What we tested and how it ended
Every idea is written down and locked before it is tested. Click a verdict to filter the table.
Daily model forecast for every coin →
All hypotheses
Which rule works on which coin
Every rule tested in any hypothesis, applied to every coin since 2022-01-01, fees included. This is exploration, not verdicts: with this many combinations some look good by pure luck, so a false-discovery check is applied to the whole table. A blue dot marks a coin the original hypothesis was about. A green ✓ on another coin is a candidate for a new test on future data.
- ✓works (passed the chance check)
- ↑better than holding, could be luck
- ✗worse than holding
- ·not enough data
- original coin of the hypothesis
| Rule | BTC | ETH | SOL | BNB | XRP | ADA | LTC | LINK | DOGE | XLM | TRX | EVER | GRAM |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| H-005Hold while price is above its 50-day average | ↑ | ↑ | ↑ | ↑ | ✗ | ↑ | ✗ | ✗ | ↑ | ↑ | ✗ | ↑ | ↑ |
| H-006Hold while stablecoin growth is above normal | ↑ | ↑ | ↑ | ↑ | ↑ | ✗ | ↑ | ↑ | ✗ | ↑ | ✗ | ✗ | ↑ |
| H-007Leave when BTC funding is unusually high | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
| H-008Leave when Fear & Greed is 75 or more | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ↑ | ↑ | ✗ |
| H-009Hold only after net inflows to BTC ETFs | ↑ | ↑ | ✗ | ↑ | ✗ | ✗ | ✗ | ✗ | ↑ | ✗ | ✗ | ✗ | ↑ |
| H-012Volatility targeting (30% a year) | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ✗ | ✗ |
| H-0147-day momentum | ✗ | ↑ | ↑ | ✗ | ✗ | ✗ | ✗ | ✗ | ↑ | ↑ | ✗ | ↑ | ✗ |
| H-015Hold only on Mondays | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
| H-016Hold only around the turn of the month | ✗ | ✗ | ✗ | ✗ | ✗ | ↑ | ↑ | ✗ | ✗ | ✗ | ✗ | ↑ | ✗ |
| H-017Hold only while BTC is above its 200-day average | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ✗ | ↑ | ✗ |
| H-020Leave on heavy BTC deposits to exchanges | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ↑ | ✗ | ✗ |
| H-021Leave on heavy ETH deposits to exchanges | ↑ | ↑ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ↑ | ↑ | ✗ |
How we test
- Idea written down
- Locked with a hash
- Tested out of sample
- Verdict
- Re-tested on new data