How Warren Buffett's Style Actually Shifted — Tracking 13F with DIVA

After Warren Buffett's retirement, we ran his investing style through DIVA Quantizer. Does the cliché "Buffett the value investor" hold up in numbers? Let's track it directly by splitting Berkshire's 13F portfolio into three eras (2001, 2011, 2024).

Any US institution managing over $100 million in equities must disclose its US-listed stock holdings on Form 13F within 45 days of quarter-end. That means the stocks held by Warren Buffett's Berkshire Hathaway become public four times a year — a rare chance for an individual investor to peer into a legend's hand on a regular basis.

Where and how to read it

It's not hard to do yourself. On SEC EDGAR (the filing search at sec.gov), search "Berkshire Hathaway," pick filing type 13F-HR, and open the information table — issuer, shares and market value, line by line (jump to Berkshire's 13F list). If raw filings feel daunting, aggregators like dataroma or whalewisdom reorganize the same 13F with weights and quarter-over-quarter changes. Every holding and weight in this article comes from those 13Fs (and Berkshire's annual reports for the relevant year).

But 13F has clear limits: ① it shows only US-listed stocks (no cash, bonds, private or foreign-listed positions); ② short positions and many derivatives are invisible; ③ it's a quarter-end snapshot with up to a 45-day lag, so things may have already changed by the time you see it. That makes 13F less a "copy-trade" tool than a way to read a manager's style — which is exactly what this article does.

Buffett's holdings have turned over slowly across decades. To see that drift we picked three points about six years (72 months) apart: the classic value/quality years (≈2001), the era of broad staples-and-banks diversification that also added IBM (≈2011), and today's Apple concentration with energy mixed in (≈2024). For each, we bundled the top 13F holdings by market-value weight into one portfolio and fed it into DIVA using that era's prices (month-end adjusted close).

One processing rule to be transparent about: we excluded names without a continuous price history and renormalized weights to 100% within the rest (the same approach as our National Pension piece). So Era I drops Gillette (absorbed by P&G in 2005), the Washington Post (reorganized into Graham Holdings) and Moody's (IPO'd Sept 2000, too short a window); Era II drops foreign listings (Munich Re, POSCO, Sanofi, Tesco) and the split-up Kraft. As a result Era I is highly concentrated in just four names (which is faithful to Buffett then), while Era II is the most spread out, at nine.

The Fama–French five-factor model decomposes a portfolio's return into exposure (beta) to five characteristics: market (MKT), size (SMB, small caps), value (HML, cheap stocks), profitability (RMW, quality), and investment (CMA, conservative). A bigger beta means a stronger "tilt" toward that trait. Run the same lens over all three eras and Buffett's drift shows up on a single page.

Slice the three eras through the Fama–French five factors and the simple label “value investor” turns out to be pretty crude. Buffett's tilt was never fixed in one direction; in each era he expressed the same principle — buy good businesses cheaply — in a different way.

Value beta: strongest in the middle era, not the “classic” one

The most telling number is value beta (HML): +0.16 in 2001, +0.66 in 2011, +0.09 in 2024. Intuitively the “classic” 2001 book should show the strongest value tilt — but it didn't. The 2001 portfolio was closer to a quality book heavily concentrated in Coca-Cola and American Express: less an academic “basket of cheap stocks” than a conviction bet on a few firms with powerful brands and durable long-term advantages.

The value tilt was strongest, by contrast, in 2011. That portfolio held Coca-Cola, Wells Fargo, American Express, P&G and IBM, and its value beta of +0.66 was statistically significant. In other words, the era where Buffett's “value investor” image shows up most clearly in the statistics was not the early classic book but the middle era, broadly spread across financials, staples and IBM.

This is the key point. Buffett's value investing is not simply buying stocks with a low P/B or P/E. In 2001 he concentrated on brands and moats; in 2011 he held a broad basket of cheap, quality names; in 2024 he concentrated on high-return mega-cap quality led by Apple. Inside the very same phrase “value investor,” the actual character of the portfolio changed completely.

Market beta: defensive only in the middle

Market beta (MKT) carries the same message: 1.15 in 2001, 0.83 in 2011, 1.17 in 2024. The only defensive era was 2011 — high value exposure, low market sensitivity. The classic picture of “Buffett-style” investing — buying good companies cheaply and swinging less than the market — was sharpest then.

The 2024 portfolio, by contrast, is hard to call a traditional value book any longer. As Apple's weight grew, market beta climbed back above 1 and value beta all but vanished. This doesn't mean Buffett abandoned value investing. It's closer to saying his definition of “value” shifted from accounting cheapness toward qualitative edges — capital efficiency, brand dominance, ecosystem control.

