How AQR combines value investing and quantitative analysis

Sergey Gurov

Nov 28, 2024

In one of the recent texts, senior partner of Movchan’s Group Elena Chirkova talked about the investment of the GEIST fund, which she advises, in one of the high-yield funds of AQR Capital Management, managed by a computer program. The stock selection in it relies on a method that combines a value approach and quantitative analysis based on the application of complex statistical methods for processing large data sets. What could be the value of such synergy?

In recent studies AQR Capital Management demonstrated, that the use of stock return prediction models, where the influence of individual significant factors is described by a simple linear function, often leads to suboptimal capital allocation in terms of the "excess return/volatility" ratio. Models that take into account real nonlinearities in the relationships between predictive variables and future asset returns generate higher Sharpe ratios. Analysts particularly emphasize that the improvement in long-short portfolio metrics occurs only when a small number of factors are included, which from a financial theory perspective should be directly related to stock returns. This is where the significance of the value approach manifests itself, capable of helping in the selection of such variables, improving the interpretability of the model, and avoiding the "garbage in, garbage out" situation (garbage in, garbage out). 

Another important result obtained by researchers: machine learning methods can be useful for solving the task of determining the appropriate timing for significantly increasing or decreasing market exposure. But even in this case, algorithmic market-timing strategies based on nonlinear models show good results when only economically significant variables are included. 

Why this matters  

Even if the computer program managing the fund from the AQR Capital Management group incorporates complex nonlinear models for predicting future asset returns, there are substantial reasons to trust it. The fact that the program follows value principles favorably distinguishes this fund from the vast majority of other quantitative funds. Many of them have high-risk strategies, often based solely on the analysis of historical asset prices without considering any fundamental economic indicators. Without this, as empirical studies show, such algorithms cannot achieve high returns without significant drawdowns over a long time horizon.