What Has Predicted Funds' Performance? The Thing That Wasn’t Supposed to Work

I was interested in the extent to which funds’ returns before fees predicted how they performed over subsequent before-fee periods. For instance, if a fund did well before fees compared with category peers over a trailing period, did that portend well for it to outperform those rivals over a subsequent period before fees?

To address that, I compiled stock, bond, and allocation funds’ rolling 60-month pre-fee returns over the past two decades.* The first rolling period was from Jan. 1, 2005, through Dec. 31, 2009, the second from Feb. 1, 2005, through Jan. 31, 2010, and so forth all the way through the five years ended Dec. 31, 2024. (This will get less boring in two paragraphs in case you’re wondering.)

I grouped the funds by the percentile ranking of their pre-fee returns versus other funds in their peer group for each five-year period. My groupings followed a normal distribution, such as, best 10% = “top”; next 22.5% = “2nd”; middle 35% = “3rd”; next 22.5% = “4th”; worst 10% = “bottom.” Then I calculated the average forward pre-fee return of each fund in those cohorts and compared them with the average return of all other funds of that type.**

Here’s what it looked like: An almost perfect sort, with the best past performers before fees going on to notch the best future performance before fees, on average, and the opposite for the worst past performers.

Needless to say, this isn’t what we’ve been conditioned to expect. It is past performance after all, and past performance is, well, you know. And so, if you’re like me, you had questions. Questions such as …

Is It a Fluke?

Here’s a look at the year-by-year average excess returns of the different cohorts. (For instance, 2018 averages the average excess returns of all funds for the rolling periods ended in any of the 12 months ended in 2018, and so on for the other years shown.)

What the chart seems to show is that this isn’t some flash-in-the-pan phenomenon. Rather, we’ve seen a similar pattern play out for quite some time.

Is It an Asset-Class Thing?

Maybe past performance has been unusually persistent in, say, stocks but not so much in bonds and multi-asset funds and because there are more stock funds that are distorting the picture? So I reran the same analysis, this time grouping by asset class to control for those differences.

Judging from the above, it doesn’t appear to be an asset-class thing. The pattern was similar across stock, bond, and allocation funds.

Is It a Risk Thing?

Even if it doesn’t appear to be ephemeral or confined to one type of fund, maybe it’s just a risk/reward thing? That is, the best past performers have been taking on more risk, risk has been rewarded, and so really all you’re seeing is different payoffs to risk?

If it’s a risk thing, then that should express itself, resulting in a poorer sort when you use a risk-adjusted measure like a Sharpe ratio. With that in mind, I compiled the funds’ pre-fee risk-adjusted returns (such as, their rolling five-year gross Sharpe ratios) and repeated the exercise described previously.

It doesn’t appear to be a risk thing either. The funds that had the best pre-fee past Sharpe ratios tended to have the best pre-fee forward Sharpe ratios (when compared with their average peer), and vice versa for the laggards.

Is It a Before-Fees Thing?

Lastly, we have to ask whether the picture looked different once we baked fees in. Here it’s a bit apples to oranges—we’re grouping funds based on their pre-fee past returns versus category peers and then measuring their subsequent performance based on net returns.

In summary, even if you ignored fees and chose funds based on their past pre-fee performance, you still tended to do better after fees over subsequent periods. The sorting isn’t quite as emphatic on this basis—this can owe to, say, some very high performers before fees charging a lot and coming up short after fees when we measure subsequent excess net returns. But on the whole, it’s not too different of a picture.

Conclusion

I’ll get into the hows and whys of this pattern in a follow-up to this article, but the evidence suggests pre-fee past performance did a very good job of predicting funds’ subsequent performance, on average, over the past few decades. If you stuck with the best past performers, you tended to be rewarded and, conversely, contrarianism didn’t pay.

Switched On

Here are other things I’m reading, listening to, and watching:

• Amy Arnott on which funds have created the most value for shareholders
• Christine Benz on how to get the most bang for your buck in retirement
• Jason Zweig takes a hard look at an ETF that’s taken a not-small percentage stake in SpaceX
• The Hard Fork podcast on whether DeepSeek really is AI’s “Sputnik moment” (if you want to go deeper, Ben Thompson has a mega-DeepSeek FAQ on Stratechery)
• Richard Ennis sees a “jumble,” not diversification, in institutions' allocations to alternatives
• Work music: Air “Talkie Walkie”
• “You have about you a purposeful savagery”—Furiosa (which I rewatched for, oh I don’t know, like the 10th time)

Feed Me!

I love hearing from you. Have some feedback? An angle for an article? Email me at jeffrey.ptak@morningstar.com. If you’re so inclined, you can also follow me on Twitter/X at @syouth1, and I do some odds-and-ends writing on a Substack called Basis Pointing.

Footnotes

* There were more than 10,300 US open-end funds and ETFs in the study coverage, including dead funds. To be included, each fund had to be classified as an equity, fixed income, or allocation fund, have at least one rolling five-year pre-fee return, and have a category classification at the beginning and end of each rolling five-year period that had a return. I only included the oldest share class of each fund, as the tests were before-fee, and the pre-fee returns of multi-share class funds would be identical and so this avoided needlessly overrepresenting multiple share class funds in the results.

** For instance, once I identified the “top” large-growth funds based on their pre-fee returns versus other large-growth funds over the five years ended March 31, 2017, I tracked their performance versus all other large-growth funds over the five years ended March 31, 2022. (To be more precise: I averaged those “top” funds’ subsequent pre-fee returns and subtracted the average return of all large-growth funds over that period to derive an average excess return for that cohort.) I did this for all cohorts, peer groups, and rolling periods. Then I averaged those averages over the full 20-year span.