ProbGap publishes independent probability forecasts and compares them with prediction-market consensus.
Every published forecast is timestamped and preserved. Forecast revisions remain visible rather than replacing earlier estimates, and losing forecasts are never removed.
The purpose of this page is to measure ProbGap’s forecasting performance over time using probabilistic scoring, not simple win/loss counts.
Current snapshot
Resolved forecasts
4
Open forecasts
7
ProbGap Brier Score
0.210
Market Brier Score
0.201
Scores currently reflect two resolved forecasts. They will become more informative as the sample grows.
Open forecasts
Forecast | Market at public. | ProbGap | Current Market | Gap | Latest forecast | Status |
|---|---|---|---|---|---|---|
PG-0002 CLARITY Act becomes law in 2026 | 7% | 5% | 5% | 0 pp | Sep 15, 2026 | Open |
PG-0005 U.S. Q3 GDP growth ≥3.0% | 51% | 64% | 70% | -6 pp | Sep 18, 2026 | Open |
PG-0006 Fed raises rates by 25 bp in October | 68% | 61% | 36% | +25 pp | Sep 25, 2026 | Open |
PG-0007 Canada–EU “associate membership” | 65% | 38% | 41% | −3 pp | Sep 19, 2026 | Open |
PG-0010 | Anthropic IPO by Nov. 30, 2026 | 63% | 52% | 61% | -9 pp | Sep 30, 2026 | Open |
PG-0011 | Bolsonaro wins Brazil presid. | 57% | 68% | 84% | -16 pp | Oct 5, 2026 | Open |
PG-0012 Bank of Canada +25 bps, Oct 28 | 82% | 65% | 82% | -17 pp | Oct 23, 2026 | Open |
Forecasts may be updated when material new information becomes available. Earlier published probabilities remain part of the historical record.
Resolved forecasts
Forecast | Market | ProbGap | Outcome | ProbGap Brier | Market Brier |
|---|---|---|---|---|---|
PG-0004 Fed raises rates by 25 bp at the September FOMC meeting | 84% | 94% | YES | 0.004 | 0.026 |
PG-0008 | U.S.–Iran diplomatic meeting by Sep 30 | 14% | 24% | YES | 0.578 | 0.740 |
PG-0003 Saudi East–West Pipeline by Sep 30 | 18% | 48% | NO | 0.230 | 0.032 |
PG-0009 U.S.–Iran Hormuz Agreement by Sep 30 | 9% | 17% | NO | 0.029 | 0.008 |
The Federal Open Market Committee raised the target federal funds rate by 25 basis points on September 16, 2026, to a range of 3.75–4.00%.
How forecasts are scored
ProbGap uses the Brier Score to evaluate probabilistic forecasts.
The score measures the squared difference between a forecast probability and the eventual outcome.
A perfectly correct forecast receives a score of 0, while larger errors produce higher scores. Lower is better.
ProbGap’s forecasts are scored alongside the prediction-market probability recorded for the same forecast event, allowing the two approaches to be compared on equal terms.
As the number of resolved forecasts increases, this page will also track longer-term calibration and aggregate performance.
Forecast revisions
Forecasting is a process, not a single static number.
When important new information becomes available, ProbGap may update its probability estimate. Updates do not overwrite previous forecasts.
Instead, the original estimate, subsequent revisions and their publication dates remain part of the record.
This makes it possible to distinguish genuine forecasting updates from hindsight.
Resolution policy
Forecasts are resolved according to the predefined resolution criteria of the underlying question or prediction market whenever possible.
Once an outcome becomes known, the corresponding ProbGap probability and prediction-market probability are scored against the final result.
Forecasts remain in the track record regardless of whether they were correct, incorrect or moved against the market.
Why this matters
A forecast can be wrong even when the reasoning was reasonable, and a forecast can be right by chance.
The goal of the ProbGap Track Record is therefore not to count wins.
It is to answer a harder question:
Over many forecasts, does ProbGap assign probabilities more accurately than the market consensus it is trying to challenge?