDLE
04
Tier ranges.
S / A / B / C bands
| Tier | Alpha range | Reading |
|---|---|---|
| S | 70 – 100 | Elite — beats BTC consistently with high specificity. |
| A | 55 – 69 | Strong — directional edge, partial specificity. |
| B | 40 – 54 | Mixed — directional hit but thin alpha. |
| C | 0 – 39 | Underperforming — frequent misses or low-conviction. |
05
Audit me.
Reproducible · open source
The recompute pipeline is reproducible. Every score traces to a transcript line, a Binance candle range, and a deterministic formula. There is no hand-tuned weighting per creator.
Disagree with a number? The source is published; clone it, rerun the score, and tell us where it diverges.