What's the Best League of Legends Draft AI?
We tested Winrate, iTero, and LoLDraftAI on 20,000 real ranked drafts. Winrate had the highest realized win rate in every Top-K bucket.
2026年4月28日
文章
關於方法論、產品決策和底層實現的長文內容。
How the Draft Model Works
How Winrate recommends champion picks during draft, and why incomplete information is the hard part.
2026年3月11日
Building an Item Recommender When Winning Explains Almost Nothing
Why predicting wins from item purchases doesn't work, what we tried, and how pairwise ranking solved it.
2026年2月27日
Stop Interpreting Predictive Models as Causal: How We Built Player Insights
Why SHAP values and feature importance can mislead when you want to decide what to change. How we use Double ML to separate correlation from causation in player insights.
2026年2月18日
What 'Similar Champions' Means in Winrate
A plain-language explanation of what "similar" means, how it's measured, and how we test if it works.
2026年2月16日
Measuring What Matters: From Stats to Win Probability
How I rebuilt my performance scoring system after discovering it was rewarding losers and punishing winners.
2026年2月4日