Volume 22, Issue 2 (9-2025)
JSDP 2025, 22(2): 31-42 |
Back to browse issues page
Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
Shaeeri Z, Kazemitabar J, Haghverdi S. Stock Market Anomaly Detection Using Behavioral Analysis. JSDP 2025; 22 (2) : 2
URL: http://jsdp.rcisp.ac.ir/article-1-1203-en.html
URL: http://jsdp.rcisp.ac.ir/article-1-1203-en.html
Babol Noshirvani University of Technology
Abstract: (38 Views)
Risk management is of crucial significance in financial institutes. Lack of labeled data and the unbalanced nature of samples is one of the challenges in risk management and fraud detection. Generally, when labels are not available, fraud detection is ranked based on the degree of anomaly. In this article, a live and unsupervised fraud detection based on behavior analysis is proposed for anomaly detection in stock market. We borrow the notion of log-likelihood ratio from detection theory to build the risk function. We find the conditional probability of a transaction with either fraudulent and non-fraudulent intents. The ratio of the two -in log scale- determines the transaction risk. In measuring the risk, we use Bayesian concepts of probability theory to ease the calculation. Specifically, we use the Bayes rule to find the conditional probability and assume Naive Bayese implying different risk components are independent of each other. We then engage characteristic traits such as time, date, and amount of transaction to build a personal profile. We use binning methods to segment the amount of transaction. Using data in a specific period of time, we build a profile for each user including his behavioral traits in stock trading. Each profile, given enough data, has the capability to explain fluctuations in the user’s behavior. In the test phase, we propose a statistical approach where for each new sample, risk is calculated in which is based on the amount of deviation from expected behavior recorded in his profile. Users with significant behavior change will get further analyzed, while normal users will have their profile updated based on new transactions. Simulations show that the proposed technique is successful in detecting front-running.
Article number: 2
Type of Study: Research |
Subject:
Paper
Received: 2021/01/11 | Accepted: 2025/03/8 | Published: 2025/09/13 | ePublished: 2025/09/13
Received: 2021/01/11 | Accepted: 2025/03/8 | Published: 2025/09/13 | ePublished: 2025/09/13
Send email to the article author
| Rights and permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |
