BR · per-country backtest
Brazil forecast vs reality
Compare published forecasts with observed outcomes for this country.
Updated every 30 min · CC BY 4.0 · Raw JSON · Current forecast →
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)
Forecasts evaluated
163
since Apr 17
Accuracy @ 0.5
57.1%
93/163 correct
Brier score
0.047
lower is better
Observed positive rate
3%
mean predicted 13%
All 163 predictions (newest first)
| Eval date | Forecast | Pred ≥ 0.5? | Observed? | Correct? |
|---|---|---|---|---|
| Sep 26 | 15.7% | — | no event | ✓ |
| Sep 25 | 14.1% | — | no event | ✓ |
| Sep 24 | 9.9% | — | no event | ✓ |
| Sep 23 | 8.7% | — | no event | ✓ |
| Sep 22 | 12.4% | — | no event | ✓ |
| Sep 21 | 32.7% | ↑ | no event | ✗ |
| Sep 20 | 8.6% | ↑ | no event | ✗ |
| Sep 19 | 33.2% | ↑ | no event | ✗ |
| Sep 18 | 30.9% | ↑ | shutdown | ✓ |
| Sep 17 | 30.5% | ↑ | shutdown | ✓ |
| Sep 16 | 8.7% | ↑ | shutdown | ✓ |
| Sep 15 | 31.5% | ↑ | shutdown | ✓ |
| Sep 14 | 30.8% | ↑ | shutdown | ✓ |
| Sep 13 | 7.6% | ↑ | no event | ✗ |
| Sep 12 | 28.9% | ↑ | no event | ✗ |
| Sep 11 | 30.4% | ↑ | no event | ✗ |
| Sep 10 | 7.3% | ↑ | no event | ✗ |
| Sep 9 | 32.0% | ↑ | no event | ✗ |
| Sep 8 | 30.5% | ↑ | no event | ✗ |
| Sep 7 | 31.3% | ↑ | no event | ✗ |
| Sep 6 | 7.3% | ↑ | no event | ✗ |
| Sep 5 | 6.5% | ↑ | no event | ✗ |
| Sep 4 | 7.0% | ↑ | no event | ✗ |
| Sep 3 | 6.4% | ↑ | no event | ✗ |
| Sep 2 | 5.7% | ↑ | no event | ✗ |
| Sep 1 | 5.6% | ↑ | no event | ✗ |
| Aug 31 | 7.0% | ↑ | no event | ✗ |
| Aug 30 | 32.5% | ↑ | no event | ✗ |
| Aug 29 | 31.2% | ↑ | no event | ✗ |
| Aug 28 | 29.8% | ↑ | no event | ✗ |
| Aug 27 | 30.1% | ↑ | no event | ✗ |
| Aug 26 | 6.8% | ↑ | no event | ✗ |
| Aug 25 | 32.2% | ↑ | no event | ✗ |
| Aug 24 | 5.1% | — | no event | ✓ |
| Aug 23 | 4.0% | — | no event | ✓ |
| Aug 22 | 4.9% | — | no event | ✓ |
| Aug 21 | 4.1% | — | no event | ✓ |
| Aug 20 | 5.5% | ↑ | no event | ✗ |
| Aug 19 | 3.8% | — | no event | ✓ |
| Aug 18 | 4.3% | — | no event | ✓ |
| Aug 17 | 4.3% | — | no event | ✓ |
| Aug 16 | 4.0% | — | no event | ✓ |
| Aug 15 | 3.2% | — | no event | ✓ |
| Aug 14 | 4.4% | — | no event | ✓ |
| Aug 13 | 4.9% | — | no event | ✓ |
| Aug 12 | 4.8% | — | no event | ✓ |
| Aug 11 | 4.7% | — | no event | ✓ |
| Aug 10 | 4.5% | — | no event | ✓ |
| Aug 9 | 2.7% | — | no event | ✓ |
| Aug 8 | 28.4% | ↑ | no event | ✗ |
| Aug 7 | 4.9% | — | no event | ✓ |
| Aug 6 | 5.3% | ↑ | no event | ✗ |
| Aug 5 | 27.9% | ↑ | no event | ✗ |
| Aug 4 | 3.2% | — | no event | ✓ |
| Aug 3 | 5.5% | ↑ | no event | ✗ |
| Aug 2 | 27.6% | ↑ | no event | ✗ |
| Aug 1 | 3.6% | — | no event | ✓ |
| Jul 31 | 36.2% | ↑ | no event | ✗ |
| Jul 30 | 36.5% | ↑ | no event | ✗ |
| Jul 29 | 34.2% | ↑ | 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/br — current 7-day calibrated forecast for Brazil with SHAP drivers
- /br — Brazil country page (current state, recent incidents)
- /sentinel/backtest — global reliability diagram + per-country comparison table
- Calibration refit writeup
About this view
Every Voidly Sentinel shutdown-risk forecast for Brazil, plotted against what actually happened. 163 (predicted, observed) pairs from the rolling 30-day evaluation window.