Quantitative Investing - Rules-Based Strategies and Systematic Edge

By EC Assets Research Team, Quantitative Strategy · Published 19 May 2026 · Updated 16 October 2026

Quantitative Investing: Quantitative investing uses mathematical models, statistical analysis, and computer algorithms to make investment decisions. The discipline ranges from simple rule-based strategies to complex machine-learning systems trading thousands of signals across global markets.

What Quantitative Investing Is

Quantitative investing makes decisions with explicit rules and models built from data. A quantitative manager specifies in advance what signals to use, how to combine them, how to size positions and how to trade, then lets the system run. The approaches range from factor investing, which tilts portfolios towards characteristics such as value or momentum, through trend following in futures, covered under managed futures, to statistical arbitrage over days and market making over seconds.

The appeal is discipline, scale and control. Rules can be tested, applied across thousands of securities without fatigue, and combined with explicit risk limits. The difficulty is that the same data used to build a rule is the data used to test it.

Machine learning has widened the toolkit, from text analysis of company filings and news to flexible, non-linear models of returns. It also sharpens the central difficulty: flexible models fit noise more easily, so out-of-sample testing, comparison with simple benchmarks and an economic reason why a signal should work matter more, not less.

How It Works

A typical process gathers data, defines signals, tests them on history, combines them in a portfolio with risk constraints, and trades it at the lowest possible cost; the last step matters because many signals are small and costs of trading can absorb them. The central statistical problem is selection. Every backtest has noise, and testing many variations makes it likely that the best one owes much of its result to chance. Harvey, Liu and Zhu (2016) counted hundreds of published factors and argued that a new one should clear a t-statistic of about 3, not the traditional 2. Bailey and López de Prado (2014) propose a deflated Sharpe ratio that adjusts a backtest for the number of trials, the length of the sample and the shape of returns.

Even genuine effects weaken once known. McLean and Pontiff (2016) found that the returns of published anomalies fell by 26 percent out of sample and by 58 percent after publication, consistent with investors trading them away. Strategies that many managers run can also fail together: Khandani and Lo (2011) show that the losses of August 2007 spread through quantitative equity funds as similar positions were unwound at once.

Worked Example

Suppose a researcher tests strategies on 10 years of data and none has any true edge. Each estimated Sharpe ratio is then noise with a standard error of about one divided by the square root of 10. The expected best result grows with the number of trials:

Expected maximum of independent standard normal draws; real strategy variants are correlated, which reduces the effect, but researchers rarely report how many they tried.

After 1,000 trials, pure noise is expected to produce a Sharpe ratio above 1, which would look excellent on a Sharpe ratio screen. The t-statistic hurdle of 3 corresponds to a Sharpe ratio of 0.95 over 10 years, and even that is cleared by chance after 1,000 trials.

Real factors show the problem from the other side. Momentum, buying recent winners and selling recent losers, was documented by Jegadeesh and Titman in 1993:

Kenneth French momentum factor, US stocks, monthly data; annualised geometric return.

The factor also lost 52.8 percent in 2009, 49.4 percent of it between March and May, when beaten-down stocks rebounded sharply; Daniel and Moskowitz (2016) show that such crashes come after market declines, when momentum is short the stocks most likely to rebound.

When It Applies (and Limitations)

Correlated trials. The table assumes independent tests; correlated variants inflate results less, but the number of effective trials is usually unknown to outsiders.

Regime change. Relationships estimated in one period may not hold in another, as momentum after 2008 shows.

Costs and capacity. Paper returns ignore trading costs and the price impact of managing large sums, which grows as a strategy attracts more capital and competitors.

Model and data risk. Errors in data, look-ahead bias and survivorship bias in backtests can create returns that never existed.

Why It Matters for Institutional Investors

Judge process over backtests. Asking how many ideas were tested, what was kept, and how results held up out of sample and live reveals more than the backtest itself.

Look through to factors. Many quantitative strategies, including those of hedge funds, load on common factors that investors could buy more cheaply; the residual is what fees should pay for.

Expect crowding. Overlap between managers means quantitative allocations can lose together; diversification across approaches and horizons helps.

Plan for decay. Assuming that published or widely used signals will earn less than their history sets realistic expectations.

References

1. Harvey, C. R., Liu, Y., & Zhu, H. (2016). ... and the Cross-Section of Expected Returns. Review of Financial Studies, 29(1), 5-68.
2. Bailey, D. H., & López de Prado, M. (2014). The Deflated Sharpe Ratio: Correcting for Selection Bias, Backtest Overfitting, and Non-Normality. Journal of Portfolio Management, 40(5), 94-107.
3. McLean, R. D., & Pontiff, J. (2016). Does Academic Research Destroy Stock Return Predictability? Journal of Finance, 71(1), 5-32.
4. Daniel, K., & Moskowitz, T. J. (2016). Momentum Crashes. Journal of Financial Economics, 122(2), 221-247.
5. Khandani, A. E., & Lo, A. W. (2011). What Happened to the Quants in August 2007? Evidence from Factors and Transactions Data. Journal of Financial Markets, 14(1), 1-46.
6. Jegadeesh, N., & Titman, S. (1993). Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency. Journal of Finance, 48(1), 65-91.