A location dependent query (LDQ) result set is valid only in a specific region called the validity region (VR). While limiting the validity of a particular result set to a given area, the VR may also be used in caching implementations to determine if cached results satisfy semantically equivalent queries. Existing LDQ caching schemes rely on the database servers to provide the VR at a cost of high computational overhead. Alternatively, a LDQ proxy cache, which approximates the VR can be employed, freeing the database servers from the high cost of calculating the VR. A LDQ proxy cache architecture is proposed to compute an estimated validity region (EVR) based on the observed querying history at the proxy server. We present an algorithm - Window_EVR - for the LDQ proxy to compute the EVR for a window query result set. The simulation results show that LDQ proxy caching using the Window_EVR algorithm significantly reduces both the window query response time and the workload at the database servers while maintaining query result set accuracy.