CU · per-country backtest
Cuba forecast vs reality
Every Voidly Sentinel shutdown-risk forecast for Cuba, plotted against what actually happened. 112 (predicted, observed) pairs from the rolling 30-day evaluation window.
Updated every 30 min · CC BY 4.0 · Raw JSON · Current forecast →
Forecasts evaluated
112
since Apr 17
Accuracy @ 0.5
41.1%
46/112 correct
Brier score
0.261
lower is better
Observed positive rate
16%
mean predicted 28%
Forecast time series
Blue line: forecast probability. Green ✓ markers: forecast was right. Red ✗ markers: forecast was wrong. Dashed line at 0.5 is the binary decision threshold.
Y axis: forecast probability · Faint tick: distance to observed outcome (0 or 1)
All 112 predictions (newest first)
| Eval date | Forecast | Pred ≥ 0.5? | Observed? | Correct? |
|---|---|---|---|---|
| Aug 6 | 80.0% | ↑ | no event | ✗ |
| Aug 5 | 77.7% | ↑ | no event | ✗ |
| Aug 4 | 76.8% | ↑ | no event | ✗ |
| Aug 3 | 73.7% | — | no event | ✓ |
| Aug 2 | 36.6% | ↑ | no event | ✗ |
| Aug 1 | 37.3% | ↑ | no event | ✗ |
| Jul 31 | 31.4% | ↑ | no event | ✗ |
| Jul 30 | 34.2% | ↑ | no event | ✗ |
| Jul 29 | 33.5% | ↑ | no event | ✗ |
| Jul 28 | 39.3% | ↑ | no event | ✗ |
| Jul 27 | 34.3% | ↑ | no event | ✗ |
| Jul 26 | 43.1% | ↑ | no event | ✗ |
| Jul 25 | 43.0% | ↑ | no event | ✗ |
| Jul 24 | 43.4% | ↑ | no event | ✗ |
| Jul 23 | 32.3% | ↑ | no event | ✗ |
| Jul 22 | 46.0% | ↑ | no event | ✗ |
| Jul 21 | 43.5% | ↑ | no event | ✗ |
| Jul 20 | 43.6% | ↑ | no event | ✗ |
| Jul 19 | 33.8% | ↑ | no event | ✗ |
| Jul 18 | 30.5% | ↑ | no event | ✗ |
| Jul 17 | 35.0% | ↑ | no event | ✗ |
| Jul 16 | 26.8% | ↑ | no event | ✗ |
| Jul 15 | 64.3% | ↑ | no event | ✗ |
| Jul 14 | 61.3% | ↑ | no event | ✗ |
| Jul 13 | 70.6% | ↑ | no event | ✗ |
| Jul 12 | 75.5% | ↑ | no event | ✗ |
| Jul 11 | 75.9% | ↑ | no event | ✗ |
| Jul 10 | 72.7% | ↑ | no event | ✗ |
| Jul 9 | 76.3% | ↑ | no event | ✗ |
| Jul 8 | 73.4% | ↑ | no event | ✗ |
| Jul 7 | 74.5% | ↑ | no event | ✗ |
| Jul 6 | 74.5% | ↑ | no event | ✗ |
| Jul 5 | 80.5% | ↑ | no event | ✗ |
| Jul 4 | 76.9% | ↑ | no event | ✗ |
| Jul 3 | 74.8% | ↑ | no event | ✗ |
| Jul 2 | 80.1% | ↑ | no event | ✗ |
| Jul 1 | 71.2% | ↑ | no event | ✗ |
| Jun 30 | 76.7% | ↑ | shutdown | ✓ |
| Jun 29 | 76.7% | ↑ | shutdown | ✓ |
| Jun 28 | 64.2% | ↑ | shutdown | ✓ |
| Jun 27 | 64.2% | ↑ | shutdown | ✓ |
| Jun 26 | 64.2% | ↑ | no event | ✗ |
| Jun 25 | 64.2% | ↑ | no event | ✗ |
| Jun 24 | 64.1% | ↑ | no event | ✗ |
| Jun 23 | 64.1% | ↑ | no event | ✗ |
| Jun 22 | 61.8% | ↑ | no event | ✗ |
| Jun 21 | 5.3% | ↑ | no event | ✗ |
| Jun 20 | 4.6% | — | no event | ✓ |
| Jun 19 | 5.4% | ↑ | no event | ✗ |
| Jun 18 | 5.2% | ↑ | no event | ✗ |
| Jun 17 | 4.9% | — | no event | ✓ |
| Jun 16 | 6.7% | — | no event | ✓ |
| Jun 15 | 3.4% | — | no event | ✓ |
| Jun 14 | 4.1% | — | no event | ✓ |
| Jun 13 | 4.4% | — | no event | ✓ |
| Jun 12 | 5.1% | — | no event | ✓ |
| Jun 11 | 5.2% | — | no event | ✓ |
| Jun 10 | 3.8% | — | no event | ✓ |
| Jun 9 | 6.3% | — | no event | ✓ |
| Jun 8 | 5.4% | — | no event | ✓ |
How to read this
- Each row is one historical forecast. We made the prediction at eval_date, then waited the 7-day horizon, then graded against the observed outcome.
- Forecast % is the calibrated probability we published that day. Post-recalibration (2026-05-20) these now match actual observed rates much better — see the refit finding.
- Pred ≥ 0.5 shows the binary alert decision. Note we usually fire alerts at a lower threshold (see /v1/sentinel/global_heatmap for the live cutoff) — the 0.5 column here is for backtest scoring consistency with the global confusion matrix.
- Correct = both (pred ≥ 0.5) and (observed) agree.
Related
- /atlas/forecast/cu — current 7-day calibrated forecast for Cuba with SHAP drivers
- /cu — Cuba country page (current state, recent incidents)
- /sentinel/backtest — global reliability diagram + per-country comparison table
- Calibration refit writeup