Can You Actually Pick a Winning Fund to Clone in Advance? Partly — But the Edge Evaporates.

When we ranked 59 funds by clone performance, we said the winners were "impossible to pick in advance." A fair reader pushed back: is that actually true, or does skill persist? It's the only question that matters for anyone choosing a manager to follow, so we tested it head-on.

The method

We took every fund with a clone record reaching back to early 2023 (60 of them) and split the timeline into two consecutive ~21-month periods:

• Period 1: Jan 2023 → Oct 2024
• Period 2: Oct 2024 → Jul 2026

For each fund we measured clone alpha (vs the S&P 500) in each period separately, then asked: does doing well in Period 1 tell you anything about Period 2?

The result: persistence is real — more than a coin flip

It does. The correlation between a fund's Period-1 and Period-2 clone alpha is +0.32 — weak-to-moderate, but clearly positive. And ranked into halves:

70% of the funds in the top half in Period 1 were still in the top half in Period 2 — well above the 50% you'd get from a coin flip.

So our earlier "impossible to pick in advance" was too strong. Past clone performance is a real, if noisy, signal. The good funds tend to keep being good. That alone separates this from the "past performance means nothing" boilerplate.

The catch: the edge reverts hard

Here's what the persistence number hides. Look at the eight funds that led Period 1 and where they landed in Period 2:

Every one of them stayed positive — that's the persistence. But four of the five saw their edge collapse by 60–80%. A clone that beat the market by 91 points reverted to 20. The rank held; the magnitude did not.

This is mean reversion doing exactly what it does to concentrated, high-variance books (which, as we showed separately, is what most of these top funds run). An extraordinary year pulls the multi-year number up; the next stretch is merely very good instead of once-in-a-decade.

What this actually means for cloning

1. Do use past clone alpha as a filter — it works. A top-half fund is meaningfully more likely than not to stay top-half. Ignoring the track record entirely leaves signal on the table.
2. Never extrapolate the number. The fund that returned +40 pt/yr will very

likely still beat the market — and very likely by a fraction of +40. Size the bet to the reverted expectation, not the headline.

1. Two periods of a growth bull is not proof. Both windows here favored the concentrated growth books that dominate our leaderboard; a regime change could compress the persistence itself. Weak signal, not a law.

The honest synthesis across this whole series: cloning the average manager matches the index, the best managers genuinely persist above it, and the size of their edge is the one thing you should assume will shrink. Skill is real and partly predictable — it's the magnitude that lies to you. Test any of it in the clone backtester.

Methodology: measured over the daily clone NAV series, no lookahead. 60 funds with records spanning both periods; survivorship flatters the sample and both windows were growth-favorable. Past performance does not predict future returns. Not investment advice.