Quantitative Investing
Finding statistically measurable edges in data and executing them systematically, with position sizing treated as a mathematical problem.
Overview
Quantitative investing replaces judgment about individual securities with statistical patterns tested across large samples. Its distinguishing feature is that the edge must be measurable before it is acted on, and that how much is staked is treated as seriously as what is bought.
How the philosophy developed
Ed Thorp established the template. He worked out a card counting system for blackjack using an IBM 704, published it in 1962, and then applied the same procedure to securities, showing that warrants were systematically mispriced relative to the shares they converted into and that the gap could be hedged.
Thorp was pricing and hedging convertible securities before the option pricing formula that later made the technique standard was published in 1973, and he brought the Kelly criterion into investment practice, making position sizing a calculation rather than a matter of nerve.
Jim Simons took the approach to its logical conclusion at Renaissance Technologies from 1982, hiring mathematicians and scientists specifically because they had no market theories to defend, and trading statistical relationships across enormous numbers of positions. Separately, Eugene Fama and Kenneth French produced the factor models that gave the academic version of the same idea its vocabulary.
Core principles
- An edge must be measurable and verified against data before capital is committed, and anything that cannot be measured is opinion.
- How much is staked matters as much as whether the position is right, because a real edge sized badly still ends in ruin.
- A small statistical advantage becomes reliable only when repeated across a large number of independent positions.
- Understanding why a pattern exists is useful but not strictly required, which is where this tradition parts company with every other approach on this site.
How decisions get made
Research begins with data rather than with a company. A candidate relationship is specified precisely enough to be tested, then examined across the longest available history and across samples it was not developed on.
Surviving signals are combined and sized. The Kelly criterion gives a mathematical answer to how much of a bankroll to stake given the edge and the odds, and practitioners commonly stake a fraction of that amount because full Kelly produces swings few can tolerate.
Execution is systematic and largely automated, which is deliberate: the point of specifying the rule in advance is that no judgment is exercised at the moment of trading, where judgment is least reliable.
How it approaches valuation
Most quantitative strategies do not value securities at all in the sense a value investor would recognise. They look for relationships in prices and related data rather than for a gap between price and intrinsic worth.
The academic branch is closer to valuation, using characteristics such as company size and the ratio of book value to market price to explain differences in long-run returns. That is a description of what has happened rather than an estimate of what a business is worth.
How it approaches risk
Risk is managed by arithmetic rather than by conviction: position size, diversification across a large number of positions, and limits set in advance and not overridden.
The specific danger the tradition worries about is ruin, meaning a drawdown large enough to end participation before the edge can assert itself. Ed Thorp's writing returns repeatedly to the point that surviving matters more than being right.
A second, subtler risk is that a measured edge was never real, having been found by searching the same data many times until something appeared. Testing on samples the signal was not developed on is the standard defence.
How portfolios are built
Very large numbers of small positions with short holding periods, so that no single position matters and the statistical average can assert itself.
Hedging is used to isolate the specific relationship being traded and remove exposure that the model has no view on, which is what Thorp's warrant work demonstrated and what most market-neutral strategies still do.
Time horizon
Usually short, from days to months, because statistical edges tend to be small and depend on repetition rather than on a thesis maturing. The academic factor version operates over years instead.
Where the approach can work well
- It forces an edge to be measured before money is committed, which rules out most of what passes for analysis elsewhere.
- It makes position sizing an explicit calculation rather than a matter of confidence.
- Systematic execution removes the emotional decisions that damage discretionary investors.
- Results can be attributed, because the rule was specified in advance and its behaviour can be examined afterwards.
Limitations and criticisms
A balanced view includes where the approach struggles, presented neutrally.
- Almost none of it is executable by an individual. The infrastructure, data and execution speed that made the leading firms work are not available outside a large organisation.
- Edges decay as they become known, so the research has to keep moving and a strategy that worked is not evidence that it will continue to.
- Backtests are easy to fool. Searching the same data repeatedly will produce patterns that were never real, and only out-of-sample testing offers protection.
- A model has no view on conditions it never saw, so unprecedented events are exactly where systematic approaches fail.
- The Kelly criterion is aggressive, and Paul Samuelson argued for years that maximising the growth rate of wealth is not what most people actually want.
Common misconceptions
- The claim
Quantitative investing removes human judgment.
What is actually the caseIt moves it upstream. Humans choose the data, design the tests, decide what counts as a signal and set the risk limits. What is removed is judgment at the moment of trading.
- The claim
A good backtest means a good strategy.
What is actually the caseA backtest shows that a rule would have worked on data it was developed against, which is a much weaker claim. Testing on samples the rule never saw is the minimum standard, and even that is not conclusive.
- The claim
Individuals can run these strategies at home.
What is actually the caseThe famous records depended on data, infrastructure and speed unavailable outside a large firm. What does transfer is the discipline of measuring an edge and sizing positions deliberately.
Investors associated with this approach
Listed because of a documented intellectual connection to the approach, not because they are well known.
In their words
“The amount you bet matters as much as whether you are right.”
Context: Thorp was writing about position sizing, worked out mathematically across many repeated bets. The betting frame belongs to that setting.
“I learned that being right is not enough; you have to be right and survive.”
“I take the market efficiency hypothesis to be the simple statement that security prices fully reflect all available information.”
Context: Fama's statement of the efficient-market hypothesis, the idea behind the case for index funds.
Strategies that put this into practice
A philosophy is what an investor believes. These are the procedures people run on the strength of it.
Related guides
Related concepts
Useful tools
Markets and hubs
Investor comparisons
Related philosophies
Frequently asked questions
What makes an approach quantitative rather than discretionary?
Identifying statistically measurable relationships in market data and trading them systematically, usually across large numbers of positions. The defining requirement is that the edge is measured and tested before capital is committed rather than argued for.
How do quantitative investors decide position size?
A formula for how much of a bankroll to stake on an opportunity, based on the size of the edge and the odds available. It maximises the long-run growth rate of wealth at the cost of accepting large swings, which is why practitioners commonly stake a fraction of what it suggests.
Do quantitative investors care why a signal works?
Traditions differ. Renaissance Technologies deliberately hired scientists with no market theories and treated the question as secondary to whether the relationship held. Academic factor researchers care a great deal, because a premium without an explanation may be a statistical artefact.
Why do quantitative edges stop working?
Because they become known. Once enough capital pursues the same relationship, the mispricing it depended on is competed away. This is why the research has to keep moving and why a historical record is not evidence about the future.
Can an individual investor do this?
Not the institutional version, which depends on data, infrastructure and execution speed unavailable outside a large firm. What transfers is the discipline: insisting an edge be measurable, sizing positions by arithmetic, and accepting that survival matters more than being right on any single position.
Get smarter about investing
Clear market insights, useful tools, and beginner-friendly investing education.
Educational content only. This page explains how an investing approach works and where it falls short. It is not a recommendation to adopt it, not investment advice, and not a claim that any approach suits your circumstances.