voidly

China CN

stable (-1.3pp)

Follow the model’s 30-day path. Read the probability and uncertainty together.

30-day trajectory with 90% conformal band

model trajectory_v1
0%25%50%75%100%d1d7d14d21d30peak d22 44%
mean P(shutdown by day)90% conformal bandpeak day
Peak day
44.4%
d22 · 2026-10-07
Floor day
3.5%
d3 · 2026-09-18
30-day delta
-1.3pp
d30 vs d1 probability

Drivers (SHAP top 3 from the 7-day model — same features feed all horizons)

day 113.2%
2026-09-16
  • critical_incident_7d+0.130
  • month-0.097
  • ooni_anomaly_7d-0.090
day 717.1%
2026-09-22
  • critical_incident_7d+0.130
  • month-0.097
  • ooni_anomaly_7d-0.090
day 144.7%
2026-09-29
  • critical_incident_7d+0.130
  • month-0.097
  • ooni_anomaly_7d-0.090
day 3011.9%
2026-10-15
  • critical_incident_7d+0.130
  • month-0.097
  • ooni_anomaly_7d-0.090

Honest caveats

  • Per-horizon XGBoost stacked into a trajectory — not a true joint distribution.
  • Conformal bands are absolute-residual q90 from LOCO predictions on 18 eval countries.
  • Long horizons (h>=14d) have wider effective uncertainty than the conformal q90 suggests; the same lag features feed all horizons.
  • Bands floor at [0,1]; probabilities below 0.05 are at the model's noise floor.
  • SHAP attribution above is from the 7-day point model. The trajectory uses 30 separate per-horizon XGBoost models that share the same feature set; we have not yet computed per-horizon SHAP.

Sample horizons

DayDateP(shutdown)90% interval
d12026-09-1613.2%[0%, 57%]
d72026-09-2217.1%[0%, 82%]
d142026-09-294.7%[0%, 73%]
d302026-10-1511.9%[0%, 90%]
Spotlight countries
About this view

30-day day-by-day shutdown-risk trajectory. Each point is a per-horizon XGBoost prediction (h=1..30) with a 90% conformal band derived from absolute LOCO residuals. Peak risk highlighted.