We Tested Whether Copying Fund Managers Actually Works

312 of the largest 13F books, replayed at the real 45-day reporting delay for six years. 16.7% beat the index. The typical book lagged by 1.9% a year. Here is the full study, the method, and what it means if you follow the smart money.

Every quarter, the internet fills with articles about what famous investors just bought. The implicit promise is always the same: see what the smart money holds, do the same, get smart-money results. We built the dataset to test that promise properly, and the answer deserves more honesty than it usually gets.

We took the 500 largest institutional investment managers in the US, loaded every quarterly holdings filing they made with the SEC since 2020, and replayed each manager's disclosed stock book the only way a real person could: buy the book the day after the filing became public, hold it until the next filing appeared, then rebalance into that one. No hindsight, no early access, no pretending the 45-day legal reporting delay does not exist.

312 managers had enough priced, validated history to score. The results: 52 of them, 16.7%, would have beaten the S&P 500. The median book lagged the index by about 1.9% a year. The middle half of the cohort landed between 3.9% a year behind and 0.5% a year behind. Following the typical large manager, at the delay the law imposes on you, was a reliable way to slightly underperform an index fund.

How the replay works

• For each manager and each quarter, we take the disclosed long US stock positions from that quarter's SEC filing, weighted exactly as the filing weights them.
• The clock starts the first trading day after the filing went public, not at quarter end. A Q1 book filed on May 15 is bought May 16. This is the delay every follower actually faces.
• The book is held unchanged until the manager's next filing becomes public, then rebalanced into the new book. Intra-quarter trades, shorts, options and cash are invisible in these filings, so they are invisible in the replay too.
• Returns use split- and dividend-adjusted prices, and the S&P 500 is measured over the exact same windows. Nobody gets credit for a calendar the other side did not have.
• A manager-quarter only counts if we could price at least 85% of the reported book value. Quarters below that render as excluded, with the reason, rather than being quietly scored on a partial book.

One methodological wrinkle deserves its own paragraph, because it is where lazy studies go wrong. Some managers legally hide positions while accumulating them, through SEC confidential treatment, and reveal the full book months later in an amendment. Berkshire Hathaway does this regularly. When the complete book only became public after the next quarter's filing had already landed, there was never a moment a follower could have held it, so we exclude that quarter and keep holding the previous book, exactly as a real follower would have. Our Berkshire page shows nine such quarters, labeled, with the reason.

What separates the 52 from the 260

The followable minority is not a list of the most famous names. Berkshire Hathaway makes the cut, but barely: following its disclosed book earned 19.1% a year against the index's 18.3% over the same windows, an edge of 0.8% a year. The strongest results in the cohort came from concentrated books that move slowly, because a slow book ages well during the 45 days you are forced to wait, and from managers whose ideas play out over quarters rather than weeks.

The regime pattern is just as instructive. Defensive, value-leaning books beat the index handily through the 2022 bear market and then lagged badly through the 2023 technology rally. Momentum-heavy books did the opposite. A track record that looks smooth from a distance is usually two opposite regimes averaging out, which is one more reason a single summary number should never be the whole story.

And the failures? High-turnover books fail the delay almost by construction: by the time you can see the book, the manager has traded away from it. Index-hugging giants fail differently, tracking the market minus nothing, plus nothing, which after any friction at all means minus something.

The biases we cannot remove, named

• Survivorship: we scored today's 500 largest managers looking backward. Managers who shrank or closed along the way are not in the cohort, which flatters the group's aggregate result.
• Selection on size: the study covers large filers. Small managers might be more followable or less; our data does not say.
• The filing is not the fund: it shows long US stock positions only. A manager whose disclosed book lagged may have thrived on shorts, options or international positions you cannot see. The study measures the follower's achievable result, not the manager's skill.
• Six years is six years: 2020 through 2026 contained a crash, a mania, a bear market and a rally, which is more variety than most study windows get, and still far from everything markets can do.

What this means if you follow the smart money

First, stop treating a famous holder as a reason to buy. The base rate says the typical big book, followed at the real delay, loses to an index fund. If a headline says a legend bought something, the honest question is not who bought, but whether that manager's book has ever been worth following at the lag, and that is a measurable question.

Second, when a book is followable, the numbers are modest. The best sustained edges in our cohort ran a few points a year, with real drawdowns along the way. Anyone promising more than that from copying filings is selling the costume, not the number.

Third, the interesting signal is often agreement, not admiration. When a company's own officers are buying in the open market during the same quarter that large institutions added the name, two independent groups with different information both moved the same direction. That co-occurrence is what our conviction screen watches for, and its own base rates are published with the same honesty.

Every number in this post is recomputed from filings and prices we store, and every manager page shows its own quarter-by-quarter replay, including the quarters we refused to score and why. That is the standard this kind of claim should meet. Historical statistics, not predictions. Insight, not advice.