In addition, because fewer funds outperform than would be randomly expected, it’s hard to know if the very small percentage of outperformers were able to produce that result based on skill or luck.

For example, Philipp Meyer-Brauns of Dimensional Fund Advisors, author of the August 2016 research paper “Mutual Fund Performance Through a Five-Factor Lens,” found that “there is strong evidence that the vast majority of active managers are unable to produce excess returns that cover their costs.” His study examined 3,870 active funds over the 32-year period from 1984 to 2015. Meyer-Brauns added that “funds do about as well as would be expected from extremely lucky funds in a zero-alpha world. This means that ex-ante, investors could not have expected any outperformance from these top performers.”

The great majority of studies on mutual fund performance have used fund-level data, examining the performance of funds rather than the performance of fund managers. And over any sample period, more than one manager may have been responsible for the management of that fund, particularly as the sample period lengthens. With this in mind, Andrew Clare, author of a September 2016 study, The Performance of Long-Serving Fund Managers explored the performance of tenured fund managers (those with at least 10 years as the sole manager of a fund) to see whether experience favorably impacted performance. It certainly seems like a reasonable hypothesis.

Of course, by restricting the pool to managers with such long experience, we have to be aware of the issue of survivorship bias – managers with poor track records aren’t likely to last 10 years. Clare’s dataset consisted of 357 managers with at least 10 years running a fund as of December 2014.

He found that for long-serving mutual fund managers, the net-of-fee return in excess of the benchmark averaged just less than 0.5 percent per year. And while the benchmark-adjusted performance figures look impressive, as noted earlier, there is an element of survivorship bias here.

Importantly, Clare also found that the excess return was negative over the last four years of the period. In other words, the collective ability of long-tenured managers to outperform their benchmarks waned over the 10-year period. Either that or their luck ran out. When he tested for persistence in performance, Clare found that the winners in one year were unlikely to be winners in the next years, as in six of the nine other years’ excess returns were negative. Within the group of long-serving mutual fund managers he studied, there was no evidence of persistence of performance.

The bottom line is that even with considerable survivorship bias in how the test was set up, deteriorating performance over the 10-year period and the lack of persistence in performance among long-serving fund managers demonstrates why Charles Ellis called active management a loser’s game. It’s not that you cannot win. It’s just that the odds of doing so are so poor it’s not prudent to try. And despite what most investors believe, it’s clear that even long and successful track records don’t provide much, if any, value for predicting future performance.