Notes from the draft data.
Short, honest reads on drafting League of Legends: synergy and matchup edges measured from a large ranked corpus, with the methodology in the open. No tier-list hand-waving.
-
Data · Pro meta
What the world's best prioritized: MSI & EWC 2026 draft trends
Ban, pick, pairing and side-selection trends from 110 international pro games across MSI and the Esports World Cup, on a single patch. Poppy banned in 61% of games and picked in 7%; blue side winning 62%. Small sample, high signal — presence read honestly, and why the pro meta isn't your soloq tier list.
-
Method · Item value
Why win rate lies about items
An item's raw win rate mostly measures the situation it gets bought in, not the item. Defensive items look weak because you buy them behind; snowball items look strong because you buy them ahead. Why we won't rank items by win rate, and the debiased measure, Win Probability Added, we're building instead.
-
Data · Itemization
The builds the ladder has solved, and the ones it hasn't
Some champions have one forced core item bought in 96% of games; others are a coin-flip where the most popular build is still a minority. A map of build consensus across 115,000 ranked games, measured as popularity, with an honest line on why that isn't the same as strength.
-
Data · Synergy & matchups
The strongest champion pairings in the current meta
Which duos actually lift each other, which drag each other down, and the hardest lane counters, measured from a verified-clean ranked corpus with a recency-weighted, shrinkage-corrected estimator. Includes the pairs the meta over-rates, and the bug we fixed to trust our own numbers.