Rule Forecast Beta

When will it actually become final?

We track every proposed benefits rule and forecast whether, and roughly when, it will be finalized, so you can get ahead of what's coming. Each forecast is a window, not a date: it widens the further out we look, because that's honest.

Backtested & calibrated

Scored against 971 past forecasts drawn from public Federal Register history. It beats the naive base-rate guess (Brier 0.244 vs 0.249) and is well-calibrated (within 7 points).

Intelligence, not legal advice: a planning signal, not a deadline.

The regulatory runway

Everything that's coming, on one timeline

Every tracked proposal, sorted by when it's likely to land. Scan it in five seconds — then click any rule to see exactly where it stands.

Sort
Jul 26Oct 26Jan 27Apr 27Jul 27Oct 27Jan 28Today

Each row: ▇ likely-final window (middle 50%) · ○ most-likely date · gold tick = comment deadline · dashed line = today · color = 12-month odds.

The full board

Every proposed rule we're tracking

40 active proposals across the benefits agencies, each with its 12-month finalization odds and predicted window.

Forecasts are generated from public Federal Register history and agency base rates, updated as new data lands. They are editorial intelligence and curation to help you plan — not legal advice, and not a guarantee of any agency's timing. Always confirm against the official record before relying on a date. Last updated Jun 16, 2026.

The track record

We grade our own homework

A forecast is only useful if it's honest about how often it's right. We replayed the model across the public Federal Register record and plotted predicted probability against what actually happened.

perfectly calibrated0%25%25%50%50%75%75%100%100%Forecast probabilityActual finalization rate
Points on the dashed line mean a forecast of X% came true X% of the time. Marker size reflects how many forecasts landed in each band. Calibration error (ECE) 7%, across 971 past forecasts.
0
Forecasts scored
Backtested on FR history
+0%
Skill over base rate
Brier vs naive guess
0 pts
Calibration error
Lower is better (ECE)
0.244
Brier score
Baseline 0.249
0%
Ever finalized
Of proposals, eventually
±0d
Timing error
Median date, typical miss