The U.S. Securities and Exchange Commission prosecutes about 50 cases of insider trading on average each year. According to a study from the University of Technology Sydney, the true number of incidents could be at least four times higher. Illegal insider trading involves buying or selling financial securities using information that is not available to the public, and although insiders are required to report their transactions to the SEC, illegal trading remains difficult to detect and even harder to prove.
Solmaz Batebi, an assistant teaching professor in the University of Washington’s School of Business, is exploring Artificial Intelligence as a tool to improve insider trading predictions. Her interest grew from early exposure to financial markets in Iran and later work in market surveillance and regulatory development at the Tehran Stock Exchange, where she helped build data-driven compliance tools and update regulations related to market manipulation and high-frequency trading. That experience pushed her to bridge finance and computer science, combining business knowledge with coding and data science skills.
Batebi later earned a doctorate in Business Administration with a concentration in Finance from the University of Texas Rio Grande Valley and focused her dissertation on whether machines can better predict insider trading. Her research found that machine learning methods substantially outperformed traditional methods in predicting both the likelihood and the magnitude of insider sales. She concluded that human decision making is complex enough that linear models often miss important variables, while Artificial Intelligence is better suited to synthesizing information and identifying behavioral patterns.
The largest benefit was speed. Before Artificial Intelligence, investigations on a single trade could take six months to a year to understand the intention. With Artificial Intelligence, much of that same workload can happen in a couple minutes. Her analysis also found differences in trading behavior across genders: male traders appeared to act more on incentive, while female traders relied more heavily on information. As a result, predictive gains were more pronounced among female insiders.
In fall 2025, Batebi joined UW Bothell, where she teaches finance and brings computer science expertise into the classroom. She encourages students to build technical and analytical skills that can distinguish them in the job market and supports student participation in research. Her next paper will examine how Artificial Intelligence can be applied in investment banking and wealth management.
