Quantitative investing: why is it not enough for a strategy to have worked in the past?
In quantitative investing, algorithms and mathematical models are used to support investment decisions. Strategies make decisions based on predefined rules that have been tested on historical data.
One of the greatest challenges, however, is that a strategy may look excellent based on past data yet still disappoint when put to the test in real market conditions. Impressive results achieved on historical data alone do not guarantee future success. This is why, when developing a strategy, the quality of the research methodology is at least as important as the investment idea itself.
In the world of quantitative investing, there are three fundamental research considerations that every investment model developer should keep in mind. To understand what makes quantitative investment strategies reliable over the long term, it is worth examining these three key principles:
- Lookahead Bias
- Curve Fitting (Overfitting)
- Survivorship Bias
Lookahead Bias: Time Travel
Lookahead bias occurs when a strategy or its underlying code accidentally uses information that was not yet available at the time the investment decision would have been made.
Example:
Suppose a strategy relies on corporate earnings data. A researcher downloads a database in which each quarter is associated with the company’s profit for that particular quarter.
For example, the company’s second-quarter 2025 earnings may appear in the database with a date of 30 June 2025. The problem is that the company did not actually publish those results until mid-August 2025.
If the strategy already uses this information in July, everything may appear perfectly reasonable at first glance: after all, the information is indeed shown in the 30 June row of the database. In reality, however, the model is using a figure that was not known to the market for another month and a half.
This can significantly improve the results of a backtest, even though the strategy is effectively relying on information that was unavailable at the relevant point in time. The error does not occur because the developer is intentionally “cheating”, but because historical databases naturally contain information that only became known later.
Curve Fitting: Overfitting the Data
Curve fitting, or overfitting, occurs when the rules of a strategy are tailored too closely to historical data. Instead of identifying a broadly applicable market relationship, the algorithm effectively ends up memorising historical noise.
Example:
Suppose you want to predict which football team will win its matches. You feed twenty years of football statistics and all kinds of alternative data into a model and spend weeks fine-tuning it. Eventually, the model discovers a peculiar rule:
“The team always wins when its captain wears neon-green socks, the temperature in the stadium is exactly 20°C, and the referee has had two cups of coffee before kick-off.”
This rule might even explain a narrow subset of winning matches in the twenty-year historical database with 100% accuracy. But does it represent a genuine underlying relationship in the sport?
Of course not. It is simply a coincidence.
When next week’s match arrives, this overly specific rule becomes completely useless. By chasing perfection in historical data, we have built a model that is extremely fragile when faced with the future.
Survivorship Bias: Looking Only at the Winners
Survivorship bias is the logical error of focusing exclusively on those that have “survived” a process while completely ignoring those that failed. This creates a distorted and overly optimistic picture of reality because the losers have simply disappeared from the dataset.
Example:
Imagine that you want to study the habits of successful restaurant owners. You walk down a busy high street and interview the owners of ten restaurants that have been operating successfully for twenty years. You discover a pattern:
They all serve pineapple pizza.
It might therefore seem reasonable to conclude that serving pineapple pizza is the secret to decades of success in the restaurant business.
But what has been overlooked?
The ninety other restaurants that opened on the same street over the past twenty years and subsequently went out of business – many of which also served pineapple pizza.
Because the analysis focused only on the “survivors”, it arrived at a false conclusion.
The same problem arises in investing when an equity strategy is tested only on companies that are still successfully operating today. This can easily exclude companies that went bankrupt or were delisted in the meantime, making the strategy’s backtest appear artificially safer and more profitable than it would actually have been.
Quantitative investing is an extremely powerful tool, but historical data is full of traps, illusions and logical pitfalls.
The purpose of research is not to create a perfect story about yesterday, but to build a robust framework capable of dealing with the unpredictable surprises of tomorrow. In quantitative investing, true value is not created by a single impressive backtest, but by the disciplined research process behind it, rigorous validation and the consistent management of risk.
Jogi nyilatkozat: A blog üzemeltetője a VIG Befektetési Alapkezelő Magyarország Zrt., a szerzői az Alapkezelő munkavállalói. A weboldal kereskedelmi kommunikációt tartalmaz. A blogon megjelenő cikkek magánszemélyek szubjektív véleményét tükrözik, tájékoztatási céllal készülnek és nem minősülnek befektetési elemzésnek vagy befektetési tanácsadásnak és nem tartalmaznak befektetési ajánlást. A blog szerzői saját nevükben kereskedhetnek olyan pénzügyi és pénzeszközzel vagy más termékkel, amelyről az általuk készített cikk közöl tájékoztatást vagy véleményt. Bár a szerzők tőzsdei vagy tőzsdén kívüli kereskedés során szerzett tapasztalata a jelen blogon szereplő írásaikban is megjelenhet, de érdekeltség nem befolyásolhatja az általuk közölt tájékoztatást. A blogon megjelenő cikkekben, hírekben és tájékoztatásokban megjelenhetnek olyan társaságok, amelyek üzleti kapcsolatot tartanak fenn a VIG Befektetési Alapkezelő Magyarország Zrt.-vel vagy a blog szerzőivel akár közvetlenül, akár a VIG Group cégcsoportba tartozó más vállalkozáson keresztül. Jelen blogon megjelent cikkek nem tartalmaznak teljes körű tájékoztatást, és nem helyettesítik a befektetés megfelelőségének vizsgálatát, amelyet csak az adott befektető egyedi körülményeinek értékelésével lehet megállapítani. A megalapozott befektetési döntés meghozatalához kérjük, hogy részletesen és több forrásból tájékozódjon!
A VIG Befektetési Alapkezelő Magyarország Zrt., a blog szerkesztői és szerzői nem vállalnak felelősséget a blogon szereplő tartalom naprakészségéért, esetleges hiányosságaiért vagy pontatlanságaiért, valamint a blogcikkek alapján hozott befektetési döntésekért és a befektetési döntésekből származó bármilyen közvetlen vagy közvetett kárért vagy költségért.