AQR: How to use Machine Learning to improve Your Investment Strategy

The folks over at AQR recently wrote a piece on how to use machine learning in stock portfolio construction. Specifically stock portfolio selection and timing. The article is quite technical but do provide some food for thought for less sophisticated investors on how AI can be implemented in an investment strategy.

Here are the key takeaways:

  • The Virtue of Complexity: Machine learning techniques can significantly improve stock selection strategies by capturing ‘nonlinear relationships’ between predictor variables and stock returns. Complex models that incorporate numerous nonlinear factors can outperform simple linear approaches by 50-100% in terms of portfolio performance.
  • Nonlinear Relationships Matter: The relationship between investment signals (like value and momentum) and stock returns is ‘complex and nonlinear’. Traditional linear models miss critical nuanced interactions between different investment signals. By using machine learning techniques to generate nonlinear predictors, investors can develop more sophisticated portfolio allocation strategies.
  • Performance Across Different Signal Sets: The research demonstrated the “virtue of complexity” across three different signal sets: Value and Momentum, Fama-French 5-Factor Model plus Momentum, Defensive-Oriented Signals

In each case, complex models generated substantially higher Sharpe ratios compared to simple linear models, with improvements ranging from 50-100%.

  • Caution Against Indiscriminate Data Mining: While complexity can improve portfolio performance, researchers warn against randomly adding predictor variables. Signal relevance is crucial – Including unrelated or noisy variables can rapidly degrade portfolio performance. The nonlinear predictors must have a meaningful relationship to expected returns.
  • Regularization is Key: To manage the challenges of complex models with many parameters, techniques like ridge regression help identify precise portfolio weights while preventing overfitting. This allows investors to leverage model complexity without sacrificing statistical reliability.

More complex models can better identify true nonlinear relationships and, thus, produce better stock selection strategy performance. […] Our results indicate performance improvements relative to a simple, linear approach in the range of 50-100%, suggesting that machine learning can help to build better stock selection portfolios.

Share the news

Disclaimer of liability

The above has been prepared by Børsgade ApS for information purposes and cannot be regarded as a solicitation or recommendation to buy or sell any security. Nor can the information etc. be regarded as recommendations or advice of a legal, accounting or tax nature. Børsgade cannot be held liable for losses caused by customers’/users’ actions – or lack thereof – based on the information in the above. We have made every effort to ensure that the information in the above is complete and accurate, but cannot guarantee this and accept no liability for errors or omissions.

Readers are advised that investing may involve a risk of loss that cannot be determined in advance, and that past performance and price development cannot be used as a reliable indicator of future performance and price development. For further information please contact info@borsgade.dk

You might also find this interesting:

Thomas Shrager: Superior Value Outside the U.S

In a comprehensive interview, Thomas Shrager and Jay Hill from the renowned New York value investing boutique Tweedy, Browne articulate their belief that the most compelling investment opportunities currently lie outside the United States. The veteran fund managers, who oversee portfolios for the 104-year-old firm, explain that international markets offer dramatically better value propositions than the overvalued U.S. equity market, which they describe as “priced for perfection” at 24 times earnings.

Aswath Damodaran: The Uncertain Payoff from Alternative Investments

Professor Aswath Damodaran’s latest analysis challenges the conventional wisdom surrounding alternative investments, revealing significant gaps between marketing promises and actual performance. Aswath examines how institutional and individual investors have increasingly embraced alternatives like hedge funds, private equity, and venture capital, often with disappointing results despite decades of compelling sales pitches.

Alternative investments have gained mainstream acceptance over the past two decades, moving beyond their traditional institutional confines to target individual investors. The core argument for these investments rests on two pillars: their supposedly low correlation with traditional stocks and bonds, and their potential to generate excess returns through superior management and market inefficiencies. However, Damodaran’s analysis suggests these benefits may be largely illusory when subjected to rigorous scrutiny.

Michael Mauboussin: How to Handle Intangibles in Modern Value Investing

Michael Mauboussin, Head of Consilient Research at Morgan Stanley, delivered a compelling keynote presentation at the Ben Graham Centre for Value Investing’s 2025 conference, addressing how the rise of intangible assets has fundamentally altered the landscape of value investing.

Drawing from nearly a century of investment wisdom while adapting to modern realities, Mauboussin argues that traditional accounting methods have become increasingly inadequate for evaluating companies in today’s intangible-heavy economy. His presentation reveals that intangible investments now represent 1.7 times tangible investments in the U.S. economy, a complete reversal from 1977 when tangible investments dominated by a factor of 1.4.

Cliff Asness: Missing the Best Days Isn’t the Real Problem

Clifford Asness of AQR Capital Management revisits his 1999 rejected paper that challenged one of the most common arguments against market timing. The widespread belief that missing just a few of the market’s best days destroys long-term returns is fundamentally flawed, according to Asness.

His analysis shows that while missing the best performing days does hurt returns, missing the worst performing days provides symmetrical benefits. The author demonstrates through both historical data and simulations that this “evidence” against market timing is mathematically obvious and essentially useless for investment decision-making.

Asness argues that legitimate criticisms of market timing should focus on investors’ lack of skill rather than cherry-picked scenarios of perfect incompetence. His 25+ years of out-of-sample data confirms these findings, showing the argument remains as flawed today as it was when first proposed.