How Could They Win? An Exploration of Win Condition for Esports Narratives in Dota 2
summary
Proposes a win condition model for Dota 2 that reverses standard win prediction: instead of mapping game state to winner, it maps a chosen winner to the game state that would let that team win. Uses a two-stage ML pipeline—neural network 1 predicts final duration from lineup and winner, neural network 2 predicts total kills—and runs separately per team (example output: Team 1 wins at ~20 minutes with N kills). Prototype exists; planned user tests with commentators and audiences remain to evaluate usefulness.
AI-written summary · about
related talks
09mWait, But Why?: Assessing Behavior Explanation Strategies for Real-Time Strategy Games
35mDecoding bias and narrative in competitive video games
32mBenefits and use cases of emerging tech for sports and e-sports
40mFrom AI Underdog to Most Valuable Player: Lead with a Champion's Mindset
08mVisualizing the Spatio-Temporal Evolution of Gameplay using Storyline Visualization: A Study with...
14mModeling Individual and Team Behavior through Spatio-temporal Analysis
42mHow Deep Learning is Revolutionizing Sports Analytics
03mThe Ludic Potentials of Player Data