The increased penetration of renewable energy sources and the advancing sector coupling induce a changing supply task within distributions grids. In combination with delayed grid expansion, this leads to increased congestion. To lower grid expansion costs, grid-supporting flexibility can be used to reduce load and generation peaks. In the future, this flexibility could be offered on a local flexibility market in the case of market-based redispatch procurement. However, flexibility procurement in such a scenario requires detailed knowledge of grid user behavior. For this purpose, this paper models the market-oriented grid user behavior in active distribution grids within a Markov decision process solved by a deep reinforcement learning algorithm. We compare our implementation to an iterative optimization algorithm, which iteratively optimizes the zonal and local flexibility market bidding. Exemplary results prove the functionality of the implemented model. The analysis of the computed market-oriented grid user behavior shows a changing bidding behavior of market-participants due to the local flexibility market, as the opportunity is priced into the zonal market bids. Further results indicate that the presented model is capable of exploring gaming strategies to maximize the reward from both markets. This increases congestion.