PK · per-country backtest
Pakistan forecast vs reality
Every Voidly Sentinel shutdown-risk forecast for Pakistan, 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
25.0%
28/112 correct
Brier score
0.580
lower is better
Observed positive rate
34%
mean predicted 59%
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 | 75.5% | ↑ | no event | ✗ |
| Aug 5 | 68.6% | ↑ | no event | ✗ |
| Aug 4 | 69.7% | ↑ | no event | ✗ |
| Aug 3 | 80.7% | ↑ | no event | ✗ |
| Aug 2 | 78.9% | ↑ | no event | ✗ |
| Aug 1 | 78.0% | ↑ | no event | ✗ |
| Jul 31 | 76.6% | ↑ | no event | ✗ |
| Jul 30 | 75.8% | ↑ | no event | ✗ |
| Jul 29 | 73.0% | ↑ | no event | ✗ |
| Jul 28 | 69.8% | ↑ | no event | ✗ |
| Jul 27 | 71.1% | ↑ | no event | ✗ |
| Jul 26 | 80.7% | ↑ | no event | ✗ |
| Jul 25 | 75.8% | ↑ | no event | ✗ |
| Jul 24 | 73.9% | ↑ | no event | ✗ |
| Jul 23 | 86.3% | ↑ | no event | ✗ |
| Jul 22 | 71.0% | ↑ | shutdown | ✓ |
| Jul 21 | 84.0% | ↑ | shutdown | ✓ |
| Jul 20 | 74.6% | ↑ | shutdown | ✓ |
| Jul 19 | 69.5% | ↑ | shutdown | ✓ |
| Jul 18 | 75.1% | ↑ | no event | ✗ |
| Jul 17 | 69.8% | ↑ | no event | ✗ |
| Jul 16 | 71.0% | ↑ | no event | ✗ |
| Jul 15 | 74.0% | ↑ | no event | ✗ |
| Jul 14 | 76.0% | ↑ | no event | ✗ |
| Jul 13 | 68.9% | ↑ | no event | ✗ |
| Jul 12 | 76.4% | ↑ | no event | ✗ |
| Jul 11 | 76.7% | ↑ | no event | ✗ |
| Jul 10 | 80.4% | ↑ | no event | ✗ |
| Jul 9 | 81.6% | ↑ | no event | ✗ |
| Jul 8 | 78.3% | ↑ | no event | ✗ |
| Jul 7 | 81.2% | ↑ | no event | ✗ |
| Jul 6 | 78.9% | ↑ | no event | ✗ |
| Jul 5 | 80.6% | ↑ | no event | ✗ |
| Jul 4 | 86.8% | ↑ | no event | ✗ |
| Jul 3 | 78.6% | ↑ | no event | ✗ |
| Jul 2 | 78.6% | ↑ | no event | ✗ |
| Jul 1 | 73.7% | ↑ | no event | ✗ |
| Jun 30 | 73.7% | ↑ | shutdown | ✓ |
| Jun 29 | 66.5% | — | shutdown | ✗ |
| Jun 28 | 68.5% | ↑ | shutdown | ✓ |
| Jun 27 | 64.7% | ↑ | shutdown | ✓ |
| Jun 26 | 69.3% | — | shutdown | ✗ |
| Jun 25 | 76.4% | — | no event | ✓ |
| Jun 24 | 76.4% | — | no event | ✓ |
| Jun 23 | 76.7% | ↑ | no event | ✗ |
| Jun 22 | 79.5% | ↑ | no event | ✗ |
| Jun 21 | 95.0% | ↑ | no event | ✗ |
| Jun 20 | 95.0% | ↑ | no event | ✗ |
| Jun 19 | 95.0% | ↑ | no event | ✗ |
| Jun 18 | 95.0% | ↑ | no event | ✗ |
| Jun 17 | 95.0% | ↑ | no event | ✗ |
| Jun 16 | 95.0% | ↑ | no event | ✗ |
| Jun 15 | 95.0% | ↑ | no event | ✗ |
| Jun 14 | 95.0% | ↑ | no event | ✗ |
| Jun 13 | 95.0% | ↑ | no event | ✗ |
| Jun 12 | 95.0% | ↑ | no event | ✗ |
| Jun 11 | 95.0% | ↑ | no event | ✗ |
| Jun 10 | 95.0% | ↑ | no event | ✗ |
| Jun 9 | 95.0% | ↑ | no event | ✗ |
| Jun 8 | 95.0% | ↑ | 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/pk — current 7-day calibrated forecast for Pakistan with SHAP drivers
- /pk — Pakistan country page (current state, recent incidents)
- /sentinel/backtest — global reliability diagram + per-country comparison table
- Calibration refit writeup