The Anatomy of Quant Equity Funds
strategies, structure, talent
In this post I will describe the strategies deployed by systematic long/short equity funds based on my experience working in Quantitative Research.
Equity Factor Investing
The last firm I worked at ran a strategy broadly known as quantitative equity portfolio management - a high-capacity factor based approach to running long/short equity portfolios. Factor based portfolio management aims to decompose equity returns into alpha factors and risk factors. Alpha factors are categorized into 4 classes:
Value
Momentum
Growth
Quality
The task of an alpha researcher is to devise increasingly accurate and granular methods of decomposing equity returns. These individual alphas are recombined downstream to create factor scores for each equity within the traded universe. These scores, along with a risk and transaction cost model are fed into a portfolio construction algorithm. The portfolio construction model spreads desired factor exposures among as many securities as possible. This approach minimizes the idiosyncratic risks present in individual stocks.
The task of the portfolio manager is to determine what mix of factor exposure is desirable based on an analysis of the market regime. If you keep up with quant news, you will know that AQR ended a multi-year stretch of poor returns by tilting their factor exposure towards value in 2022.
A quantitative equity portfolio is akin to devising the perfect stock for the current market conditions, and building a characteristic portfolio that represents that perfect stock using thousands of equities as small pieces. The weight of each position acts as a coefficient in a linear combination to attain target exposures. Without highly sophisticated technology, it is impossible to manage the number of positions required to build such a portfolio.
While alpha factors represent exposures to be maximized, risk factors are to be minimized. Risk factors encompass any set of attributes that the portfolio wants to avoid. Examples include
Beta
Volatility
Market capitalization
Sector-specific correlation
The goal is to have neutral weighting towards risk factors. Some risk factors are positive predictors of equity performance over time. Nonetheless, if a risk factor falls outside the mandate of an investment portfolio, its desired exposure is as close to 0 as possible.
Institutional investors differ from retail investors in that they already carry intrinsic exposure to risk factors. Suppose you are Jeff Bezos with a large portion of your net worth tied up in Amazon. Amazon is a high beta stock with an enormous market cap. The last thing Jeff is willing to pay 2 and 20 for is a portfolio that fluctuates in tandem with his net worth. He desires an investment vehicle with statistical independence.
For further reading on this class of strategies I recommend this textbook.
Talent
Quantitative hedge funds broadly employ three types of investment professionals:
Researchers
Developers
Traders
Researchers
The research team is responsible for the generation of intellectual property - typically in the form of new models, improvements upon old models, or insights into the underlying properties of financial markets. The daily tasks of a quant researcher span:
Finding new ways to minimize market impact/transaction costs
Research and development of new alpha factors
Analyzing the contribution of risk vs alpha to past portfolio returns
Studying strategy simulations on past market data
Researchers are skilled in the following:
Probability and Statistics
Algorithms and data structures
Time series analysis and econometrics
Machine learning
Numerical Analysis (scientific computing)
Programming (R/Python)
SQL (data manipulation)
Finance and portfolio theory
Alpha Research
The goal of alpha research is to conduct a cross-sectional post risk analysis of stock characteristics that forecast under-performance or out-performance. The cross-sectional study is later extended to a panel-wise analysis.
Post Risk: After adjusting for the effects of risk factors
Cross-section: A set of returns spanning the entire equity universe at a single point in time
Panel-wise: Returns spanning the entire universe across an extended time period
Once research teams have verified the validity of a proposed factor, it is integrated into a production alpha system.
Transaction Cost Research
Transaction costs eat away at the returns of a hedge fund once the AUM scales. One of the properties of transaction costs is that they are quadratic rather than linear. T-cost researchers will study new ways of executing orders so that positions can be entered whilst minimally impacting the market. T-cost models forecast transaction costs of proposed trades.
Risk Research
Risk researchers primarily develop models that forecast volatilities of equities within the traded universe. The same factor/cross-section/panel cycle from the alpha research applies to risk research. Risk researchers also analyze tail-risk scenarios and run stress tests of the portfolio on simulated market conditions. An execution risk model forecasts the effect of entering/exiting positions on portfolio volatility.
Portfolio Construction Research
The portfolio construction model is what generates the trade decisions. It’s job is to synthesize information from the alpha, risk, and transaction costs models and determine whether proposed trades will bring the portfolio closer to its performance targets.
Let’s illustrate this with a toy example
Your portfolio holds 20% of the daily volume of a stock. The forecasts for the returns of this stock are negative
Exiting the position ahead of negative returns will generate steeper losses through transaction costs and market impact
Therefore, it is best to leave the position unchanged, or liquidate it partially
Portfolio construction researchers often have expertise in decision theory, operations research, and optimization. The task of building portfolios is a multivariate constrained optimization problem.
Development Team
The development team is responsible for building the software and data infrastructure that supports the operation of the entire fund. This is a broad set of responsibilities that include:
Designing research and production databases
Automating the cleaning and formatting of data
Integrating data vendor API’s into existing software systems
Building statistical libraries to streamline research
Optimizing high performance computing systems to run market simulations
Implementing alpha/risk/T-cost models for live trading systems
Building applications that display live portfolio analytics
Designing applications used by traders/portfolio managers
A skilled development team is crucial to the success of the hedge fund. While the researchers build the models; developers implement them into productionized software systems. Furthermore, software errors can be catastrophic to PnL. Quantitative developers are skilled in the following:
Algorithms and data structures
High performance computing
Networks and system design
Scientific computing (numerical analysis)
Software validation and testing
Cloud computing
Probability and Statistics
C++ or some other low level language
There is frequent cross collaboration between researchers and developers. Oftentimes a developer has a specific skill crucial for a research project and vice versa. Without good data, reliable systems, and high performance models; the fund is little more than a stack of jupyter notebooks.
Traders
The responsibilities of a trader vary. This depends on the level of automation present within the firm. Traders monitor live performance and adjust risk parameters. They might act as a buffer between orders generated by the portfolio construction model and live execution. This buffer can entail manually entering, managing, and exiting positions to optimize alpha transfer.
If the fund trades equities in emerging/frontier markets, they will manually manage illiquid positions.
Traders often collaborate with portfolio construction researchers on projects to bridge the gap between theory and applicable market dynamics.
Traders are typically skilled in the following
Probability and Statistics
Market microstructure
Quickly piecing together qualitative and quantitative information
Finance/Portfolio Theory
Rational decision making
Conclusion
This article covers the structure of factor oriented quant equity funds. There are many other types of hedge funds such as statistical arbitrage, relative value, and HFT. From an AUM standpoint - equity funds account for the lion share of capital deployed in markets through quantitative strategies.