voidly
ET · per-country backtest

Ethiopia forecast vs reality

Every Voidly Sentinel shutdown-risk forecast for Ethiopia, 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
58.9%
66/112 correct
Brier score
0.211
lower is better
Observed positive rate
17%
mean predicted 26%

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 625.3%no event
Aug 520.4%no event
Aug 423.3%no event
Aug 322.7%no event
Aug 214.5%no event
Aug 119.6%no event
Jul 3117.2%no event
Jul 3014.7%no event
Jul 2928.6%no event
Jul 2835.6%no event
Jul 2728.3%no event
Jul 2635.1%no event
Jul 2529.9%no event
Jul 2443.8%no event
Jul 2340.4%no event
Jul 2240.8%no event
Jul 2154.8%no event
Jul 203.8%no event
Jul 1945.4%no event
Jul 1869.8%no event
Jul 1757.6%no event
Jul 1654.1%no event
Jul 1558.5%no event
Jul 1462.5%no event
Jul 1360.6%no event
Jul 1268.2%no event
Jul 1167.0%no event
Jul 1059.0%no event
Jul 959.0%no event
Jul 871.3%shutdown
Jul 769.3%shutdown
Jul 672.7%shutdown
Jul 571.1%shutdown
Jul 473.2%shutdown
Jul 361.2%no event
Jul 269.3%no event
Jul 175.6%no event
Jun 3066.0%no event
Jun 2966.1%no event
Jun 2869.1%no event
Jun 2768.0%no event
Jun 2673.3%no event
Jun 2562.4%no event
Jun 2461.0%no event
Jun 2359.7%no event
Jun 2268.6%no event
Jun 214.9%no event
Jun 203.4%no event
Jun 194.1%no event
Jun 183.6%no event
Jun 174.5%no event
Jun 164.4%no event
Jun 154.3%no event
Jun 145.2%no event
Jun 133.9%no event
Jun 124.2%no event
Jun 114.2%no event
Jun 105.3%no event
Jun 93.9%no event
Jun 85.5%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