In short, Buffett's style moved from “buying value stocks” to “concentrating in companies that compound value over time.” The 2001 Buffett was a concentrated investor betting big on a few quality names; the 2011 Buffett was a value/defensive investor in the traditional sense; and the 2024 Buffett is closer to a quality-concentration investor betting hard on one giant compounding machine — Apple.

The Markowitz lens shows yet another shift in Buffett's portfolio. The point is simple: over time the book didn't spread out into wider diversification — the quality of its concentration improved.

Annual volatility was 24.4% in 2001, 18.8% in 2011 and 21.7% in 2024. 2011 was the steadiest; 2001 and 2024 carried relatively more risk. Yet the Sharpe ratio improved, 0.26 → 0.49 → 0.91. 2024 looks good not because volatility fell but because it is a structure that earned a bigger reward while accepting high volatility.

Here is where Buffett's change shows. The 2001 concentration was somewhat rough. His conviction in Coca-Cola and American Express was strong, but from a portfolio standpoint there were too few names, leaving large single-company risk. So the diversification effect was a low 20.2%.

2011 was the most balanced era. As the number of names grew, the diversification effect rose to 28.8% and market beta fell. This is when Buffett's book came closest to a “textbook stable value portfolio”: high value exposure, low market sensitivity, the best diversification. If you want to describe Buffett as a defensive value investor, this era is the most convincing evidence.

2024, though, has a different texture. Diversification is 26.9%, lower than 2011, and Apple's weight makes the concentration risk obvious. Yet the Sharpe ratio is the highest, at 0.91. This does not mean the 2024 portfolio became safer. More precisely, risk wasn't removed — it was deliberately piled into the place where it was taken on.

Plotting today's Era III portfolio directly on the efficient frontier makes plain what the Apple concentration did to its risk and efficiency.

This is the heart of today's Buffett portfolio. He is not a diversifier in the traditional sense. Rather than spreading out to reduce risks he doesn't understand, he manages risk by loading heavily into companies he believes he understands. The 2024 Apple concentration is the latest version of that method. On the numbers, market beta is high and value beta is low — but from Buffett's vantage, Apple was likely not a mere growth stock but a long-term compounding asset with enormous cash flows and ecosystem dominance.

So the conclusion from this analysis is clear. Buffett did not become a more conservative investor over time. If anything, he became a more selective concentration investor. In 2001 he concentrated on brand moats; in 2011 he held traditional value and defensiveness broadly; in 2024 he concentrated on capital-efficient mega-cap quality.

Buffett's true consistency is not in any single factor. Value beta, market beta and the diversification effect all changed era to era. Only one thing stayed constant — the principle of allocating capital, for the long term, to a handful of quality companies he judges he understands. His style, in other words, is better read not as “a preference for value stocks” but as “long-term concentration in high-conviction quality companies.”

Sources: SEC EDGAR 13F-HR (Berkshire Hathaway) and Berkshire annual reports / shareholder letters for the relevant years. Prices: yfinance month-end adjusted close — Era I 1997–2002, Era II 2008–2013, Era III 2019–2024 (72 months each). Factors: Ken French data library FF5 (North America). Risk-free rate: current US 13-week T-bill (3.69% annual), held common across models and eras.

① 13F is for reading style, not copy-trading

A 13F is a lagging document, public up to 45 days late. Buy the same stock after the filing and your price already differs from Buffett's — and he may have sold in the meantime. What a 13F does show well is which risks a legend is tilted toward right now. As we saw, that tilt clearly changed era to era — something to read and learn from, not to copy.

② “Below the frontier” doesn't mean inefficient

In all three eras Buffett's point sat below the efficient frontier. But that curve is an optimistic (overfit) estimate, lifted by past high-flyers — so “below the curve” doesn't equal inefficient. DIVA grades efficiency against an estimation-error-corrected benchmark, and on that basis the broadly diversified 2011 (The Impregnable Fortress) and 2024 (The Veteran Foreman) read as “efficient,” while only the dot-com-crash-era, four-name 2001 (The Mapless Explorer) shows “room to improve.” And even where efficiency reads low, it doesn't mean "Buffett was wrong" — Markowitz judges one axis (return relative to past volatility), while Buffett also weighs management quality, business moats, long holding periods and taxes. Remember too that our analysis is a cross-section of part of the US-listed book, not the whole enterprise.

③ You can track your own portfolio the same way

What we did here used nothing exotic — just a public 13F, public prices, and the same two models. You can apply the identical procedure to your own holdings. Enter your tickers and weights into DIVA Quantizer and you'll get your value, size and quality exposures (β) plus risk, diversification and frontier position — in the same format as Buffett's era cards above. Turning the lens you used on a legend onto your own portfolio is the real point of this piece.

"Buffett the value investor," it turns out, wore a different face in each era once you look at the numbers. The habit of decomposing a cliché instead of accepting it — that starts with running your own portfolio once.

This article was written with the help of AI.