163
16.4. Summary
This step is the make or break step of the entire design process. In the step, the product
team will have produced an algorithm that can beat the benchmark or they have proven that
the current strategy is wrong and they need to go back to Stage 1 to change the algorithm.16.4.1. Best Practices
● Use fishbone diagrams to identify special causes of variation.
● To ease reuse, separate raw data from the returns and from the indicators.
● Screen instruments and lock down the investable universe.
● Estimate transaction costs and slippage and incorporate them into backtests.
● Plan and document the trading/investment rules and backtesting methodology before
you begin.16.4. SUMMARY