Unexpected Value - Poker · Daily Fantasy · Sports Betting #45: The Black Box
6 snips
Aug 17, 2021 In this discussion, Daniel Hutchings, a software engineer and creator of Pro Poker Tools, shares insights from his journey in transitioning from poker software to developing winning Daily Fantasy Sports (DFS) strategies. He explores the nuances between balanced and exploitative strategies, and how automated simulations revolutionize lineup creation. Daniel also touches on the impact of ownership in different sports, the challenges of late swaps, and the future of DFS in a world of solvers and automation, all while balancing his diverse career in tech and music.
AI Snips
Chapters
Books
Transcript
Episode notes
Iterate By Averaging Thousands Of Candidates
- Hutchings iterated by generating thousands of candidate lineups and analyzing their average EV post-contest to refine models.
- He used stakeholder backing to scale once simulations consistently showed positive average EV.
Showdown Rewards Uniqueness Over Median EV
- Large variance in showdown-style events means a few unique lineups can take massive prizes, so uniqueness matters more than median EV.
- Mass-enter strategies can lose to a single contrarian solo lineup when chalk ties explode the prize split.
Collect Real CSVs To Validate Models
- When building automation, keep an information-first approach: enter minimal real tickets to collect outcome CSVs for feedback.
- Use those real-results CSVs to validate and refine simulation assumptions before committing volume.
