Voidly
Multi-horizon shutdown forecast

AO: available horizons.

Current-regime risk at three horizons, each with its own uncertainty. These are not shutdown-onset predictions.

API calculation time: 2026-10-12 06:59:26.319594 (timezone not provided) · model multi-horizon-h-v1 · raw JSON

Calculation time is not the date of the underlying evidence.

Probabilities

1 day
1d
36.4%
Interval unvalidated; no band shown.
7 days
7d
58.8%
Interval unvalidated; no band shown.
30 days
30d
70.0%
Interval unvalidated; no band shown.
Cite this forecast

Consistency check

OK — monotonicP(1d) ≤ P(7d) ≤ P(30d)

The three horizons are internally consistent: longer windows capture at least the risk of shorter windows. This is the expected behavior — a 30-day window must contain any 7-day or 1-day shutdown.

36.4% ≤ 58.8% ≤ 70.0%

Top-5 SHAP features per horizon

HorizonRankFeatureSHAP
1-day#1week_of_year-0.927
#2day_of_week-0.342
#3block_rate_roll7_std-0.131
#4block_rate_roll7_mean-0.089
#5block_rate_roll30_mean-0.087
7-day#1week_of_year-0.651
#2block_rate_roll7_std-0.156
#3recent_shutdown+0.116
#4month-0.086
#5gdelt_unrest_30d-0.070
30-day#1week_of_year-0.775
#2risk_tier-0.153
#3block_rate_roll7_std-0.150
#4recent_shutdown-0.087
#5gdelt_unrest_30d-0.065

Each horizon has its own learned drivers. Positive SHAP values push the prediction up; negative values push it down. Features differ across horizons because near-term risk responds to acute signals, while month-out risk responds to slower structural factors.

Honest caveats

  • Each horizon is a separate XGBoost model with its own features and its own honest evaluation. LOCO AUC across spotlight countries: 1d 0.91 · 7d 0.88 · 30d 0.84.
  • The previous LOCO residual bands have no validated 90% coverage for these probability scores. We withhold them until an eligible evaluation can measure coverage.
  • These numbers are not guarantees. A 70% probability does not mean a shutdown is certain; a 20% probability does not mean you're safe.
  • See /atlas/multi-horizon for the cross-country index, or /atlas/models for the full model registry.
About this view

How the current-regime shutdown-risk level for AO reads over 1 day, 7 days, and 30 days. Each horizon uses its own XGBoost model with per-horizon feature attribution. Intervals are withheld because coverage is unvalidated. These are current-regime risk signals, not onset predictions — see the scope note below.

Forecast scope & limitations