Actually, there are many traps in the back-testing. historic data, time frames, time ranges, data quality, subjective selection of markets, stocks, and errors in using historical data, software problem, etc, these are all problems that causes often too-good-to-be-true results.