voidly
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.

0.000.250.500.751.00threshold 0.50Apr 17Aug 6

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

All 112 predictions (newest first)

Eval dateForecastPred ≥ 0.5?Observed?Correct?
Aug 680.0%no event
Aug 577.7%no event
Aug 476.8%no event
Aug 373.7%no event
Aug 236.6%no event
Aug 137.3%no event
Jul 3131.4%no event
Jul 3034.2%no event
Jul 2933.5%no event
Jul 2839.3%no event
Jul 2734.3%no event
Jul 2643.1%no event
Jul 2543.0%no event
Jul 2443.4%no event
Jul 2332.3%no event
Jul 2246.0%no event
Jul 2143.5%no event
Jul 2043.6%no event
Jul 1933.8%no event
Jul 1830.5%no event
Jul 1735.0%no event
Jul 1626.8%no event
Jul 1564.3%no event
Jul 1461.3%no event
Jul 1370.6%no event
Jul 1275.5%no event
Jul 1175.9%no event
Jul 1072.7%no event
Jul 976.3%no event
Jul 873.4%no event
Jul 774.5%no event
Jul 674.5%no event
Jul 580.5%no event
Jul 476.9%no event
Jul 374.8%no event
Jul 280.1%no event
Jul 171.2%no event
Jun 3076.7%shutdown
Jun 2976.7%shutdown
Jun 2864.2%shutdown
Jun 2764.2%shutdown
Jun 2664.2%no event
Jun 2564.2%no event
Jun 2464.1%no event
Jun 2364.1%no event
Jun 2261.8%no event
Jun 215.3%no event
Jun 204.6%no event
Jun 195.4%no event
Jun 185.2%no event
Jun 174.9%no event
Jun 166.7%no event
Jun 153.4%no event
Jun 144.1%no event
Jun 134.4%no event
Jun 125.1%no event
Jun 115.2%no event
Jun 103.8%no event
Jun 96.3%no event
Jun 85.4%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