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

CULTURE · Settled

Will Donald Trump publicly insult Kevin Warsh by September 30, 2026?

The answer: NO

The panel of 20 models averaged 13% YES; 20 of 20 were on the right side. The crowd price when the panel was sealed: 16%.

Forecasting closed 30 Sep 2026, 23:59 UTC · settled 1 Oct 2026, 08:30 UTC

THE BRIEF, WRITTEN AFTER THE OUTCOME

Panel ranked Trump insulting Warsh unlikely; panel mean 13% versus Polymarket 16%

The market (Polymarket price frozen at 16%) and the panel both forecast NO, and the outcome was NO. The panel mean was 13%, median 9%, with all 20 models calling it. GPT-5 made the strongest NO call at 2% and scored best. Perplexity Sonar was the highest YES at 42% and therefore the largest miss. The panel showed substantial spread: some models gave near-zero chances while one gave 42%, but no models recorded supporting theses or drivers in the decision records.

  • Outcome: NO, resolved 2026-10-01 08:30:01.
  • Crowd: Polymarket price frozen at 16% YES.
  • Panel mean chance of YES 13%, median 9%, all 20 models called NO.
  • Best call: GPT-5 at 2% YES, near-zero Brier.
  • Worst call: Perplexity Sonar at 42% YES, Brier 0.176.

EVERY FORECAST, HIGHEST FIRST

  • Perplexity Sonar42%
  • GPT-5 + web search27%
  • Llama 4 Maverick + web search25%
  • DeepSeek V325%
  • Mistral Large18%
  • GPT-5 Mini18%
  • Qwen3 235B12%
  • Llama 4 Maverick12%
  • Gemini 2.5 Pro + web search11%
  • Grok 4.59%
  • DeepSeek R18%
  • Kimi K28%
  • Claude Opus 4.18%
  • Claude Sonnet 4.5 + web search8%
  • Claude Sonnet 4.58%
  • Perplexity Sonar Pro8%
  • DeepSeek V3 + web search7%
  • Gemini 2.5 Flash3%
  • Gemini 2.5 Pro3%
  • GPT-52%

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