A fund advertises average annual returns of 11 percent since 1998. The number is honest. The scoreboard it sits on has been quietly removing the losers for a quarter century, and the fund in the ad is the survivor of a field that started with several hundred entries.
The exits nobody puts in the brochure
Run a database of mutual funds forward for twenty years and watch what happens to the bad ones. They do not sit at the bottom of the rankings forever. They close, or they merge into a stronger fund and vanish from the record entirely. A share class that underperforms gets folded into one that outperforms. The record of the merged class ends on the day of the merger.
What remains is a population that was selected for having survived. The average return of that population rises a little every time a bad fund leaves, without a single portfolio manager making a better trade. Mutual fund families have consolidated aggressively since the 2000s, and every wave of mergers does the same cleanup job on the numbers.
This is survivorship bias, and it is old news in academic circles. A 1992 study by Stephen Brown, William Goetzmann, Roger Ibbotson, and Stephen Ross looked at the funds that had died and found they trailed the survivors by a margin wide enough to make a mediocre manager look talented. The dead funds were not in the average. That was the trick.
The scale of the cleanup is worth seeing once. Scorecards with twenty-year windows routinely open with a couple of thousand funds in a category and close with barely half of them still filing reports. The other half left no corpse in the data, just a quiet gap where a bad decade used to be.
The Steady Eddie tell
Some of the worst damage comes from the returns that look calmest. Bernie Madoff’s fund reported gains of around 10 to 12 percent a year with almost no losing months, and that texture was the tell. Real markets leave scars. A record that never has a bad month has either found something extraordinary or is drawing the line in pencil.
The trader who sees a beautiful track record should ask three questions. How many funds were in this category when the record started. What happened to the ones that left. And is the average shown here a survivor average or a whole-cohort average. S&P Dow Jones Indices publishes a scorecard that keeps both numbers side by side, and the gap between them is the price of the cleanup.
Buying the graveyard’s returns
One portfolio manager, Ted Aronson, ran a public paper portfolio for years and published every trade. Most managers do not, so the public record keeps the winners and the memory. The practical move is to grade a manager against the cohort: every fund launched the same year, alive or dead. If the pitch leans on the phrase “since inception,” ask when inception was. A 1998 start date and a 2021 start date describe different markets.
The bias works one level up too. A fund of funds, a model portfolio, a newsletter’s recommended list: each one inherits the same housekeeping, the quiet removal of whatever disappointed. The scoreboards of finance are maintained by the people whose funds are still playing.