The Goldman CEO’s biggest-ever sale is a filing in which he got 30% bigger David M. Solomon · Goldman Sachs Group · Form 4
- Who
- David M. Solomon, Chairman of the Board and Chief Executive Officer, The Goldman Sachs Group, Inc.
- What he did
- Took delivery of 109,896 shares on 23 October from a performance-based RSU granted in October 2021 — the Shareholder Value Creation Award — which paid at 150% of target, its maximum. The company withheld 60,773 shares at $891.30 for tax. On 27 October he sold 12,000 shares in three lines at $910.46, $911.49 and $912.11, for $10,928,694.
- The document
- Form 4, earliest transaction date 10/23/2026, accepted by EDGAR 27 October 2026 at 17:12 ET, signed by attorney-in-fact Jamie A. Greenberg. Footnote F1 states the delivered shares generally cannot be sold or transferred before October 2027. Footnote F2 identifies the 60,773 shares as tax withholding. F6 and F7 record 16,171 shares in a family trust (beneficial ownership disclaimed) and 15,205 held as trustee of a grantor retained annuity trust.
| Date | Code | Shares | Price | Direct owned after |
|---|---|---|---|---|
| 22 Oct (derived) | — | — | — | 122,057 |
| 23 Oct | M (acquire) | +109,896 | — | 231,953 |
| 23 Oct | F (withhold) | −60,773 | $891.30 | 171,180 |
| 27 Oct | S (sell) | −9,181 | $910.46 | 161,999 |
| 27 Oct | S (sell) | −2,382 | $911.49 | 159,617 |
| 27 Oct | S (sell) | −437 | $912.11 | 159,180 |
| Net, 22–27 Oct | — | +37,123 | — | +30.4% |
Consensus read
A 2021 award paid out at its maximum and the CEO immediately monetised part of it — $10.9m, the biggest sale of his career, struck within $4 of the all-time high, with no 10b5-1 plan to explain the timing. MarketBeat scored it a 7.01% cut to his position.
Steelmanned, this is the strongest version of the insider signal: discretionary sales by a sitting CEO into a record print are the subset of insider activity with the most documented predictive content, and the absence of a plan removes the usual defence that the trade was set months ago.
Contrarian read
This is an accumulation filing that contains a sale, and it is being reported as a sale filing.
He finished the week holding 37,123 more shares directly than on 22 October. He sold roughly one share for every four he gained. The 30.4% rise is the largest single-document increase in his direct stake in his recorded history — larger, in shares, than any sale he has ever made.
The “7.01% decrease” is arithmetically correct and analytically empty: it is measured from 171,180, a denominator that already contains the acquisition. It describes the second half of a two-part document.
Evidence that discriminates
- The starting number is in the same table and nobody used it. Row 1 reports 109,896 acquired with 231,953 owned after, so the 22 October position was 122,057. Any percentage computed from 171,180 has silently booked the acquisition and then reported only the disposal. This separates the two readings directly: the consensus number cannot be produced from the full document, only from half of it.
- The comfortable benign explanation fails too. Withholding was 60,773 of 109,896 shares — 55.30%, at or above the full combined federal, New York State, New York City and Medicare marginal rate on supplemental income. The vest was funded at source. He did not need to sell a single share to pay tax on it, so “it was just the taxes” is not available. The sale was discretionary. Both the easy bullish story and the easy bearish story lose their easy version on this one field.
- “Largest sale ever” is a price artifact, not a behaviour change. 12,000 shares is 56.5% of the 21,252 he sold on 29 January 2018. GS was $271.65 then and $910.72 now — 3.35×. Measured in the unit he actually controls, this is a below-record trade.
- The award shares cannot have been the source. Footnote F1 locks them until October 2027. The 49,123 net new shares are held whatever he thinks of $910. What he sold was free stock he already owned, which is the only stock he could sell.
- Base rate from his own record. Across eight sales between April 2022 and January 2026 his reported post-sale holding went from 96,467 to 125,527 — up 30% while selling in five of those years. The trend is not monotone (it dips in 2024 and again marginally in January 2026), but the direction is unambiguous: selling annually and ending each cycle larger is his pattern, not a departure from it.
Stated as inference, not as document: none of this shows Solomon is bullish. The increase is compensation retained, not capital deployed, and a man who has never made an open-market purchase in seventeen filings is not expressing a view by failing to sell. The claim is narrower and harder — his ownership rose, and the published percentage describes the wrong half of the filing.
What would falsify me
A Form 4 inside the next 90 days showing further non-plan sales that take his direct holding below 122,057 — his 22 October starting point. That would make this week the first tranche of a distribution and the accumulation framing an artifact of looking at one filing in isolation.
Equally: disclosure of a Rule 10b5-1 plan adopted now and timed to the October 2027 release of the 49,123 restricted shares. That would show the retention is a lock-up, not a choice, and the contrarian reading would be describing a constraint as a decision.
Pattern vs his own history Seventeen recorded transactions since February 2017. All sales. No open-market purchase on record. Share counts: 6,250 / 3,260 / 2,400 / 6,104 / 12,707 / 21,252 / 3,497 / 8,382 / 8,488 / 9,768 / 7,400 / 10,000 / 4,200 / 6,550 / 6,608 / 272 / 12,000. Median about 7,400; this one is roughly 1.6× the median — elevated, well inside the historical range, and smaller than two prior trades. The genuine outlier in the series is the 272-share sale of 29 January 2026, which is 4% of his median and has no obvious explanation in the record available here.