We developed an algorithm capable of detecting traders exploiting latency between stock exchanges. As a result, it was possible to block practices that were technically legal but distorted the market and harmed regular investors.
Trading thus became safer and fairer, which strengthened trust and attracted new clients.
The client faced a problem in financial markets where millions of dollars were traded within an hour. They were able to detect automated trading bots, but not individuals using custom tools for extremely fast trading. These traders could buy on one exchange and sell on another within a fraction of a second at a better price. They exploited technical latency between exchanges, which was legal but contrary to fair-trade principles and harmful to regular investors.
The goal was to create an algorithm capable of identifying and blocking such traders. We used behavioral analysis and contextualized information from over 50 characteristics of trading accounts. We implemented unsupervised learning algorithms because there was no existing “ground truth” for training classical models. The results were presented through an interactive application that allowed experts to explore various scenarios and validate relevant parameters.
Thanks to our solution, the client can now identify aggressive traders in real time and prevent manipulations that distort the market. Trading has become safer and fairer, increasing investor trust and attracting new clients. The algorithm continues to be used to this day.
