No. You don't need to backtest a trading strategy.

About this video

• Video Title: No. You don't need to backtest a trading strategy.
• Channel: Roman Paolucci
• Speakers: Roman Paolucci
• Duration: 00:25:13

Overview

This video argues against the necessity of traditional backtesting for trading strategies. The speaker, Roman Paolucci, emphasizes that true portfolio success comes from understanding and combining "games" (orthogonal return streams) that are physically and stochastically independent, rather than trying to predict past performance. He illustrates this with code examples, demonstrating how combining assets like SPY (equities) and DBMF (managed futures) can lead to better risk-adjusted returns and reduced drawdowns, especially when leverage is applied thoughtfully, by mitigating volatility drag.

Key takeaways

• Rethinking Backtesting: Traditional backtesting is criticized as a form of overfitting historical data, as future market behavior is inherently unpredictable.
• The "Casino Principle": The core idea is that a casino's success isn't based on individual player performance but on the underlying "game" mechanics. Similarly, investment success relies on understanding the underlying "games" of different asset classes.
• Orthogonal Return Streams: The key to superior portfolio construction lies in combining assets that are "physically and stochastically independent" (other games), which offer diversification benefits beyond simple correlation.
• Managed Futures as an "Other Game": ETFs like DBMF (managed futures) are presented as an example of an asset class with low correlation to equities, offering a diversification benefit.
• Leverage and Volatility Drag: Applying leverage to a diversified portfolio of uncorrelated assets can amplify returns and improve the geometric growth rate, provided it's done strategically to combat volatility drag, unlike naive leverage on single asset classes.
• Portfolio Engineering over Prediction: Professional investors focus on engineering portfolios with robust structures and uncorrelated return streams rather than trying to predict market movements.