Leaderboard
Top forecasters
Lower Brier score = better. We rank by accuracy, not by volume. Minimum 25 resolved markets to qualify for ranking.
All timeThis yearThis monthPoliticsFiscalPolicy
| Rank | Forecaster | Brier | Resolved | Streak | Region |
|---|---|---|---|---|---|
| 01 | parliamentarian | 0.137 | 247 | 12 🔥 | London |
| 02 | whitehall_old | 0.142 | 189 | 8 | London |
| 03 | budget_watcher | 0.151 | 304 | 5 | Edinburgh |
| 04 | manchester_civic | 0.156 | 156 | 11 | Manchester |
| 05 | backbench_bench | 0.158 | 92 | 3 | Cardiff |
| 06 | historical_dynamics | 0.162 | 211 | 7 | Oxford |
| 07 | city_economist | 0.167 | 78 | 4 | London |
| 08 | policy_wonk | 0.171 | 134 | 2 | Bristol |
| 09 | edinburgh_reader | 0.174 | 67 | 6 | Edinburgh |
| 10 | leeds_civic | 0.178 | 89 | 1 | Leeds |
What is Brier score?
Brier score measures the accuracy of probabilistic predictions. A perfect forecaster scores 0; a forecaster who always guesses 50/50 on binary questions scores 0.25; a forecaster who is reliably wrong scores 1. Lower is better. We use Brier rather than profit because it rewards calibration over luck — being right with appropriate confidence, repeatedly. See full methodology.
How to climb the leaderboard
- Predict consistently — only resolved markets count, and you need 25 to qualify for ranking
- Calibration matters more than direction — say 70% if you mean 70%, not 95%
- Specialise — separate leaderboards for Politics, Fiscal, Policy reward depth
- Streaks are bonus signal, not scoring — 5+ consecutive correct adds visibility