"Concentrate in your best ideas." "Conviction is edge." Every stock-picking philosophy — and every 13F-following pitch — leans on the same premise: the manager who bets big on a handful of names knows something, and you should ride it. We have the clone-performance data to test the premise directly, so we did.
The method
For all 58 ranked funds with both a 3-year clone record and a disclosed book, we measured two numbers:
- Concentration: the share of the portfolio in its five largest positions.
- Clone alpha: the 3-year return of mechanically replicating the fund's top-20 holdings, minus the S&P 500, rebalanced at filing dates with no lookahead.
Then we asked the obvious question: does more concentration predict more alpha?
The result: the correlation is negative
Across the 58 funds, the correlation between concentration and clone alpha is −0.27. Not zero — negative. Sorted into three buckets, the pattern is stark:
| Portfolio | Funds | Avg clone alpha | Beat the S&P |
|---|---|---|---|
| Diversified (top-5 < 40%) | 21 | +6.1 pt/yr | 76% |
| Middle (40–60%) | 12 | −3.0 pt/yr | 33% |
| Concentrated (top-5 ≥ 60%) | 25 | −3.8 pt/yr | 32% |
The most concentrated funds' clones didn't just underperform the diversified ones — they lost to a plain index fund more than two times out of three. The "conviction is edge" premise, tested on three years of real clones, is backwards.
But read the reason before you draw the wrong lesson
This is not "diversify and you'll win." Two things drive the result, and neither is a free lunch:
- Concentration is variance, not skill. The concentrated bucket contains the best clone in our whole sample — CAS Investment Partners at +39 pt/yr, essentially a single enormous Carvana bet that paid — sitting right next to Abdiel and Worldly Partners at −21 pt/yr each. A five-stock book is a five-stock outcome: when it's right it's spectacular, and when it's wrong there's nothing to cushion it. The average is negative because the blow-ups outnumber the moonshots. High conviction widens the distribution; it doesn't shift the mean up.
- Cloning flatters the diversified funds — for a reason that won't repeat. The diversified bucket is dominated by quant giants: Point72, Renaissance, Millennium, Citadel — funds with thousands of positions. Cloning their
top-20 disclosed longs isn't cloning their strategy (market-neutral, high-turnover, levered); it's just buying the twenty mega-caps they happened to hold long, which rode the same megacap rally the index did — plus a little. Their real edge lives in the 3,000 positions and the shorts a 13F never shows.
So the honest takeaway isn't about concentration versus diversification at all. It's this: a concentrated 13F clone is a leveraged bet on one manager being right about a few names, and the base rate for that bet is worse than the index. The tail is real — CAS is real — but you are buying a lottery distribution, not a conviction premium.
What we'd actually do with it
- Treat a concentrated manager's clone as one high-variance position, sized accordingly — not as "the smart money's best idea, therefore safe."
- Don't read a diversified quant fund's top-20 as a strategy you can copy. The part you can see is the part that looks like the index.
- Weigh the clone drawdowns as heavily as the returns. Every concentrated winner in our set drew down 30–50% along the way.
Conviction makes a great story. Across 58 funds and three years, it did not make alpha — it made variance. Knowing the difference is worth more than any single pick. Run the numbers yourself in the clone backtester, and see why cloning even the legends only matched the market.
Methodology: clone NAV with no lookahead; concentration measured from each fund's top holdings. 58 funds with ≥3 years of history and a disclosed book. Survivorship flatters the sample. Past performance does not predict future returns. Not investment advice.