Interview, Podcast
The Shift to Systematic Trading — Building Backtested Confidence · Dave Mabe
- Automated trading systems mitigate emotional decision-making during drawdowns, which are the primary triggers for traders quitting, whereas discretionary approaches require rigorous pre-planning to prevent least profitable choices when P&L fluctuates.
- Scaling trading performance is achieved by transitioning to automation to increase trade frequency from hundreds to thousands per year, though this should be executed gradually starting with automatic position sizing and exit orders rather than an overnight switch.
- Traders are advised to build a "path to confidence" through backtesting while acknowledging that real-world slippage and transaction costs will cause live results to underperform pristine backtest references.
- Strategies often deteriorate over time, requiring continuous optimization or the addition of rules that tell a coherent story to improve performance or filter out the poorest trades during drawdowns.
- Diversifying into multiple uncorrelated strategies is identified as the superior method for scaling a business, as relying on a single strategy creates vulnerability to prolonged drawdowns and a "fossilization" risk that limits growth.
- Live trading with tiny position sizes is recommended to quickly identify discrepancies between simulation and reality, allowing for faster learning compared to attempting to perfect theoretical backtests before going live.
- The use of AI tools like LLMs for coding is encouraged, provided the trader understands the logic and can audit the output, to overcome the limitations of intuitive or "on-the-fly" strategy development.
- Moving stops to break-even or using very tight stops in backtests can significantly reduce profitability or create false performance signals by masking precision issues that occur in live market execution.
- Traders who fail to create tools for rapid idea generation and backtesting, or who rely solely on intuition, struggle to distinguish between modelable strategies and purely discretionary insights.
- Building strategies based on actual live usage patterns, analogous to constructing sidewalks where foot traffic exists, is presented as a more effective approach than theoretical planning or committee voting.
- A pre-defined "back pocket" plan for handling drawdowns is essential to prevent traders from feeling flat-footed and making emotional errors when a strategy hits a significant loss.
- The gap between simulation and live results, particularly regarding fill precision, is a primary area for learning, and understanding this difference is necessary to avoid biases where tight stops artificially inflate backtest results.
- Continuous optimization processes are planned to run even after a strategy goes live to prepare for future deterioration and the unknowable aspects of market randomness.
- Traders must spend time at a specific account size to "earn their keep," as mistakes made with smaller sizes often recur when position sizes increase to generate substantial income.
- Scaling "wide" by taking more trades across uncorrelated strategies is expected to be a better long-term approach than scaling "up" by increasing position size on a single strategy.
- Strategies that have been on the shelf are often inferior because highly profitable systems would likely be utilized otherwise, suggesting that active development and testing are required for sustained success.
- The art of trading involves managing the unknowable aspects of drawdowns and randomness, where a "humbling" realization that systematic methods outperform intuition serves as a critical turning point for many.
- Relying on a single strategy without a plan to handle deterioration leads to emotional decision-making, whereas having a pre-planned approach allows traders to navigate drawdowns without quitting.
- Partial profit-taking strategies are expected to result in significantly lower profitability compared to letting positions run to natural exits, based on backtest comparisons.
- Creating a process for reconciling backtests with live trades is necessary for establishing a path to confidence, which is critical for trading the significant sizes required to make substantial money.
- Understanding that backtests lack the precision to catch all stop-outs in real life is necessary to avoid overconfidence in theoretical models and to anticipate the "unknowable" nature of future market behavior.