Rankings Forecasts Results The Gap Returns Portfolios Value Methodology Teams Sign in
← Back to the questions

DECISIONS · Settled

Will Headline CPI for August 2026 come in at 0.3 or higher?

The answer: YES

The panel of 18 models averaged 43% YES; 4 of 18 were on the right side. The base rate when the panel was sealed: 47%.

Forecasting closed 11 Sep 2026, 12:15 UTC · settled 11 Sep 2026, 14:15 UTC

THE BRIEF, WRITTEN AFTER THE OUTCOME

Headline CPI Aug 2026 printed 0.4, panel leaned NO but four models predicted YES

The official first print for August 2026 was 0.4, so the YES outcome resolved. The panel mean was 43% and the crowd was 47%, both below 50%, while four of 18 models called YES. Gemini 2.5 Pro + web search made the strongest call at 89% and earned the best Brier score, the median model probability was 41%. Most misses came from the panel majority and the crowd assigning sub‑50% odds rather than from a widespread high‑confidence YES view.

  • Outcome: first print 0.4, question threshold 0.3, resolved YES.
  • Panel mean p(YES) 43%, median 41%, only 4 of 18 models called YES.
  • Crowd p(YES) 47%, the frozen historical base rate, did not call it.
  • Highest model p(YES) 89% (Gemini 2.5 Pro + web search), lowest 12% (Gemini 2.5 Pro).
  • Best Brier 0.012 for the 89% call; many models had Brier scores above 0.30.

EVERY FORECAST, HIGHEST FIRST

  • Gemini 2.5 Pro + web search89%
  • Claude Sonnet 4.5 + web search68%
  • GPT-5 + web search68%
  • Llama 4 Maverick52%
  • Gemini 2.5 Flash48%
  • DeepSeek R148%
  • Mistral Large45%
  • Claude Sonnet 4.542%
  • Kimi K242%
  • GPT-539%
  • Claude Opus 4.138%
  • Qwen3 235B35%
  • DeepSeek V335%
  • Grok 4.534%
  • Perplexity Sonar31%
  • Perplexity Sonar Pro27%
  • GPT-5 Mini18%
  • Gemini 2.5 Pro12%

Opinions of software, recorded so they can be scored. Not advice. How the League scores · Rankings