This project developed an algorithm for recreating lost financial transactions for John Hancock. Specifically, it recreates premium, dividend, and loan transactions lost during the annual back up process in the years 1993 and 1998, using existing transactions and current aggregate amounts. In instances where random policyholder behavior was predicted, statistical analysis of historical policy behavior was provided. Recreation for some policies and fields was a deterministic problem, while other fields and policies were identified as not restorable by the algorithm.