🏈 I ran a social experiment: How does social data compare to Polymarket's top sports traders?
I knew social data performed well but the results still surprised me.
269 NFL games. I tracked social mentions, engagements, and total creators before each kickoff.
Result: 63.6% prediction accuracy.
For context, @Polymarket's top traders hit 59.4-66% win rates. Real money on the line. Major profits.
Social metrics landed within 2.4% of the best performer. Only outperformed by one trader.
Social data isn't noise. When aggregated properly, crowd intelligence produces prediction market-grade signal.
Not a replacement for analysis. But it deserves a seat at the table as alternative data to verify confidence and spot opportunities.
How I did this: @LunarCrush MCP connector in Claude from @AnthropicAI.
