The stock market is a central aspect of modern economies, and the ability to detect price manipulation in the market is crucial for maintaining fair and efficient trading. The purpose of this paper is to provide an effective method for detecting price manipulation in the stock market. This research includes some insightful results in which the developed model is able estimate the probabilities of execution for the different orders and provides a probability trend which suits the behavior of each agent who has placed the order. It may seem a bit odd but the true contribution of the research is not the accuracy achieved in predicting if an order will get executed or cancelled. The most significant contribution is the estimated probability of execution plotted against the number of market updates since the order entered the market which act as the proxy for the intention of the client placing the order. To tackle the issue of limited data availability, the thesis employs synthetic data generation and creates code-based market participants (Agents) that can simulate market conditions. By generating synthetic data, the classifier can be trained on a much larger dataset, which increases its accuracy and makes it more robust. Moreover, the use of code-based agents allows for a more comprehensive solution as it provides a realistic simulation of the stock market, which can be used to test the classifier. The novel approach of this thesis lies in predicting the probability of execution of an order and using it as a proxy to understand the intention of the client who placed the order. By predicting the likelihood of execution, the classifier can identify instances where the client’s intention is to manipulate the stock price.