The most-cited Buffett line on holding period is: "Our favorite holding period is forever." Which is true. It's also misleading taken alone, because the same investor who said that has also sold positions for sound, value-investing-aligned reasons - and the why of those sales is the meaningful part.
This post walks through the three legitimate reasons to sell a value-investing position and the four illegitimate reasons that destroy retail returns. With worked examples of when Buffett himself has sold and why.
The three legitimate reasons to sell
Reason 1: The framework verdict has changed
If a stock passed 7-of-7 frameworks when you bought it and now passes 4-of-7, something material has changed. Investigate. If the change is structural (eroding moat, declining gross margin trend, deteriorating capital allocation), that's a sell signal.
Worked example: Berkshire's reduction in Apple (2024-2025). Buffett did not exit Apple - but he reduced from ~50% of equity book to ~30%. The framework consensus on Apple's valuation pillar had moved from Strong Fit to Partial Fit as the multiple expanded. The 5-pillar Buffett-Brain verdict had downgraded. Buffett re-sized accordingly. That's the framework-update sell, properly applied.
Reason 2: A materially better opportunity has emerged
Capital is constrained. If you hold Stock A passing 5-of-7 frameworks, and Stock B with the same risk profile passes 7-of-7 at a comparable price, the rational move is to swap. This is opportunity-cost selling, and it's perfectly legitimate.
Worked example: Berkshire's accumulation of the Japanese trading houses (2020-2024). Buffett took ~5% positions in five Japanese sogo shosha (Itochu, Marubeni, Mitsubishi, Mitsui, Sumitomo). The capital came from trimming positions where the framework consensus had weakened. Better opportunity → portfolio rotation.
Reason 3: The business has fundamentally changed beyond what's recoverable
This is the rarest legitimate reason. A business undergoing a fundamental structural change (regulation, technology, market structure) that the framework consensus can't accommodate may need to be exited.
Worked example: Buffett exited his banking positions (Wells Fargo, Goldman, JP Morgan) in 2020-2022. The thesis: low interest rates + intensifying regulatory pressure + Buffett's specific concerns about Wells Fargo's culture made the original moat case un-recoverable. Not a framework update, not opportunity cost - a structural conviction reset.
These three reasons combined account for the vast majority of good value-investing sells.
The four illegitimate reasons retail investors sell
Bad reason 1: The price went down
If your thesis hasn't changed and the framework consensus hasn't changed, a falling price means the future return is higher (you're getting more cash flow per dollar of market cap). Buffett has been explicit that he prefers stocks to go down before he buys more.
The retail mistake is selling on a 20% drawdown because it "feels bad." If the business is fine, the falling price is opportunity, not warning. Do not sell because the price went down.
Bad reason 2: The price went up and you want to "lock in gains"
If a 7-of-7 framework consensus stock is up 50%, the rational response is to check whether the framework consensus has changed. If yes (the rally pushed valuation from 7 to 4-of-7), sell. If no (the framework still passes), the higher price is the market catching up to the underlying business value - not a sell signal.
The retail mistake is selling because "I'm up 50%, I might as well take it." That's anchoring to entry price, which has no relationship to current value.
Bad reason 3: You need the money
Liquidity-driven selling is the silent return-destroyer. If your time horizon is unclear and you might need the capital in 18 months, you should not have been in individual stocks in the first place - that's a public-equity timeline mismatch.
The legitimate response is to never put capital you might need short-term into individual equities. Index funds + cash buffer cover the unpredictable life events. Individual stocks are for capital you genuinely won't touch for 5-10 years.
Bad reason 4: The news cycle is loud
Pandemic, election, recession headlines, geopolitical conflict, a famous CEO leaving - every news event has historically been a poor selling trigger. The 2020 COVID drawdown is the cleanest recent example: investors who sold in March 2020 generally never returned. Investors who held generally fully recovered by August.
Buffett's published view on this: "If you aren't willing to own a stock for ten years, don't even think about owning it for ten minutes." The news cycle is not your friend. It's noise on a 10-year horizon.
What Buffett's longest-held positions tell us
Buffett's holding history is the data:
- American Express (AXP) - bought 1964, still held 2026. 62 years.
- Coca-Cola (KO) - bought 1988, still held 2026. 38 years.
- Moody's (MCO) - bought 2000, still held 2026. 26 years.
- Geico (Berkshire subsidiary) - bought 1976, still held. 50 years.
The common thread on these long-held names: framework consensus has remained intact. AmEx still passes 7-of-7. Coca-Cola still passes 6-of-7. Moody's still passes 7-of-7. Geico is a private business but its insurance-float economics have improved over time.
When the framework remains, the position remains. When the framework changes (Kraft Heinz, AT&T historically, several others), the position changes.
How to apply this to your own holdings
The honest workflow:
- Annually re-score every position against the 7-framework consensus on invest-like. (Open each /buffett/[ticker]/ page.)
- Compare to last year's score. If consensus has dropped by 2+ tiers (e.g., from 7-of-7 to 4-of-7), investigate.
- Decide if the drop is structural or temporary. Structural = sell. Temporary = hold and watch.
- Look for opportunity-cost sells. Is there a higher-conviction position you could rotate into?
- Ignore the price. Both directions. Price is noise; framework consensus is signal.
The discipline is in the what of selling, not the when.
Where invest-like helps
Every stock has a /buffett/[ticker]/ page that surfaces:
- Current 7-framework consensus
- 5-pillar Buffett-Brain verdict
- Historical scoring versions (so you can see how the verdict has changed)
- Boardroom debate with the four investor personas
- Risk & Upside scenarios
The annual review workflow is 5-10 minutes per stock. For a 20-position portfolio, that's a 3-hour annual sit. Less work than rebalancing a portfolio of mutual funds, more rigorous than what most retail investors do.
Disclosure
Educational tool. Examples named (AmEx, Coca-Cola, Moody's, Apple, Wells Fargo, Kraft Heinz) are historical and analytical, not recommendations. Past holding history is not a strategy template - Buffett's long horizon depended on his specific framework consensus on each name. Each investor has their own horizon and constraints.
Author: Zaid Ghazal, founder of invest-like, indie SaaS, Kiel, Germany. Not a registered investment adviser.