Why Copying 13F Trades Usually Loses Money: the Strategies, Backtested on This Site's Own Data

The naive strategy, stated precisely

Each quarter, buy what institutional managers disclosed as brand-new positions in the quarter that just ended, value-weighted, and hold until the next filing. The rule carries a built-in one-quarter signal lag: you act only on data at least 45 days old. That is exactly the rule on our strategy backtest page, and every number below comes from the same engine that powers that page — same price book, same lag, same quarter-skipping logic.

The results on this site's data

The tracked universe is 98 managers; quarterly prices reach back to 2020, so each strategy is compared with SPY bought and held over the identical segments. Total return includes the running quarter through the latest close (snapshot date at the top of this page).

Read the second column as the honest benchmark: the more “newsy” the signal — fresh new positions, famous names — the further it landed behind a plain index fund. The two “boring” approaches did relatively better: requiring a second consecutive accumulation quarter, and cloning the oldest top names of a patient book.

Why the gap: four structural reasons

• The 45-day lag is a positioning tax. By the time a NEW position is public, the re-rating it was betting on has often already happened. The two worst stretches for the follow-everything rule were 2021-Q4 (-12.6%) and 2022-Q1 (-12.2%) — quarters when crowded institutional favorites unwound after being fully public for months.
• Position size is not conviction. Value-weighting new positions piles into whatever was largest that quarter — frequently momentum at its peak. A 13F tells you what someone owned, not what they would pay today; see what 13F cannot show.
• You only see the long, visible book. The Tiger Global top-10 clone fell 63.5% in 2022 with a 55.2% max drawdown: its hedged book was invisible, and the visible half was the wrong half to own that year. Copying the long-only slice of a hedged book is not copying the fund.
• Passive giants flood the new-positions list. A new BlackRock or Vanguard line is index arithmetic, not analysis — see smart money vs passive money. Even restricting to star and hedge funds did not save rule three: famous names were exactly the crowded trades of the 2021–2022 growth unwind.

What did work, honestly

• Consecutive accumulation beat fresh news. Requiring a second consecutive ADD/NEW quarter — the streak rule — returned +76.8% with a 17.4% max drawdown, *lower* than SPY's 24.9%, and a Sharpe ratio (0.91) within rounding of SPY's (0.88). The patience filter removes the one-quarter punt.
• Copy the process, not the quarter. The Berkshire top-10 clone beat SPY over the same window (+95.2% vs +81.2%) — but its top names are positions held for years, not disclosures from last quarter. What is copyable is the holding discipline, visible on the Berkshire fund page, not the latest filing's news.
• Screen the copyee before following. Combine turnover (the next chapter), top-10 concentration and active share: a low-turnover fund's NEW position is an event; a high-turnover fund's NEW position is just Tuesday.
• Use rankings built for the lag — consecutive accumulation, star consensus and smart-money signals normalize for fund count and manager type instead of rewarding whoever filed the loudest headline.

Reproduce any row on the backtesting tools pages: value-weighted portfolios, one-quarter signal lag, and quarters with no priceable positions skipped identically. Figures are a snapshot as of the date at the top of this page and will drift as daily prices update. None of this is investment advice — it is a measurement of how the most common copying instincts perform once the 45-day lag is priced in.