How to Read a 13F Filing for Institutional Concentration

When Berkshire Hathaway filed its 13F for the quarter ended December 31, 2023, one line jumped out: Apple accounted for 49.54% of the company's entire disclosed equity portfolio, per the filing itself. Financial media ran the number as proof of conviction — Buffett's biggest bet, still growing. The next several 13Fs told a different story. By the quarter ended September 30, 2024, Berkshire's Apple share count sat roughly 67.2% below where it stood a year earlier, according to contemporaneous reporting on the filings.

That's the trap a 13F sets for anyone reading it casually. The document is real and the numbers are accurate, and it still misleads more often than it informs — not because it's wrong, but because it's a snapshot of a decision made weeks before you're allowed to see it, dressed up to look like a live signal.

I want to walk through what a 13F actually discloses, what it structurally can't tell you, and — this is the part that matters for your own portfolio — how much weight a filing like Berkshire's Apple position deserves when you're deciding what to do with your own concentration.

What a 13F Filing Actually Discloses

Form 13F is a quarterly disclosure the SEC requires of any institutional investment manager exercising discretion over $100 million or more in what the rule calls Section 13(f) securities — U.S.-exchange-listed stocks, ETFs, and certain convertible instruments and options. Cross that threshold even briefly during a calendar year, and the manager has to keep filing through the third quarter of the following year, whether or not assets stay above the line.

The filing itself is a security-level table: issuer name, share class, CUSIP, number of shares, and market value, aggregated across every client account the manager oversees. It's due within 45 calendar days of the quarter's end — Q4 numbers land in mid-February, Q1 numbers in mid-May — and it's public the moment it hits EDGAR. Anyone can pull Berkshire's, or any large hedge fund's 13F history, directly from the SEC's site at no cost.

That accessibility is exactly why 13F season has become its own recurring event on financial media and on aggregator sites that repackage the raw EDGAR tables into leaderboards of what "smart money" bought and sold. None of those sites are doing anything wrong — the data is public and the aggregation is genuinely useful for scanning quickly. But the format itself nudges readers toward treating a quarterly regulatory filing like a stock tip, because a ranked list of buys and sells reads like one whether or not the underlying document was ever meant to function that way.

What the Filing Leaves Out

The gaps matter more than the disclosure, and they're not edge cases. A 13F shows long positions only. A manager can run an equally large or larger short position against the same stock, and none of it shows up — the filing would show a bullish-looking long position sitting on top of a bet that's actually market-neutral or outright bearish. Managers aren't required to report short positions on Form 13F at all.

Options are similarly incomplete. A manager may report put or call options they hold on a security covered by the SEC's official 13(f) list, but options they've written — sold, rather than bought — don't appear, which means short-equity-like exposure can hide inside an options book the filing never surfaces. Add in that non-U.S. securities, most fixed income, cash, and anything outside the official 13(f) list are excluded entirely, and a 13F stops looking like "a fund's portfolio" and starts looking like a specific, incomplete slice of it.

Then there's the lag, which investors underweight the most. A 13F for the quarter ended December 31 isn't public until mid-February at the earliest. A manager can exit a position entirely on January 5, and their 13F still shows the full stake six weeks later — because the form reports where the portfolio stood on the last day of the quarter, not where it stands the day you're reading it.

There's a subtler gap too: a single 13F blends together every strategy a manager runs. A multi-strategy fund with a dozen semi-autonomous portfolio managers files one combined table, and a large position can just as easily be the sum of several unrelated, moderate-sized bets from different desks as it can be one person's high-conviction call. Berkshire is unusually easy to read precisely because it's run as a small number of concentrated, long-held decisions rather than dozens of internal books netted together — most 13F filers aren't that legible, and the filing gives you no way to tell the difference from the outside.

A Framework for Weighing 13F Data in Your Own Position Sizing

None of that makes 13F data useless. It makes it a single, stale, partial input — and the real question is what job you're asking it to do before you let it move your own portfolio.

• Corroboration over conviction. One manager concentrating in a name tells you about that manager's judgment, not about the business. A cluster of unrelated managers building positions in the same name over consecutive quarters is a weaker but more interesting signal — closer to independent verification than any single filing.
• Position size relative to the manager, not relative to the market. A stock at 5% of a $2 billion fund and the same stock at 5% of Berkshire's equity book mean very different things about how much capital is actually chasing the idea. Read the percentage in context, not in isolation.
• Treat the filing date, not the report date, as the operative fact. Anything you read in mid-February reflects decisions made as of December 31 at the latest, possibly reversed since. Don't anchor a position-sizing decision today to a filing that's already been overtaken by two more months of trading you can't see.
• Ask what your own conviction is, independent of the filing. A 13F can add context to a thesis you already hold. It's a much weaker foundation for building a new position from scratch — the [concentration your own portfolio can absorb](/blog/concentration-vs-diversification) has to come from your own risk tolerance and time horizon, not someone else's disclosed one.
• Check whether the manager is even worth reading in the first place. A fund that quietly hugs an index while charging active fees produces a 13F that looks like a hundred small, low-conviction decisions rather than a handful of real ones — following it tells you almost nothing, no matter how large the fund is. A concentrated, long-holding-period manager with a public, consistent process is a far more informative filer than size alone would suggest.

The Berkshire Example: Reading a Concentration Number Correctly

Berkshire's Apple position is as good a case study as exists, precisely because the raw numbers are dramatic enough to tempt a bad read. At 49.54% of the disclosed equity portfolio as of December 31, 2023, Apple wasn't just Berkshire's largest position — it was nearly half of everything the filing showed, an extraordinary concentration by any conventional diversification standard.

Read only that number, and the obvious conclusion is maximum conviction, keep buying. But the next several 13Fs said otherwise. Berkshire trimmed the stake by roughly 13% in the first quarter of 2024, nearly halved what remained in the second quarter, and kept selling into the third — leaving the share count down about 67.2% from a year earlier, even as Buffett and Greg Abel repeatedly said publicly that Apple remained a business they liked.

A business the manager still likes, sold down aggressively anyway. That combination is exactly the nuance a single 13F snapshot erases. Commentary at the time attributed the selling to tax and portfolio-concentration considerations rather than a changed view of Apple's business quality — I'm not fully convinced that's the whole story, and a 13F table alone can't confirm or refute a manager's stated motive either way. The number told you what changed. It didn't tell you why, and reading a percentage without the why is how a lot of "the smart money is buying/selling X" narratives get built on incomplete information.

The lag compounds the problem. The Q4 2023 filing showing that 49.54% figure wasn't public until mid-February 2024 — by which point Berkshire had already begun the first-quarter trimming that wouldn't itself be disclosed until mid-May. Anyone who read the headline concentration number in February and sized a position around "Buffett still loves Apple" was, without knowing it, already a quarter behind the manager's own decision-making. That's not a hypothetical risk unique to Berkshire. It's the structural condition of every 13F, for every filer, every quarter.

The lesson isn't that 13F data is worthless. It's that a concentration figure is a fact about a filing, not a verdict about a business. And that gap is exactly where the framework above earns its keep.

Key Takeaways

• Form 13F discloses long U.S. equity-type positions for any manager overseeing $100 million or more, filed within 45 days of quarter-end — a real, useful, but structurally partial picture.
• The filing excludes short positions, written options, non-U.S. and non-13(f) securities, and anything that's changed since the quarter-end snapshot — which can be six weeks stale by the time you read it.
• Weigh 13F data by corroboration across managers and by position size relative to the filer, not by treating any single filing as a signal to replicate in your own portfolio.
• Berkshire's Apple position shows the pattern clearly: a dramatic concentration number and a subsequent reversal can both be true at once, and the filing alone never explains which matters more for your own decision.