The Risk Takers Podcast How to Make a Living Betting on What Someone is Going to Say w/ Foster| Ep 147
18 snips
Mar 18, 2026 A deep dive into mention markets and live trading tactics. Stories about building announcer databases and pricing speeches. How coding, bots, and AI speed cross-exchange arbitrage. Practical talk on live mispricing, RFQs, and exploiting correlated award and political markets.
AI Snips
Chapters
Transcript
Episode notes
Discord Play Almost Wiped Group Over Contract Mismatch
- Foster's group nearly lost big after misreading a Kalshi contract: title/summary suggested one outcome but the actual contract covered a different time window.
- They accumulated large positions, got adverse action, then Kalshi sided with title/summary and refunded losses.
Broadcast Context Drives Mention Frequencies
- Broadcasters and production differences create predictable mention patterns; e.g., primetime NFL broadcasts and scorebugs influence whether announcers say 'inches'.
- Segmenting by broadcast crew, night, and team's style revealed strong non-random strike frequencies.
Feature Engineer Mentions Beyond Raw Transcripts
- Don't rely on raw historical transcripts alone; feature-engineer predictors (team, matchup, broadcast crew) to beat markets once transcripts become common.
- Adjust baseline frequencies for context like run/pass balance and primetime crew tendencies.