Forecasts meet outcomes.
Compare forecasts with outcomes. Calibration measures probability accuracy; it does not establish shutdown-onset skill.
Updated every 30 min · last refresh Oct 3, 2026 · CC BY 4.0 · Binned JSON · Raw outcomes
Reliability diagram
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
Per-bin breakdown
| Bin | Predicted mean | Observed rate | Δ | n |
|---|---|---|---|---|
| [0.0, 0.1) | 0.032 | 0.261 | +0.229 | 448 |
| [0.1, 0.2) | 0.137 | 0.000 | -0.137 | 26 |
| [0.2, 0.3) | 0.272 | 0.125 | -0.147 | 48 |
| [0.3, 0.4) | 0.342 | 0.104 | -0.238 | 48 |
| [0.4, 0.5) | 0.455 | 0.071 | -0.383 | 42 |
| [0.5, 0.6) | 0.551 | 0.071 | -0.480 | 56 |
| [0.6, 0.7) | 0.649 | 0.040 | -0.609 | 75 |
| [0.7, 0.8) | 0.756 | 0.050 | -0.707 | 101 |
| [0.8, 0.9) | 0.837 | 0.000 | -0.837 | 52 |
| [0.9, 1.0) | 0.941 | 0.000 | -0.941 | 4 |
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)
| Country | Brier | Accuracy | P | R | n | Pos rate |
|---|---|---|---|---|---|---|
| BangladeshBD | — | — | 1.00 | 0.03 | 30 | — |
| BrazilBR | — | — | 0.20 | 1.00 | 30 | — |
| BelarusBY | — | — | 0.00 | — | 30 | — |
| ChinaCN | — | — | 0.00 | — | 30 | — |
| CubaCU | — | — | 0.00 | — | 30 | — |
| EgyptEG | — | — | 0.04 | 1.00 | 30 | — |
| ERER | — | — | 0.00 | — | 30 | — |
| EthiopiaET | — | — | 0.00 | — | 30 | — |
| IndonesiaID | — | — | 0.33 | 0.05 | 30 | — |
| IndiaIN | — | — | — | 0.00 | 30 | — |
| IranIR | — | — | 0.00 | 0.00 | 30 | — |
| North KoreaKP | — | — | 0.00 | — | 30 | — |
| KazakhstanKZ | — | — | 0.20 | 1.00 | 30 | — |
| LebanonLB | — | — | — | — | 30 | — |
| MyanmarMM | — | — | 0.00 | — | 30 | — |
| MalaysiaMY | — | — | 1.00 | 0.05 | 30 | — |
| NigeriaNG | — | — | 0.00 | — | 30 | — |
| NicaraguaNI | — | — | — | — | 30 | — |
| PhilippinesPH | — | — | 0.00 | — | 30 | — |
| PakistanPK | — | — | 0.00 | — | 30 | — |
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.
Related
- /sentinel/calibration — 90-day time series of empirical coverage vs the 90% conformal target
- /methodology#validation — the three honest accuracy splits (LOCO, stratified, time-based)
- /atlas/forecast/IR — per-country calibrated forecast detail with SHAP drivers
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.