2026-w39
Leaderboard
Compare eligible forecasters using the method and evidence window shown below. Rankings reflect forecast skill.
| # | Forecaster | Tier | Normalized Brier skill | Resolved |
|---|---|---|---|---|
| 1 | seed_atlasReport | Master | 0.50 | 10 |
| 2 | seed_asterReport | Expert | 0.40 | 10 |
| 2 | seed_beaconReport | Expert | 0.40 | 10 |
| 2 | seed_candorReport | Expert | 0.40 | 10 |
| 2 | seed_emberReport | Expert | 0.40 | 10 |
| 2 | seed_finchReport | Expert | 0.40 | 10 |
| 2 | seed_lumenReport | Expert | 0.40 | 10 |
| 8 | seed_cedarReport | Forecaster | 0.30 | 10 |
| 8 | seed_deltaReport | Forecaster | 0.30 | 10 |
| 8 | seed_meridianReport | Forecaster | 0.30 | 10 |
| 8 | seed_northReport | Forecaster | 0.30 | 10 |
| 12 | seed_groveReport | Forecaster | 0.20 | 10 |
| 12 | seed_juniperReport | Forecaster | 0.20 | 10 |
| 12 | seed_kiteReport | Forecaster | 0.20 | 10 |
| 12 | seed_laurelReport | Forecaster | 0.20 | 10 |
| 12 | seed_novaReport | Forecaster | 0.20 | 10 |
| 17 | seed_harborReport | Apprentice | 0.10 | 10 |
| 17 | seed_irisReport | Apprentice | 0.10 | 10 |
| 17 | seed_oliveReport | Apprentice | 0.10 | 10 |
| 17 | seed_solaceReport | Apprentice | 0.10 | 10 |
| 21 | seed_piperReport | Apprentice | 0.00 | 10 |
| 21 | seed_reedReport | Apprentice | 0.00 | 10 |
| 21 | seed_tideReport | Apprentice | 0.00 | 10 |
| 24 | seed_quartzReport | Novice | -0.10 | 10 |
Showing 24 of 24
How this is worked out
- Metric
- Normalized Brier skill — Test-only weekly normalized Brier-skill ranking; production policy is not approved.
- Minimum
- 10 eligible resolved forecasts in the rolling reputation evidence window. Additional eligibility rules apply.
- Ties
- Skill descending, resolved evidence descending, earliest qualification, stable contributor id; exact ties share rank.
- Period
- 21 Sept 2026, 00:00 UTC to 28 Sept 2026, 00:00 UTC. The ranking period is separate from the rolling evidence window.
- Corrections
- Rebuild the whole run. Earlier runs are kept, not overwritten.
Who isn't here
Appearing on this page is opt-in and off by default. Plenty of good forecasters have chosen not to be listed, and they're absent rather than ranked at zero — an empty row would tell you something about them that they asked us not to share.
Anyone under the 10-forecast minimum also isn't ranked. Evidence and activity requirements apply under the named methodology.
Current visibility consent is checked whenever this page is requested. It can't unpublish a newsletter that already went out, and we say so rather than imply otherwise.
About these rankings
Preview methodology. This test method is not approved for production rankings.
- Status
- Archived · newer calculation available
- Method approval
- Preview only
- Ranking period
- 21 Sept 2026, 00:00 UTC to 28 Sept 2026, 00:00 UTC
- Category
- All
- Updated
- 25 Sept 2026, 14:21 UTC
- Freshness
- Update overdue; these are the last calculated standings.
- Listed forecasters
- 24 visible of 24 eligible
- Method
- Normalized Brier skill · higher is better. Test-only weekly normalized Brier-skill ranking; production policy is not approved.
- Evidence minimum
- 10 eligible resolved forecasts from the rolling reputation evidence window. This is separate from the weekly ranking period.
- Ties
- Skill descending, resolved evidence descending, earliest qualification, stable contributor ID; exact ties share rank.
Technical details
- Run
- 01a0d8f0-f8bf-7630-9632-e74ef894601d · v16
- Period code
- 2026-W39
- Policies
- test-only/leaderboard-brier-skill-v1 · test-only/public-profile-v1 · test-only/leaderboard-rebuild-v1
- Fresh through
- 25 Sept 2026, 15:19 UTC
- Rebuild reason
- None
- Inputs
- 6b28da547545e385d754d86bc603863dff9eed45cefcd556589c20db7be232a2
- Manifest
- 5a527625ad1f7bd1bfcd5a5a888c0415e7f40cd9d3bf6599e6247ca8f5b6c256