This paper seeks to define a model based on discriminant analysis to categorize a stock as manipulated or non-manipulated based on certain key variables that capture the characteristics of the stock. The model helps identify stocks witnessing activities that are indicative of potential manipulation irrespective of the type of manipulation – action-based, information-based or trade-based. The proposed model helps investigators to arrive at a shortlist of securities that are potentially manipulated and which could be subject to further detailed investigation to detect the type and nature of the manipulation, if any. In a market like India, where there are about 5000 plus securities listed on its major exchanges, it becomes extremely difficult to monitor all securities for potential market abuse. Authors who have earlier used discriminant analysis have used the Linear Classification Function without validating the assumption that governs the model. In this paper we have tested the assumption on data from the Indian capital market and have found that the data does not comply with the assumptions that govern the use of the linear classification function. This resulted in us using the Quadratic Classification Function, which is the appropriate technique for instances where the data does not meet the sated assumptions, to categorise stocks into two categories, namely manipulated and non-manipulated.