The Persistence Problem: Does GP Track Record Actually Predict Future Performance?

The intuitive case for backing experienced GPs is strong: if a manager delivered top-quartile returns last time, surely that is informative about what they might deliver next. The academic evidence on this question is more nuanced than the intuition suggests, and the nuance matters for how LPs should approach manager selection.

What the research actually says

Early private equity performance persistence work found meaningful positive autocorrelation in fund returns, particularly at the top end of the distribution. Subsequent studies refined that picture: persistence is strongest in venture, weaker but still detectable in buyouts, and has attenuated as the industry has matured and capital has become more mobile.

Two caveats are worth internalising. First, persistence is estimated in aggregate; the signal for any individual manager is noisier than the cross-sectional average suggests. Second, much of what looks like persistence at the fund level is absorbed once you control for deal-level factors like sector exposure and entry multiple, which argues that part of the effect is structural rather than pure GP skill.

What this implies for allocation

The practical implication is that track record is informative, but it is not sufficient. An allocator looking at two GPs with similar historical IRRs should be doing additional work to separate structural tailwinds from residual manager contribution. The deal-level decomposition that underpins QFT's rating methodology is one way to do this at scale.

Persistence is real, but it is a signal to be calibrated, not a ranking to be followed. The allocators who outperform do so by being honest about how much of a track record is evidence of skill versus evidence of being in the right vintage at the right time.