Suggested Approach for Phase 1 Portfolio Creation with IS and OOS Filtering
In phase 1, the goal is to create several robust portfolios by filtering and ranking strategies to achieve high performance and stability, with a specific focus on their behavior in both in-sample (IS) and out-of-sample (OOS) testing.
- IS and OOS Filtering:
- IS: Ensure strategies perform consistently during the in-sample period by prioritizing parameters like CAGR/Max Drawdown (DD%), Sharpe Ratio, and Stability. These metrics demonstrate the strategy’s profitability and risk management in the training phase.
- OOS: Use filters that emphasize robustness and adaptability to unseen data. High values for metrics like Sharpe Ratio, Stability, and low OOS drawdowns indicate resilience. Also, analyze parameters such as Open DD% and consecutive losses to evaluate risk in live scenarios.
- Prioritize Key Performance Metrics for Filtering and Ranking:
- CAGR/Max DD%: A high ratio here is essential for long-term profitability.
- Sharpe Ratio: Select for a Sharpe Ratio > 2 in both IS and OOS to ensure that returns consistently outweigh risk.
- Profit Factor: A Profit Factor > 1.5 across IS and OOS indicates that profitable trades outweigh losses.
- Stability: Aim for Stability > 0.8, ensuring strategies are less likely to be overfitted.
- Open DD% and Max DD%: Filter for Open DD% < 15% to limit real-time exposure to high drawdowns, with Max DD% providing an additional safeguard.
- Net Profit and R-Squared (R²): These are supportive metrics, with higher values indicating stronger trends in profit generation.
- Portfolio Structuring:
- From the filtered strategies, construct 4–6 portfolios, each containing 4–8 strategies that exhibit the highest OOS Sharpe Ratios, Stability, and low drawdowns. Ensure each portfolio balances a mix of high-return and stable strategies.
- Once these portfolios are built, remove selected strategies from the list for the next round of portfolio creation, or, alternatively, create all portfolios from the initial list to maximize the use of all high-performing strategies without deletion.
This approach ensures each portfolio is diversified in both strategy type and performance metrics, optimizing for long-term stability across various market conditions.
