Draft simulator
A draft assistant that plays out bans and picks in the official order and, at each turn, suggests what professional play would actually do — with the reasoning behind it. It runs in the interactive explorer; this page explains how it works. Data current as of August 6, 2026.
How suggestions are made
- Popularity in the current window. How often a champion is picked on the recent patches, older ones weighing less.
- Affinity with what is already on the board. Not a raw co-pick count, but how much more often two champions appear together than chance alone would explain. A rare champion that systematically accompanies an existing pick therefore outranks a popular one with no link to the composition.
- Archetypes as a fallback. When a champion lacks history on the window, its Data Dragon roles (tank, mage, marksman…) take over rather than inventing a figure.
What it is worth
Checked by replaying real professional drafts: the pick actually made is first in the suggestions 24% of the time, in the top five 59%, in the top twelve 85%. Clearly better than simply proposing the most played champions, and a long way from an oracle — there is no single right answer to a draft turn.
It also holds outside its training ground: trained without a single LEC game, it still puts 20% of LEC picks first. It is not replaying memorised compositions.
Known limits
What this ranking cannot tell you. Worth reading before quoting it.
- It describes what professional teams do, not what wins. A popular pick is not necessarily a good one.
- A brand-new champion, or one reworked last patch, has no usable history.
- Player comfort picks, scrim preparation and opponent scouting are invisible to it — and they decide a lot of real drafts.
Current meta
Most picked champions across the dataset. Recent patches covered: 26.15, 26.14, 26.13, 26.11.