voidly
Sentinel · 30-day backtest

Forecasts meet outcomes.

Compare forecasts with outcomes. Calibration measures probability accuracy; it does not establish shutdown-onset skill.

Updated every 30 min · last refresh Sep 9, 2026 · CC BY 4.0 · Binned JSON · Raw outcomes

Reliability diagram

0.000.000.250.250.500.500.750.751.001.00perfectn=448n=15n=40n=49n=37n=58n=92n=131n=30Predicted probability (bin mean)Observed positive rate

Each point is one prediction bin. X axis is the mean predicted probability inside the bin; Y axis is the fraction of those forecasts where the real outcome actually happened. Perfect calibration is the diagonal line — points above the line mean the model UNDER-estimates risk; points below mean it OVER-estimates.

Bubble area scales with bin count · Red = model under-estimated · Blue = model over-estimated

Brier score
0.296
lower is better
Calibration MAE
0.344
0 = perfect
Accuracy
57.0%
900 evaluated
F1 (binary 0.5)
0.11
P=0.08 R=0.15

Per-bin breakdown

BinPredicted meanObserved rateΔn
[0.0, 0.1)0.0320.237+0.205448
[0.1, 0.2)0.1600.000-0.16015
[0.2, 0.3)0.2560.150-0.10640
[0.3, 0.4)0.3440.204-0.14049
[0.4, 0.5)0.4530.162-0.29137
[0.5, 0.6)0.5520.069-0.48358
[0.6, 0.7)0.6520.065-0.58792
[0.7, 0.8)0.7460.061-0.685131
[0.8, 0.9)0.8340.100-0.73430

Δ = observed − predicted. The 0.1 bin holds 448 of the 900 forecasts — this is where most action happens, and where the May 20 isotonic recalibration was aimed. See /sentinel/calibration for the time-series view of how this gap evolves day over day.

Per-country backtest (worst Brier first, n ≥ 5)

CountryBrierAccuracyPRnPos rate
BangladeshBD0.0030
BrazilBR0.0030
BelarusBY0.0030
ChinaCN30
CubaCU0.0030
EgyptEG0.201.0030
ERER0.0030
EthiopiaET0.0030
IndonesiaID0.0030
IndiaIN0.0030
IranIR0.130.4030
North KoreaKP0.0030
KazakhstanKZ0.520.7530
LebanonLB30
MyanmarMM0.0030
MalaysiaMY0.0030
NigeriaNG0.0030
NicaraguaNI0.0030
PhilippinesPH0.0030
PakistanPK0.0030

Countries where the forecast is currently performing worst — useful for targeting feature engineering or seeking expert review.

How to read these numbers

  • Brier score — mean squared error between predicted probability and actual 0/1 outcome. Lower is better. Less than 0.10 is excellent; 0.10-0.30 is OK; above 0.30 is concerning.
  • Calibration MAE — average gap between predicted-mean and observed-rate across bins. 0.00 means the model's probabilities are exactly right on average.
  • Reliability diagram — the visual version of calibration MAE. Bubble size = bin sample count.
  • F1 (P + R) — binary classification metrics at the 0.5 threshold. Useful when downstream decisions are binary (alert / no-alert).
  • The May 20, 2026 isotonic recalibration targeted the 0.1 bin specifically — see the recalibration finding.
About this view

When the Sentinel model says “5% risk,” does the real outcome actually happen ~5% of the time? Below is the answer: 900 live (predicted, observed) pairs from the last 30 days, binned into a standard reliability diagram.

Calibration and onset skill