Quantitative investing involves disciplined rule-based approaches to help investors structure optimal portfolios that balance return and risk. How has such investing evolved? In their June 2018 paper entitled “The Current State of Quantitative Equity Investing”, Ying Becker and Marc Reinganum summarize key developments in the history of quantitative equity investing. Based on the body of research, they conclude that:
- Rigorous quantitative investing commences in the early 1950s with modern portfolio theory, which employs historical mean return to estimate future return and historical return variance to estimate future risk.
- The capital asset pricing model (CAPM) follows in the 1960s, specifying the first factor model by introducing market beta. Contemporaneous computing advances enabled considerable data analysis and an ongoing debate over market efficiency.
- By the late 1970s/early 1980s, discovery of stock return anomalies (particularly those related to size, value and seasonality) complicate the explanatory landscape.
- By the 1990s, researchers began incorporating anomalies (re-named factors) into benchmark models of stock returns, most prominently as a market-size-value 3-factor model and a market-size-value-momentum 4-factor model.
- Research discovering other widely accepted factors including liquidity, quality, volatility, profitability and investment subsequently appears. Rapid expansion of discovered factors triggers backlash regarding research methods (data snooping bias).
- Practical factor investing (smart beta) then emerges to exploit a growing set of factors via passive weighting tilts that, based on past research, offer superior return-risk outcomes.
- Active quantitative investors currently seek to outperform relatively passive smart beta indexes via: (1) factor timing based on external conditions or factor performance; and, (2) fresh and subtle big-data insights (often focused on near real-time investor sentiment).
In summary, the body of research suggests that quantitative investing has evolved toward practical factor investing, factor timing and subtle big-data exploitation.
The list of references in the paper offers a long list of other papers that make major contributions to quantitative investing.
Cautions regarding this historical overview include:
- Much of the cited research is gross, not net. Findings may differ for an analysis including estimated implementation frictions.
- As noted in the paper, factor discovery is subject to data snooping bias via model variations, parameter settings and sample selections. Thousands of researchers building on past research compounds concern about bias.
