Detecting Deceptive Behavior in Online Chess: A Machine Learning Approach

Yanyu Chen; Yijian Che; Md Maruf Hasan; Affan Yasin

doi:10.4018/979-8-3373-5062-2

Detecting Deceptive Behavior in Online Chess: A Machine Learning Approach

Yanyu Chen
, Yijian Che
, Md Maruf Hasan
, Affan Yasin

School of AI and Advanced Computing

Xian Jiaotong-Liverpool University

Research output: Chapter in Book or Report/Conference proceeding › Chapter › peer-review

Abstract

Online chess tournaments face persistent threats from engine-assisted cheating, undermining fair competition. This study evaluates six machine learning models - Support Vector Machine (SVM), Random Forest, Decision Tree, Logistic Regression, Naive Bayes, and Long Short-Term Memory (LSTM) - to detect suspicious behaviors such as abnormal move timing, high accuracy, and sudden rating changes. Using the Chess Cheating Dataset under simulated tournament conditions, the Random Forest model achieved the highest accuracy (76.15%), followed by SVM (74.90%) and LSTM (75.78%). While ensemble methods provided strong overall performance, interpretable models like Decision Trees and temporal models like LSTM offered distinct advantages. The findings suggest that combining multiple approaches may
enhance detection reliability in online chess.

Original language	English
Title of host publication	Computational and Deep Learning Models for Advanced Behavioral Analysis
Publisher	IGI Global
Chapter	8
DOIs	https://doi.org/10.4018/979-8-3373-5062-2
Publication status	Published - Apr 2026

Keywords

Machine learning (ML)
Cheating Detection
Behavioral Analytics
Online Gaming Security
Fair Play Mechanisms

Access to Document

10.4018/979-8-3373-5062-2

Cite this

@inbook{fbcbe170fbf749ae80c7cd28ec0e1c56,

title = "Detecting Deceptive Behavior in Online Chess: A Machine Learning Approach",

abstract = "Online chess tournaments face persistent threats from engine-assisted cheating, undermining fair competition. This study evaluates six machine learning models - Support Vector Machine (SVM), Random Forest, Decision Tree, Logistic Regression, Naive Bayes, and Long Short-Term Memory (LSTM) - to detect suspicious behaviors such as abnormal move timing, high accuracy, and sudden rating changes. Using the Chess Cheating Dataset under simulated tournament conditions, the Random Forest model achieved the highest accuracy (76.15\%), followed by SVM (74.90\%) and LSTM (75.78\%). While ensemble methods provided strong overall performance, interpretable models like Decision Trees and temporal models like LSTM offered distinct advantages. The findings suggest that combining multiple approaches may enhance detection reliability in online chess.",

keywords = "Machine learning (ML), Cheating Detection, Behavioral Analytics, Online Gaming Security, Fair Play Mechanisms",

author = "Yanyu Chen and Yijian Che and Hasan, \{Md Maruf\} and Affan Yasin",

year = "2026",

month = apr,

doi = "10.4018/979-8-3373-5062-2",

language = "English",

booktitle = "Computational and Deep Learning Models for Advanced Behavioral Analysis",

publisher = "IGI Global",

}

TY - CHAP

T1 - Detecting Deceptive Behavior in Online Chess: A Machine Learning Approach

AU - Chen, Yanyu

AU - Che, Yijian

AU - Hasan, Md Maruf

AU - Yasin, Affan

PY - 2026/4

Y1 - 2026/4

N2 - Online chess tournaments face persistent threats from engine-assisted cheating, undermining fair competition. This study evaluates six machine learning models - Support Vector Machine (SVM), Random Forest, Decision Tree, Logistic Regression, Naive Bayes, and Long Short-Term Memory (LSTM) - to detect suspicious behaviors such as abnormal move timing, high accuracy, and sudden rating changes. Using the Chess Cheating Dataset under simulated tournament conditions, the Random Forest model achieved the highest accuracy (76.15%), followed by SVM (74.90%) and LSTM (75.78%). While ensemble methods provided strong overall performance, interpretable models like Decision Trees and temporal models like LSTM offered distinct advantages. The findings suggest that combining multiple approaches may enhance detection reliability in online chess.

AB - Online chess tournaments face persistent threats from engine-assisted cheating, undermining fair competition. This study evaluates six machine learning models - Support Vector Machine (SVM), Random Forest, Decision Tree, Logistic Regression, Naive Bayes, and Long Short-Term Memory (LSTM) - to detect suspicious behaviors such as abnormal move timing, high accuracy, and sudden rating changes. Using the Chess Cheating Dataset under simulated tournament conditions, the Random Forest model achieved the highest accuracy (76.15%), followed by SVM (74.90%) and LSTM (75.78%). While ensemble methods provided strong overall performance, interpretable models like Decision Trees and temporal models like LSTM offered distinct advantages. The findings suggest that combining multiple approaches may enhance detection reliability in online chess.

KW - Machine learning (ML)

KW - Cheating Detection

KW - Behavioral Analytics

KW - Online Gaming Security

KW - Fair Play Mechanisms

U2 - 10.4018/979-8-3373-5062-2

DO - 10.4018/979-8-3373-5062-2

M3 - Chapter

BT - Computational and Deep Learning Models for Advanced Behavioral Analysis

PB - IGI Global

ER -