SkyWatcher
Member
I’ve been experimenting with a lightweight Python model to track value in Dota 2 underdog odds—just to see if there’s a consistent edge worth exploiting. Mostly pulling data from Liquipedia + odds APIs and feeding it into a simple logistic regression (for now). Surprisingly, I’ve found a few patterns where certain Tier 2 teams perform better than their odds suggest, especially in BO3 qualifiers.
Last week, I tested it live during some SEA Dota matches and hit 3/4 underdog bets. Nothing life-changing but enough to make me want to refine the model and maybe explore xG-style predictive inputs (like kill diff, avg gold lead, etc).
Curious if anyone here’s tried building algo models for esports betting? Or do you mostly go by feel and matchup analysis? Happy to share code if there’s interest.
Last week, I tested it live during some SEA Dota matches and hit 3/4 underdog bets. Nothing life-changing but enough to make me want to refine the model and maybe explore xG-style predictive inputs (like kill diff, avg gold lead, etc).
Curious if anyone here’s tried building algo models for esports betting? Or do you mostly go by feel and matchup analysis? Happy to share code if there’s interest.