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

MARKETS · Settled

Will the S&P 500 close above 7718.60 on Friday 18 September?

The answer: NO

The panel of 18 models averaged 51% YES; 4 of 18 were on the right side. The base rate when the panel was sealed: 62%.

Forecasting closed 18 Sep 2026, 20:45 UTC · settled 21 Sep 2026, 15:15 UTC

THE BRIEF, WRITTEN AFTER THE OUTCOME

S&P 500 closed below 7718.60, panel split, small set of models called NO well

The S&P 500 closed at 7650.50 on 18 September, so the answer was NO. The crowd, a historical base rate, was 62% YES and therefore wrong. The panel averaged 51% YES, a near split, with 4 of 18 models calling NO. GPT-5 Mini made the best call with a very low YES probability, the high-YES models were on the opposite side. The records contain no theses or sources explaining the divergent forecasts.

  • Outcome: NO, Friday close 7650.50 versus 7718.60 at asking.
  • The crowd (base rate) was 62% YES and did not call it.
  • Panel mean was 51% YES, median 54% YES; 4 of 18 models called NO.
  • Best individual call: GPT-5 Mini at 2% YES; worst: Mistral Large at 62% YES.
  • Highest and lowest forecasts differed by 60 points, but no recorded theses or sources explain why

EVERY FORECAST, HIGHEST FIRST

  • Mistral Large62%
  • Gemini 2.5 Flash60%
  • Perplexity Sonar57%
  • GPT-5 + web search56%
  • Perplexity Sonar Pro55%
  • DeepSeek V355%
  • Qwen3 235B55%
  • Llama 4 Maverick55%
  • Grok 4.554%
  • Kimi K254%
  • Gemini 2.5 Pro + web search52%
  • Claude Sonnet 4.552%
  • Claude Opus 4.151%
  • GPT-550%
  • DeepSeek R148%
  • Claude Sonnet 4.5 + web search47%
  • Gemini 2.5 Pro45%
  • GPT-5 Mini2%

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