We study the problem of mean retrieval score estimation for query performance prediction (QPP). We propose an enhanced estimator which estimates the mean based on calibrated retrieval scores. Each document score is adjusted based on features that model potential tradeoffs that may exist in the retrieval process of that specific document. Using the proposed estimator, we derive several previously suggested QPP methods, from which we gather an initial set of calibration features. Based on these features and few additional ones, we propose two estimator instantiations. Using an evaluation over several TREC benchmarks, we demonstrate the effectiveness of our estimation approach.