The single most cited statistic in manager selection is also one of the least predictive. A fund that landed top-quartile in one vintage is only marginally more likely to repeat than to slide, and yet allocators keep crowding into last cycle's winners as though the ranking were a promise about the future.
The problem is not that past performance is meaningless. It is that the headline quartile bundles together skill, luck, leverage, and timing, and it takes real work to unbundle them. Allocators who skip that work are buying a label, not an edge.
What actually persists
What tends to persist is not the return itself but the capability behind it: a stable team, a repeatable sourcing engine, a genuine operating bench. Those are harder to measure than a quartile rank, which is precisely why the lazy proxy endures.
The allocators who outperform are the ones willing to back a manager whose last fund was merely good but whose process is excellent, over one whose last fund was spectacular for reasons that will not recur.
The discipline to look deeper
Doing this well means attributing returns honestly, meeting the team behind the headline partner, and asking what would have to be true for the next fund to disappoint. It is slower and less comfortable than following the league table. It is also the only version of selection that survives a change in the weather.
Chasing last cycle's top quartile feels safe because it is defensible in a committee meeting. Earning next cycle's returns requires the harder, lonelier work of judging capability rather than reputation.