A Fast and Effective Solution to the Problem of Look-ahead Bias in LLMs