Methods and data limitations

1. What enters the sample

The main analysis uses retrieved Ranked Solo / Duo matches (queue 420). The initial request is for at most 20 matches. Games shorter than five minutes and games flagged as early surrenders are excluded, so the usable sample may be smaller. Missing required data can prevent an analysis update. Additional match history, the initial analysis sample and season ranked win/loss totals are separate aggregates.

Match-V5 supplies match records, League-V4 supplies rank at observation time, and Account-V1 identifies accounts. Cached responses may be reused; opening a page does not always cause a fresh Riot data request.

2. Aggregate calculations

KDA is not the mean of individual match KDAs. DPM (champion damage per minute), GPM, CS/min and Vision/min are arithmetic means of the individual per-minute rates. For 100 CS in 20 minutes and 280 CS in 40 minutes, mean CS/min is (5+7)/2 = 6.0, not 380/60 = 6.33.

Damage share and gold share are percentages of the player's own team totals. KP is also a percentage on a 0–100 scale. Resource-conversion index E is mean damage share / max(1, mean gold share) × 100. It does not directly measure value such as peel or decision-making.

3. CORE SCORE formulas

C(x) clamps a value to 30–98 and rounds to the nearest integer. Although shown on a 100-point scale, these formulas do not span the full 0–100 range. The coefficients are internal design choices, not a statistically validated model of player skill.

CW = control wards placed per game; WK = wards killed per game; D = deaths per game; TD = time dead in seconds per game. KP is kill participation: use Riot's killParticipation as a percentage, or (kills + assists) / max(1, team total kills) × 100 if unavailable, then average across matches. CORE SCORE is the rounded, equally weighted mean of available scores in these four domains. The auxiliary lane and objective domains are not included.

If the four domains are 70, 80, 60 and 90, CORE SCORE is 75. Role-specific references shown beside some individual metrics are fixed values; these domain formulas are not role-adjusted. They are not standardized against all players of the same rank.

4. Missing data and interpretation

• Optional challenge or opponent data may be unavailable. Related auxiliary metrics can be undetermined. Missing is not the same as zero.
• Lane gold and CS differences compare end-of-game totals. They are not values at 15 minutes and do not establish the cause of a win or loss.
• Data-quality values describe sample size and field coverage, not the probability that an interpretation is correct.
• LP values are observations. Across ordinary tier or division changes, a simple LP difference is withheld. The plotting coordinate is not cumulative earned LP. Even a one-match interval can contain adjustments such as dodges.