The convergence of quantitative and discretionary strategies

Wall Street has always had its tribes, and when it comes to investment styles, discretionary and quantitative investors part the land. Both are heavily data-driven. Discretionary investors carefully analyze fundamental data (economic, sector, and company data) to make buy or sell decisions, often holding positions until stocks become overvalued. Their human touch provides more precise signals, which makes it reasonable for them to focus on specific, tailored investment portfolios. Quantitative investors rely as well on fundamental data and utilize mathematical models along with extensive datasets to diversify signals at a large scale, seeking statistical opportunities.

Quants, as they are often called, trace their roots back to the early 20th century when mathematicians and physicists like Louis Bachelier applied their expertise to finance. Quant trading gained traction in the 1950s and 1960s with the advent of computers and mathematical models. Deregulation, better data access, and technological advancements made it soon afterwards a mainstream feature of the hedge funds and institutional investors landscape.

If investing was like driving a car, discretionary investors would be driving stick-shift, reviewing the map, and gaining in-depth knowledge of the townships along their way. For them, the emergence of "quantamental" approaches is akin to adding side radars and night vision to their windshield: they remain entirely in control, but they gain more insights.

For much of its history, quantitative trading has operated in isolation from discretionary portfolio managers. That’s notably because running quantitative algorithms requires substantial data, computing infrastructures, and data scientists with PhDs to harness them. Quantitative insights thus remained inaccessible to discretionary investors who lacked resources to create them. That was not the only reason though.

Discretionary investors typically hold stocks for months to years, while quantitative traders use mathematical models to trade stocks from milliseconds up to several weeks. Technical hurdles aside, the overlap was minimal. Until it wasn’t.

The rise of quant-like approaches in discretionary investing

Recent years have ushered in a new era for discretionary investors, with two simultaneous shifts. On the one hand, quantitative models embraced new approaches that were more likely to produce insights relevant to discretionary investment decisions, like predicting inflation figures in real time, or dynamically evaluating the ESG profile of companies. On the other hand, the tools of quantitative research began making their way to the dashboards and models used by discretionary investors. Rather than deciding to buy or sell bonds and stocks, these quant insights were simply better informing discretionary traders who retained control over their decisions. This evolution has given rise to a new cohort of investors known as "quantamental" practitioners.

If investing was like driving a car, discretionary investors would be driving stick-shift, reviewing the map, and gaining in-depth knowledge of the townships along their way. For them, the emergence of quantamental approaches is akin to adding side radars and night vision to their windshield: they remain entirely in control, but they gain more insights.

At a technical level, joining the dots between heaps of data and discretionary investing presented a double challenge: improving quantitative relevance, and facilitating access.

Quantitative analysts began developing signals that could be used in the long run, not just in the hours or days that followed their inception. For instance, they looked for macroeconomic signals to assess economic regimes and identify shifts in market cycles.

At a technical level, joining the dots between heaps of data and discretionary investing presented a double challenge: improving quantitative relevance, and facilitating access.

Moreover, advancements in natural language processing (NLP) and machine learning (ML) have enabled the development of new signals based on alternative data sources. These provide more timely insights beyond traditional announcement and filing data, offering a holistic view of companies and economies, especially during periods of uncertainty and market turmoil. For instance, it is now possible, using natural language processing approaches, to read through thousands of pages of investor call transcripts and produce a synthetic view on company or economic outlook in minutes rather than weeks. The signal summarizing these views complemented company and sector-level insights, and could refine the views of stock analysts.

The format and delivery of quantitative insights have also evolved. While Excel remains the workhorse of Wall Street, its limit of just over 1 million rows per spreadsheet is not sufficient for data scientists, and in many cases, the size of alternative datasets so far precluded any contribution to the discretionary decision-making process. Data accessibility has evolved however, and today, terabytes of data can be condensed into simpler, more actionable data points, opening new possibilities for discretionary investors.

While they were already used to analyzing specific drivers and dimensions of the investment workflow as factors, these investors have seen the emergence of new dimensions, from media attention and sentiment to nowcasts and controversies, that augment the slate of factors available to them, and better enable them to integrate sophisticated data models into the discretionary decision-making process.

How sentiment-based insights can enhance stock-selection and risk-management

High-frequency, real-time market sentiment indicators provide fundamental investors with a powerful tool to enhance stock selection. These indicators have been proven to have a direct impact on equity returns. Sentiment analysis tools can scan news articles and textual data to uncover trends in market sentiment. This information, often hidden from traditional fundamental analysis, aids in identifying undervalued stocks.

Quantitative strategies based on alternative data also offer a systematic framework for assessing and mitigating risks. For example, by processing news data at scale, some models provide valuable signals for risk models. These signals assist in forecasting equity market volatility and improving beta forecasts. News-based signals also shed light on market behavior, including the spillover effect from firm-level news to broader market drivers, which is valuable for portfolio managers.

Driven by technological progress and accessibility, alternative data is fueling a new generation of quantitative factors. Meanwhile, discretionary traders are gaining broader access to quantitative tools. This is leading to a natural convergence between the two worlds, combining traditional approaches with cutting-edge quantitative insights to create a more comprehensive and effective investment landscape. This convergence is reshaping the investment landscape and enhancing the effectiveness of investment strategies across the board.

Key Takeaways:

• Discretionary investors are increasingly adopting quantitative datapoints to augment their fundamental analysis with alternative insights;
• Nevertheless, integrating quantitative methods into discretionary investment decision making processes still presents logistical challenges, including the need of a robust data infrastructure and testing the signal relevance;
• Lately, a new generation of quant-originated factors has emerged. Harnessing alternative data, like news and sentiment, they make actionable and timely insights more accessible and enhance stock selection, asset allocation, and risk management.
• Quantitative insights like GDP and inflation nowcasts, as well as stock-specific factors such as ESG controversies and sentiment indicators offer unexpected advantages that extend beyond short-term gains, providing support for both timely decisions and long-term strategies.