voidly
CN · per-country backtest

China forecast vs reality

Every Voidly Sentinel shutdown-risk forecast for China, plotted against what actually happened. 60 (predicted, observed) pairs from the rolling 30-day evaluation window.

Updated every 30 min · CC BY 4.0 · Raw JSON · Current forecast →

Forecasts evaluated
60
since Apr 17
Accuracy @ 0.5
25.0%
15/60 correct
Brier score
0.370
lower is better
Observed positive rate
37%
mean predicted 12%

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.

0.000.250.500.751.00threshold 0.50Apr 17Jun 8

Y axis: forecast probability · Faint tick: distance to observed outcome (0 or 1)

All 60 predictions (newest first)

Eval dateForecastPred ≥ 0.5?Observed?Correct?
Jun 85.4%no event
Jun 74.9%no event
Jun 64.9%no event
Jun 55.0%no event
Jun 410.0%no event
Jun 39.6%no event
Jun 212.1%no event
Jun 112.8%no event
May 3160.0%no event
May 3110.7%no event
May 3014.3%no event
May 2913.5%no event
May 2880.0%no event
May 2827.9%no event
May 2815.3%no event
May 2780.0%no event
May 2711.0%no event
May 269.6%no event
May 2526.2%no event
May 258.2%no event
May 2413.9%no event
May 248.3%no event
May 2360.0%no event
May 237.1%no event
May 226.6%no event
May 219.2%no event
May 208.2%no event
May 196.5%no event
May 189.0%no event
May 175.6%no event
May 167.5%no event
May 155.0%no event
May 146.5%no event
May 136.6%no event
May 125.1%no event
May 116.1%shutdown
May 106.3%shutdown
May 95.9%shutdown
May 85.2%shutdown
May 76.8%shutdown
May 64.9%shutdown
May 54.6%no event
May 46.7%no event
May 36.4%no event
May 25.5%shutdown
May 15.4%shutdown
Apr 305.9%shutdown
Apr 296.7%shutdown
Apr 285.2%shutdown
Apr 274.5%shutdown
Apr 264.2%shutdown
Apr 254.9%shutdown
Apr 246.2%shutdown
Apr 234.5%shutdown
Apr 223.7%shutdown
Apr 213.7%shutdown
Apr 204.6%shutdown
Apr 195.3%shutdown
Apr 183.7%shutdown
Apr 173.9%shutdown

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