The narrow finding
In a 2017 PLOS ONE paper, Kokkinakis and colleagues found that League of Legends rank correlated with performance on the WASI-II Matrix Reasoning subtest in a small laboratory sample of experienced players.
That is a real result. It is also exploratory, correlational, and far from an individual IQ test.
Study 1: controlled measures, small sample
The laboratory study recruited 56 experienced League players around several UK universities:
- 51 participants were male;
- the mean age was 20.5;
- every participant had played more than 100 ranked and unranked matches;
- fluid intelligence was represented by the WASI-II Matrix Reasoning subtest;
- participants also completed three working-memory tasks and a theory-of-mind task;
- expertise was represented by League rank.
The reported Spearman correlation between rank and matrix reasoning was about 0.44. A partial correlation controlling for age remained significant.
A correlation of this size means the variables shared some variation. It does not mean rank and intelligence are the same construct, or that rank accurately predicts a score for one person.
Study 2: a different kind of evidence
The paper also compared age and performance patterns in large datasets from League, Dota 2, Destiny, and Battlefield 3. The authors argued that the age profile of MOBA performance resembled the known age profile of raw fluid intelligence more closely than the first-person shooter data did.
Large observational datasets improve precision but introduce other uncertainty. Age can be self-reported inaccurately, player populations are selected, game systems change, and rank reflects the rules of a particular competitive ladder.
The authors explicitly described the 56-person laboratory work as exploratory and the large datasets as noisier.
Rank contains much more than cognition
League rank depends on winning over time. Winning depends on:
- practice and learning history;
- champion and role familiarity;
- mechanical execution;
- teamwork and communication;
- motivation and time available;
- patch knowledge and adaptation;
- connection, hardware, and match conditions;
- the behaviour of teammates and opponents.
Fluid reasoning may contribute without dominating. The 2017 paper acknowledges practice, dedication, learning, social factors, and possible third variables.
A related 2018 PLOS ONE study of Dota 2 expertise found that time on task, grit, and age predicted performance in its model, while fluid intelligence did not add explanatory value in the tested subsample. That does not cancel the League result. It shows why one exploratory correlation should not become a universal rule.
Read the 2017 League paper and the 2018 time-on-task paper together.
Why this is not a Saiki validation study
The researchers administered a standardized matrix-reasoning task under laboratory conditions. Saiki does not administer that task. Its Strategic Play panel reports economy and tempo from recent matches and explicitly states that it is not an IQ score.
The paper did not test:
- Saiki;
- HEXACO self-report scores;
- champion or role variety as personality measures;
- vision, assists, deaths, healing, or surrender rate as trait indicators;
- an algorithm for predicting individual intelligence from Riot data.
Calling this paper the scientific basis of a future Saiki cognitive score would move ahead of the evidence. A new product feature would need its own development sample, held-out validation, reliability analysis, error bounds, fairness testing, and external review.
The useful takeaway
League is a complex task environment, so researchers can reasonably ask how expertise relates to cognitive measures. The 2017 study offers an interesting first result: rank and matrix reasoning were moderately correlated in 56 experienced players.
The right next sentence is “replicate and validate.” It is not “your rank reveals your IQ.”