How the analyzer works

1. Every game is split into three phases

Riot's match timeline records every player's gold, CS, experience and position once a minute, plus every kill, ward, objective and tower. We split that timeline at 14:00 (when tower plates fall and laning usually ends) and 25:00 (when Baron-centred late game takes over). Only Summoner's Rift games of 15 minutes or longer are analyzed; remakes, ARAM and Arena are skipped.

2. Six skills are measured in each phase

• Farming: CS per minute in that phase (lane minions plus jungle camps).
• Gold lead: Your gold compared with the enemy player in the same role at the end of the phase.
• Fight impact: Kill participation in that phase, combined with kills plus assists per death.
• Survival: Deaths per 10 minutes, with extra weight on isolated deaths (no teammate within about 2,500 units).
• Vision: Wards placed plus enemy wards cleared, per 10 minutes.
• Objectives: Share of your team's dragons, heralds, Barons, grubs and towers in that phase that you took part in.

3. Each measurement becomes a 0–100 score

Every measurement is compared with a target for your role and phase, based on what strong (roughly Diamond and above) players achieve. 50 means on target. Each step of the “spread” above or below the target moves the score 50 points, capped at 0 and 100. So a score of 75 means you're clearly better than the target, and 25 means clearly worse. Your role is the one you played most often in the games being analyzed.

Phase targets used by the analyzer (Mid lane example)

Whole-game targets by role (Champions page)

These targets are deliberately simple reference points rather than exact averages for any one rank. Think of them as “what a strong player in this role usually manages”.

4. Findings come from patterns across many games

The analyzer combines your per-game scores (from the last 20, 50, 100 or all stored games) to produce findings:

• Strengths and weaknesses: your highest and lowest phase-and-skill combinations, such as “Mid game vision” or “Early farming”.
• Trends: skills whose score moved by 10 or more points from your older games to your recent ones, measured with a trend line through every game.
• Caught alone late: when more than a third of your deaths after 25:00 happen with no teammate nearby.
• Lead conversion: how often you win when you're 500+ gold ahead of your lane opponent at 14:00.
• Death spikes: a particular minute of the game where you die far more often than usual.

Opening a finding shows how much it actually affects your results, for example the difference in win rate between games where you hit the target and games where you didn't. It also gives timed, in-game actions and a practice drill. If a weakness barely changes your win rate, the analyzer says so, so you can focus on what matters.

5. Data is fetched once and stored

When you look up a player, Leaguealyzer downloads their recent games from the Riot API in the background and stores them. Pressing Refresh only downloads games played since the last update, and Load 100 older games reaches further back for long-term trends. This keeps pages fast and keeps our use of Riot's API to a minimum.

Limits worth knowing

• Statistics can't see intent: a “bad” death might have been a deliberate trade for Baron.
• Small samples are noisy. Findings get more reliable with 50+ games.
• Lane comparisons need an opponent in the same role, so games where roles were swapped mid-game may skew the gold-lead score.

Questions about the method? See the FAQ.