Every forecast is written to the database at generation time. When the window ends, the live BTC price is stored and the directional call is marked correct or incorrect. Nothing here is backfilled with fake wins.
Total evaluated
4
Correct
0
Incorrect
4
30-day accuracy
—
Overall accuracy: 0% · 4 pending
Next 5 Minutes
—
0/0 direction
Range —
Next 15 Minutes
—
0/0 direction
Range —
Next 30 Minutes
—
0/0 direction
Range —
Next 45 Minutes
—
0/0 direction
Range —
Next Hour
0%
0/1 direction
Range 0%
Next 4 Hours
0%
0/1 direction
Range 0%
Next 24 Hours
0%
0/1 direction
Range 0%
Next 7 Days
0%
0/1 direction
Range 100%
| Date | Window | Bias | BTC then | After | Result |
|---|---|---|---|---|---|
| Nov 14 | 7d | Bearish — 42% | $66,648 | Pending | Pending |
| Nov 14 | 24h | Bearish — 41% | $66,648 | Pending | Pending |
| Nov 14 | 4h | Bearish — 41% | $66,648 | Pending | Pending |
| Nov 14 | 1h | Bearish — 38% | $66,648 | Pending | Pending |
| Aug 19 | 7d | Neutral — 50% | $69,418 | $66,648 | ✗ Incorrect |
| Aug 19 | 24h | Bullish — 65% | $69,418 | $66,648 | ✗ Incorrect |
| Aug 19 | 4h | Bullish — 65% | $69,418 | $66,648 | ✗ Incorrect |
If the model was bullish, Bitcoin must finish the window higher than the price at generation. If bearish, it must finish lower. Range hits are tracked separately.
There is no history until snapshots age past their evaluation time. That is intentional.
| Aug 19 |
| 1h |
| Bullish — 56% |
| $69,418 |
| $66,648 |
| ✗ Incorrect |