MARKETS · Settled
No change flips Hike 25 bps for October Fed by October 15?
The answer: YES
The panel of 20 models averaged 42% YES; 3 of 20 were on the right side. The crowd price when the panel was sealed: 53%.
THE BRIEF, WRITTEN AFTER THE OUTCOME
Outcome YES; crowd 53% vs panel mean 42%, three models foresaw the YES
The question resolved YES. The crowd, the Polymarket price frozen when asked, sat at 53%. The panel mean was 0.42 and the median was 0.41, with three models calling YES: Gemini 2.5 Pro at 98%, Gemini 2.5 Pro + web search at 68%, and DeepSeek R1 at 60%. The best call was Gemini 2.5 Pro, which placed almost certain probability but recorded no thesis or drivers. Many models leaned NO, the lowest being Grok 4.5 at 22%, also with no recorded supporting notes.
- Outcome: YES.
- Crowd (Polymarket frozen price) 53% YES, called it.
- Panel mean 42% YES, median 41% YES; 3 of 20 models called YES.
- Highest model: Gemini 2.5 Pro 98% YES, lowest: Grok 4.5 22% YES.
- Panel forecasts contain no recorded theses or drivers for the top and bottom calls.
EVERY FORECAST, HIGHEST FIRST
- Gemini 2.5 Pro98%
- Gemini 2.5 Pro + web search68%
- DeepSeek R160%
- GPT-547%
- DeepSeek V345%
- Mistral Large45%
- Claude Sonnet 4.542%
- Qwen3 235B42%
- Llama 4 Maverick42%
- Kimi K242%
- Llama 4 Maverick + web search40%
- Claude Sonnet 4.5 + web search38%
- Perplexity Sonar Pro35%
- DeepSeek V3 + web search35%
- Perplexity Sonar31%
- GPT-5 + web search29%
- Claude Opus 4.128%
- GPT-5 Mini27%
- Gemini 2.5 Flash25%
- Grok 4.522%