Is there any benefit to investing in active funds?

For most investors, individuals and pension funds alike, the choice is pretty simple. You can put your money in a passive fund—a fund that simply tracks the performance of a market index, an S&P 500 say, with no ambition to beat it. Or you can gamble on higher returns by investing in an actively managed mutual fund, run by expert stock pickers with their sights set on alpha – returns that outperform the index by capitalizing on under-priced assets. Active funds are de facto a riskier proposition than passive funds, but are they a gamble worth taking? Not if you look at recent performance, they’re not.

According to Standard and Poor’s Index vs. Active scorecard, a whopping 65% of all actively managed large-cap US equity funds underperformed the S&P 500 in 2024 – and this despite the kind of geopolitical and economic volatility that typically augurs well for shrewd active managers looking to exploit turbulence to make fast gains. In fact, compared to index funds, active mutual funds have had relatively poor innings for the last quarter of a century, according to the S&P scorecard. Which begs the question: is it worth investing in active funds at all?

Weighing in here with new research is LBS Professor of Management Science and Operations, Victor DeMiguel. In a forthcoming paper co-authored by colleagues from the Universitat de Pompeu Fabra in Barcelona and Madrid’s CUNEF and CarlosIII universities, he argues that actively managed funds can yield positive net alpha—but that getting your hands on the treasure at the end of the rainbow takes highly sophisticated prediction methods; machine learning methods that he and his colleagues outline in their paper. That said, the majority of investors will take some convincing.

The shift from active to passive

The last decade or so has seen a seismic shift away from actively managed mutual funds. Since 2011, more than $3 trillion worth of assets has moved from active to passive funds as investors forego the “smart guys” in favour of safer bets – or not “doing anything fancy with their money,” says Victor.

It’s a move that has prompted many market observers to call time on active fund investing – a debate that is “essentially over,” as a well-known business columnist puts it. According to Michael Hiltzik, the debate “isn’t really about whether index funds perform better than actively managed funds — that debate is essentially over, and indexing wins, hands down.”

This does not come as a surprise to DeMiguel. “We know that index funds have consistently outperformed actively managed funds for more than 10 years. And when you factor in the cost of managing them – most charge investors around 1% in annual fees compared to 0.5% or less for passive funds—it’s hardly surprising that most people will walk away.”

So is there any benefit to investing in active funds?

There is some indication that a small proportion actually do outperform the market each year, yielding outsize returns for investors intrepid enough to have entrusted their cash. While around 90% of active equity funds underperform their index, a minority are still able to beat the trackers. And they do so on the basis of their ability to spot under or over-priced stock in a given index – an expertise that can also help drive efficiency in the market by figuring out the true value of assets. Nonetheless, for investors, the issue is identifying these outliers in advance, and this is notoriously difficult because past performance is not a reliable indication of future market returns.

“Knowing that some of these funds do beat the market, you might be tempted to look at which of them have performed well in the last five or 10 years,” says Victor. “The problem is that there’s no persistence in performance. The funds that did brilliantly over the last five or 10 years won’t necessarily be the ones beating the market in the next five or 10 years. The past alone isn’t a reliable marker of future performance.”

So what other markers might there be?

Predicting performance

To pinpoint which active funds are most likely to outplay the market, Victor and his colleagues have built an algorithm that can factor in a whole slew of characteristics besides past performance. Their algorithm integrates 17 different features from things like how many assets a fund has under management to beta – its sensitivity to market fluctuations – onto things like the tenure of its mangers, their length of time managing the fund, turnover, expense ratio, what the fund charges for its services and more. And that’s not all. The algorithm can also analyse synergies or complex relationships between these variables and process the different ways that they interact to make accurate predictions about future outcomes.

“Machine learning gives us the capacity to do a bunch of really smart things that help us determine outcomes with real certainty,” says Victor. “Our algorithm can tune out the noise – all the extraneous stuff – and fixate on those coefficients or factors that really matter to future performance. And by using non-linear predictive models in its analysis, it’s able to integrate how these different factors relate to and affect each other. So things like how market beta or risk exposure interacts with value added – differences between the current market value of a firm and how much capital investors have put in, that show us whether stock is overpriced, underpriced or accurately priced.”

The ability to crunch this wealth of data means that their algorithm can pinpoint those (few) actively managed funds that are most likely to perform better than the index, he says.

To test if all this works in the real world, he and his colleagues put the algorithm through backtesting or sliding window testing – running the models on historical data and then checking the results against actual, real-world performance at routine intervals.

“For instance, we look at data from the year 2000 and run it through our algorithm to predict how different funds would have performed by 2021, say. And we find that it’s accurate. When we use the right coefficients and the right non-linear predictive models – in our algorithm, these are called Random Forest and Gradient Boosting – we can pinpoint just how much alpha an investor would have made from an active fund. And our predictions absolutely map to what happened with these funds, which points to the accuracy of the algorithm.”

A crystal ball for active funds?

Does this mean that Victor and his colleagues have found some kind of crystal ball for investors looking for outsize returns? There are caveats, he says. And it’s principally to do with complexity.

“We’ve proved that it’s theoretically possible to predict outperforming funds, but it’s only possible with machine learning – and this is a complex world. You need access highly sophisticated methods and a real wealth of historical data covering multiple and diverse characteristics, so it’s not easy. Not everyone can do this by any means.”

That being said, there should be scope for funds of funds, pension funds and financial advisors looking to harness the potential here to work with specialists and integrate machine learning with the other tools they are using to select funds. Some savvy funds are already using machine learning to calibrate their investment strategies, but they remain the minority for now.

The main takeaway here, says Victor, is that there is “hope” for active fund management.

“We have proved that it is possible to pinpoint those active funds that do perform. If you use high-dimensional data and the right methods you can find them. This means that investors will be able to channel their capital to the right places – to the good active managers that really do know what they are doing. And this is important for the market too,” says Victor. “We need these experts to identify which stocks are overpriced or underpriced. Among other things, market efficiency depends on it.”