The legal machinery: what 10b5-1 is supposed to do
Rule 10b5-1 was designed to solve a practical problem. Corporate insiders often hold large amounts of stock and need to sell for ordinary reasons, but they are almost always at risk of being accused of trading while aware of material nonpublic information. The rule offers an affirmative defence if trades occur under a binding contract, instruction, or written plan adopted when the insider was not aware of material nonpublic information, and if the plan specifies the amount, price, and date, or includes a formula or algorithm that determines them.
That sounds neat in theory. In practice, the history of 10b5-1 is a long argument over whether it became too neat.
The original bargain
The original regulatory bargain was simple. If an insider commits in advance to a trading programme and then stays out of the way, the eventual trades should not be treated as suspicious simply because the insider later comes into possession of inside information before execution. The plan acts as a sort of legal time capsule.
For founders, this is attractive for obvious reasons. They can diversify without trying to guess blackout windows, investor sentiment, or whether every sale will be interpreted as a coded message. For boards and counsel, it reduces litigation and optics risk. For the market, at least in theory, it improves predictability.
The loophole problem
The criticism, well documented over the years by academics, journalists, and regulators, is that some insiders used 10b5-1 plans too flexibly. If a plan can be adopted, modified, cancelled, or overlapped with another plan too easily, then the supposed pre-commitment starts to look less like a restraint and more like a costume.
That criticism eventually led to the SEC’s 2022 amendments. Those changes imposed cooling-off periods for directors and officers, restricted overlapping plans, limited single-trade plans, and added disclosure requirements for issuers. The SEC’s message was not subtle: if you want the protection of a pre-arranged plan, it should actually be pre-arranged.
Why modern investors should care about the rule change
This matters for any “Bezos pattern” analysis because older founder-selling case studies often span years when the rule was looser in practice. Investors should be careful not to map every historical pattern directly onto the post-2022 regime.
A modern 10b5-1 plan for an officer or director now generally requires a cooling-off period of the later of 90 days after adoption or modification, or two business days after disclosure of the issuer’s financial results for the quarter in which the plan was adopted, subject to a cap of 120 days. There are also certifications for directors and officers, plus restrictions on overlapping plans and repeated single-trade plans.
In short, the legal plumbing has become less permissive. Good. It was due for maintenance.
How Bezos structures the sales, at least from the outside
Outside investors do not get the full plan document. They get filings, disclosures, timing, and patterns. That is enough to infer a fair amount, if not enough to satisfy the nosier corners of the internet.
Planned, not impulsive
The defining feature of Bezos’s stock disposals has been their systematic nature. Press coverage and SEC filings have repeatedly indicated that sales were made pursuant to Rule 10b5-1 trading plans. That means the execution dates and quantities were generally set in advance, or determined by formula, before the trades occurred.
This distinction is not cosmetic. A founder who wakes up, sees a strong quarter, and decides to dump stock that afternoon is making an active valuation call. A founder whose broker executes a scheduled sale under a standing plan may simply be following a calendar.
There is still room for judgement, of course. The real discretion may lie in when the plan is adopted, how much stock it covers, and whether the founder chooses to renew or suspend future plans. But that is a slower, more structural form of signalling.
Cadence matters more than any single sale
Investors tend to overweight one filing and underweight the sequence. With Bezos, the better question is whether sales appear in a recurring cadence that looks detached from immediate news flow. If so, the market should discount the informational content of each individual disposal.
A useful framework is to classify founder sales into three buckets:
| Bucket |
Description |
Likely informational value |
| Systematic plan sales |
Repeated sales under a disclosed 10b5-1 plan, with regular cadence |
Low to moderate |
| Semi-systematic sales |
Plan-based, but irregular in size or clustered around notable events |
Moderate |
| Discretionary sales |
No clear plan disclosure, sudden timing, unusual size |
Moderate to high |
Bezos typically sits in the first bucket more often than the third. That does not make the trades meaningless. It makes them different.
Size relative to ownership
A founder can sell billions and still not be “leaving”. This is one of the more annoying arithmetic facts in mega-cap founder analysis. Because Bezos’s stake has historically been so large, even substantial sales often represented only a modest fraction of his total holdings.
That is why post-transaction ownership matters. If a founder sells 2 percent of a position and retains overwhelming exposure, the sale is more plausibly interpreted as liquidity management. If the founder sells 30 percent in a compressed period, the conversation changes.
The SEC filing regime gives investors enough information to do this math. Most people simply prefer not to, because “sold $n/a billion” is easier to tweet than “reduced stake by a low-single-digit percentage while retaining alignment”.