We analyse the essential features of collusive clique in financial market.
We model collusive trading actions by directed and undirected graphs.
The trader and transaction are transformed and classified into clusters by k-means.
The clique is identified by a dynamic programming based approach.