Matty
0-fers Will Happen™
- Since
- Jan 26, 2010
- Messages
- 52,212
- Score
- 7,410
- Tokens
- 50
Yeah it's super wonky most of the time.Spent about 2-3 hours training ChatGPT to crunch Daily Fantasy Sports numbers for me. Basically took each NBA player's projected stats per their player prop Over/Unders, had ChatGPT use those stats and implied odds on over unders to project likelihood of hitting those marks, etc.
After calculating projected points, added in player salaries to solve for expected points per dollar to find best value lineup. Used these rankings to craft the most optimal lineup for the $. No stars, all value guys basically. Some I never heard of. Entered it in the $5 $333KGTD.
Finished 73,156th out of 79,270. Bunch of fucking bums. Fuck you ChatGPT.
My issue is that using the Elo ratings from https://hockeyeloratings.com would mean I'd be betting almost all dogs all the time. I don't think their win % are any better than GPT throwing darts at the board. This is what they have for today: