Decoding Berkshire's 13F: How to Read Buffett's Quarterly Playbook

Why every Berkshire 13F is the most-watched filing in markets

Most 13F filings disappear into the SEC database the day they are filed. Berkshire Hathaway's does not. Within minutes of a typical Berkshire 13F hitting EDGAR, business television cuts to it live, financial Twitter spins up, sell-side analysts put out same-day notes, and at least three retail brokerage apps queue push notifications. There is a reason for the attention — and a reason most of it is misplaced.

The reason for the attention is performance. Frazzini, Kabiller, and Pedersen (2018) studied Berkshire's record from 1976 to 2017 and found that Buffett's gross Sharpe ratio over four decades was approximately 0.79, roughly double the market's. Their decomposition showed the alpha was real and persistent, attributable to a tilt toward "cheap, safe, quality" stocks combined with patient, leveraged-via-insurance-float exposure. Whatever you think of Buffett's style, it is one of the few publicly documented track records that the academic literature credits as genuine alpha rather than luck.

The reason most attention is misplaced is timing. The 13F is reported with a 45-day delay after quarter end. By the time you see the May 15 filing covering March 31 holdings, the price action you would react to is six weeks gone. The "buy what Buffett bought" trade is structurally late.

That does not mean the filings are useless — it means using them well requires understanding what they actually are.

What a 13F includes (and what it doesn't)

Form 13F is required of any institutional investment manager with more than $100 million in qualifying US equity assets. It must be filed within 45 days of each calendar quarter end. The filing includes:

• Long positions in US-listed equities and ETFs, including ADRs.
• Long calls and warrants on US-listed equities (denoted as such).
• Convertible bonds with embedded equity options.
• Position counts and dollar values as of quarter end.

It explicitly does not include:

• Short positions. Berkshire does not short much, but that is a structural limit of the form.
• Foreign-listed equities. Berkshire's stakes in BYD (Hong Kong) and the five Japanese trading houses are not in the 13F.
• Wholly-owned subsidiaries. GEICO, BNSF Railway, See's Candies, Dairy Queen — the operating businesses are invisible in the 13F. Berkshire's annual report covers them; the 13F does not.
• Cash, treasuries, and corporate debt holdings. Berkshire's $150B+ treasury position is one of the most consequential lines on the balance sheet and never appears in the 13F.
• Derivatives like equity-index put options Berkshire wrote in the 2000s.

Reading the 13F as the portfolio is a category error. It is one ledger among several. For the publicly-traded equity slice it is authoritative, and that slice is what most retail trackers care about, so the limitation is mostly cosmetic. But it matters when people draw conclusions like "Berkshire is entirely in financials" without acknowledging that BNSF and the energy businesses dwarf any single 13F line.

The concentrated five

Buffett has spent his career making concentrated bets. The 13F reflects this: typically the top five positions account for 75–85% of the equity book. Through 2025 those positions have rotated — American Express has been a near-permanent fixture for three decades, Coca-Cola similarly long, Bank of America was a 2011-vintage warrant exercise, Apple a 2016–2018 build-up, and Chevron a more recent addition. The names change; the structure does not.

This concentration shapes how to read the filing. A 5% trim in Apple is a $5B+ liquidity event for that name. A new 1% position is a thesis-tier investment, not a flier. If a name shows up that is less than 0.5% of the book, it is almost certainly a Todd Combs or Ted Weschler trade — the two portfolio managers who run the smaller "research mandate" inside Berkshire — rather than a Buffett decision. The filing does not label which manager initiated which trade, but the size threshold is well-known among Berkshire watchers.

Reading position changes

Every 13F is best read as a delta to the previous one. The four columns to track:

• New positions. A name that did not appear last quarter and now does. Initial sizing matters — small new positions are exploratory, large ones are conviction.
• Position adds. A name with a higher share count this quarter than last. Common during accumulation phases.
• Position trims. A name with a lower share count, but still present. Often the early stage of an exit, but not always — sometimes pure rebalancing.
• Exits. A name that disappears entirely. The hardest signal to read because the timing of the final sale within the quarter is unknown — you only know it was zero by quarter end.

The Apple example illustrates the timing issue. Berkshire began accumulating Apple in early 2016. By the time the 13F revealed the position, Apple had already rallied. By the time most retail buyers acted on the disclosure, Apple had rallied further. Yet the position kept growing for two more years before stabilizing — meaning the disclosure was a useful signal not for the original entry, but as confirmation that Berkshire continued to add. The repeated quarterly add was the actionable read, not the first appearance.

The post-13F drift

The academic question is whether buying after a 13F discloses a Berkshire purchase actually works. Cohen, Polk, and Silli (2010) studied a related question — whether mutual fund managers' "best ideas" (their largest active bets) outperform — and found significant alpha in the top-conviction bucket, which would predict that following Berkshire's largest bets should help. More directly, several practitioner studies have shown a measurable post-13F drift in Berkshire-disclosed names: roughly 1–3% in the week after a major add appears, with no clear reversal.

This drift is real but small relative to the latency. The trade is "buy on disclosure day, sell two to four weeks later" — a low-volatility, low-edge swing trade rather than a long-term thesis. Most retail buyers who follow Berkshire trades do not execute it that way; they buy on disclosure and hold for years, which is closer to outsourcing portfolio construction to Buffett than capturing the drift.

Confidential treatment requests

Buffett has occasionally filed requests with the SEC to keep specific positions confidential while Berkshire is still building them. The mechanism: file the 13F as required, but ask the SEC to redact selected lines for a defined window (typically one year, sometimes longer). This is rare and reserved for positions where premature disclosure would meaningfully impair Berkshire's ability to accumulate at favorable prices. IBM in the late 2000s and Chubb (Marsh & McLennan) at various times have been cited as examples.

For retail trackers, this means there is sometimes a Berkshire position you cannot see for several quarters. When it finally surfaces, it is generally already large — the gap between thesis formation and disclosure was the whole point of the request.

Side-by-side: 13F vs the full Berkshire portfolio

Practical reading checklist

1. Compare to the previous quarter, not to your priors. The interesting data is in the delta.
2. Filter by position size threshold. Anything under ~$1B is probably Combs or Weschler, not a Buffett decision — treat accordingly.
3. Watch for repeated adds. A new position is interesting; the same position growing for three consecutive quarters is conviction.
4. Note exits, but discount them. The timing within the quarter is unknown, and tax considerations or rebalancing can drive small full exits without a thesis change.
5. Cross-reference the annual letter. Buffett's commentary in the May annual report often retroactively explains 13F changes from the prior twelve months.
6. Ignore index ETFs. Berkshire occasionally holds SPY or IVV as cash management. They are not signals.

Bottom line

Berkshire's 13F is the highest-resolution view we get into one of the few documented genuine-alpha track records in modern finance. It is also a partial view, lagged by 45 days, and stripped of context (subsidiary operating businesses, derivatives, foreign holdings, debt) that any honest portfolio analysis would require. Read carefully, it is one of the most informative filings in markets. Read carelessly, it produces an entire industry of "buy what Buffett bought" content that systematically lags the trade and misses the structure. The difference is whether you treat it as a signal among signals or as a portfolio to copy.