voidly
BD · per-country backtest

Bangladesh forecast vs reality

Every Voidly Sentinel shutdown-risk forecast for Bangladesh, 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
46.4%
52/112 correct
Brier score
0.520
lower is better
Observed positive rate
54%
mean predicted 4%

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 61.4%shutdown
Aug 51.6%shutdown
Aug 42.7%shutdown
Aug 31.1%shutdown
Aug 21.5%shutdown
Aug 11.7%shutdown
Jul 312.6%shutdown
Jul 302.3%shutdown
Jul 291.1%shutdown
Jul 280.9%shutdown
Jul 270.8%shutdown
Jul 260.7%shutdown
Jul 250.7%shutdown
Jul 240.9%shutdown
Jul 231.0%shutdown
Jul 221.3%shutdown
Jul 211.1%shutdown
Jul 204.4%shutdown
Jul 190.7%shutdown
Jul 181.8%shutdown
Jul 171.0%shutdown
Jul 161.5%shutdown
Jul 150.5%shutdown
Jul 141.3%shutdown
Jul 131.3%no event
Jul 120.7%no event
Jul 111.1%shutdown
Jul 101.8%shutdown
Jul 91.8%shutdown
Jul 81.7%shutdown
Jul 71.9%shutdown
Jul 61.2%shutdown
Jul 51.4%shutdown
Jul 42.2%shutdown
Jul 31.4%shutdown
Jul 21.8%no event
Jul 12.0%no event
Jun 303.2%shutdown
Jun 292.5%shutdown
Jun 281.3%no event
Jun 272.2%no event
Jun 262.5%no event
Jun 251.4%no event
Jun 242.2%no event
Jun 231.5%no event
Jun 221.3%no event
Jun 214.4%no event
Jun 204.9%no event
Jun 195.7%no event
Jun 184.3%no event
Jun 174.2%no event
Jun 165.9%no event
Jun 153.7%no event
Jun 145.0%no event
Jun 134.7%no event
Jun 122.8%no event
Jun 113.8%no event
Jun 103.3%no event
Jun 93.1%shutdown
Jun 83.8%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