Quant’s New Frontier

Jina Yoon | Chief Alternative Investment Strategist

Last Updated:

Additional content provided by Michael McClain, AVP, Research.

Famed investor and philanthropist James Simons passed away in May. Simons was the founder of Renaissance Technologies and led the Medallion Fund, likely the most successful hedge fund ever. Over a 30-year period from 1988 to 2018, estimates are the Medallion Fund annualized at 66% before fees and 39% after fees1. While details on the firm’s investment process are highly guarded, it’s accepted that their use of advanced machine learning techniques was instrumental in their process. As an accomplished mathematician and former Cold War code breaker, Simons' technical expertise led to incredible success and launched widespread interest in quantitative investing. Today, as the investment industry continues to play catch-up with techniques likely utilized by Renaissance decades ago, assessing how artificial intelligence (AI) and machine learning (ML) are employed is at the industry’s forefront.  

It should be noted that while AI and ML are often used interchangeably, there are distinctions. AI is a larger field that aims to build computers capable of mimicking human cognitive functions such as reasoning, learning, and recommendations. ML is a branch of AI that involves teaching machines to learn from data without being explicitly programmed. While the number of use cases and future applications are only going to grow as computing power and market interest increases, let’s review a few relevant examples as many investment firms often promote how they utilize ML in a rather vague manner. As an incredibly dense field without an accepted list of best practices, specific examples should help cut through the everyday jargon.  

Return Predictions Using Neural Networks and Decision Trees. Investment firms use neural networks and decision trees as part of their predictive modeling tools to help forecast investment returns. Neural networks are a type of ML that can identify complex patterns and relationships in data. Firms will train neural networks using historical market data, such as stock prices and economic indicators, to predict future market movements and investment returns. These models can consider a wider range of variables and factors that may impact investment performance. Decision trees are another type of predictive model that use a tree-like structure to make decisions based on input variables. Decision trees can also be used to analyze historical market data and identify key factors that are correlated with investment returns.   

Risk Management with Random Forests. Random forest is an ML algorithm based on combining multiple decision trees to make predictions. They may be employed for: 

  • Portfolio Optimization: By assigning weights to different assets based on their expected returns and risks, the algorithm can help identify the most efficient allocation of assets to maximize returns while minimizing risk. 
  • Risk Assessment: Assess the risk of individual investments or the overall portfolio. By analyzing historical market data and key risk factors, the algorithm can help identify potential risks and forecast the likelihood of different risk scenarios. 
  • Stress Testing: Used to simulate different market conditions and stress test portfolios. By running scenarios with varying levels of market volatility or economic downturns, the algorithm can help assess the resilience of the portfolio and identify potential vulnerabilities. 

Alternative Data Analysis with Natural Language Processing (NLP). Alternative data refers to non-traditional data sources such as social media sentiment, web scraping, satellite imagery, and credit card transactions. By combining alternative data with NLP, firms can develop unique insights that are not captured by traditional financial data. NLP gives computers the ability to interpret, manipulate, and comprehend human language.   

  • Sentiment Analysis: Used to analyze text data from sources like social media, news articles, and earnings call transcripts to gauge investor sentiment towards specific companies or industries. By understanding market sentiment, firms can make more informed investment decisions. 
  • Earnings Forecasting: Extract key information from financial reports and earnings call transcripts to predict company performance and earnings. By analyzing language patterns and sentiment in these documents, firms can forecast earnings with more accuracy. 
  • News and Event Analysis: Analyze news articles and events that could impact market movements, such as mergers and acquisitions, regulatory changes, or product launches. By extracting actionable insights from text data, firms can react quickly to market-moving events. 

Trade Execution with Reinforced Learning. Reinforced learning can be used to optimize trading strategies and execute trades more efficiently. Reinforcement learning involves training an algorithm to make decisions by trial and error, receiving feedback in the form of rewards or penalties based on the actions taken.   

  • Trade Execution Optimization: Optimize trade execution strategies by training algorithms to make decisions on when and at what price to place trades. The algorithm learns from past trading data and market conditions to determine the most effective way to execute trades while minimizing transaction costs and market impact. 
  • Market Making: For firms with market-making strategies, where they provide liquidity by continuously buying and selling assets, the algorithm learns to adjust bid and ask prices based on market conditions and order flow, optimizing profits while managing risk. 
  • Automated Trading: An algorithm can continuously learn and adapt to market conditions, making timely decisions to take advantage of trading opportunities. 

While this is not meant to be an exhaustive list, ML in the investment industry will continue to grow. It’s not expected all asset managers will move to a systematic process, nor should they, as poor model design and overfitting may be more harmful than excluding ML in their process. However, with so many use cases, we believe at some point all firms will benefit by employing them in some part of their investment process.   

1 “The Man Who Solved the Market: How Jim Simons Launched the Quant Revolution” by Gregory Zuckerman (2019) 

Jina Yoon profile photo

Jina Yoon

Jina Yoon is LPL Financial’s Chief Alternative Investment Strategist. Her investment career includes over 15 years of experience.