One Form 144 field — the three-month lookback table — carries today’s loudest signal, and it reads “Nothing to Report” eleven days before the largest director sale on the tape.
The Form 144 filed 2 October carries a “securities sold in the past 3 months” table that reads Nothing to Report. That is the filing’s own statement that CMXI B.V. sold nothing between early July and 1 October.
Then: 1,500,000 shares at $90.13 on 5 October, taking the line from 6,004,600 to 4,504,600. Seven days later, 2,000,000 at $94.15 — $188.30m, the single largest insider disposal filed this week — taking it to 2,504,600. A 44.40% cut in one trade; 58.3% of the position in a week.
Neither Form 4 reports a Rule 10b5-1 plan. These were discretionary trades, timed by a sitting director, with no quarterly trail behind them to be a tranche of.
The ladder runs $20.38 → $22.94 → $23.48 → $24.41 → $26.98 across 10 Sep, 21 Sep, 29 Sep, 2 Oct and 13 Oct. Every tranche cost more than the one before it; the average price rose 32% while roughly $97m went in. Accumulators normally average down.
The 13 October Form 4 shows 777,000 of the 844,130 shares printed at a flat $27.00 — the highest price he has paid — held direct, with no 10b5-1 plan box checked. On the same day’s podcast he said investing in GameStop “was a screw up,” described having “no exit hatch,” and told listeners to “bet against us.”
The headline is “director goes active at 12.2%.” The filing itself says “The Reporting Persons have no plans or proposals as of the date of this filing,” and that he intends to remain a director. He has sat on the board throughout. What changed is that Drew Holdings bought more than 2% in twelve months, which is the mechanical end of passive-filer eligibility. See The Log.
Each block is an argument, not a summary. The crux field cites only facts that separate the two readings; evidence consistent with both is left out on purpose.
Ryan Cohen · Chairman, President & CEO, GameStop Corp. · purchase 13 October 2026
Two open-market buys on 13 October: 67,130 shares at $26.75 and 777,000 at $27.00. Total 844,130 shares for about $22.77m. Holdings went from roughly 41,648,522 to 42,492,652 Class A shares, all held direct. Including roughly 3.7m warrant shares he reports beneficial ownership of about 46.2m shares, near 9.1%, per an amended Schedule 13D.
FORM 4 · GAME STOP CORP (GME) · Cohen Ryan — Director, Officer (Pres, CEO, COB)
Txn date 10/13/2026 · Code P (open-market purchase) · Class A Common, $0.001 par
67,130 @ $26.75 → 41,715,652 owned · Form of ownership: D (direct)
777,000 @ $27.00 → 42,492,652 owned · Form of ownership: D (direct)
Table II (derivatives): none · Explanation of Responses: blank · 10b5-1 plan: not reported
Signed 10/13/2026 · Accepted EDGAR 10/13/2026 16:06 EDT
Cohen is doing what he has done since September: buying his own stock with his own cash, no plan, no hedge, at escalating size. The self-deprecating podcast line is the trolling he is known for — the $22.8m is the real message, and it says conviction. Insider buying is the cleanest bullish signal in the filing record precisely because there is only one reason to do it.
Take him literally instead. “No exit hatch” is the operative disclosure, and the buying is what that condition produces rather than what conviction produces. At 9.1% of a company whose equity story is partly him, Cohen cannot sell without breaking the thing he owns. The purchase is 2.03% of the stake he already held — too small to change his economics, large enough to change the tape’s reading of him. This is expenditure on a signal, not an investment decision.
The slope of the ladder. Five tranches, 33 days, average price $20.38 → $22.94 → $23.48 → $24.41 → $26.98. A discretionary buyer deploying ~$97m on a value view scales down into weakness; the one buying a signal pays whatever the screen shows, because the point is that the filing exists. Cohen paid his series high for his second-largest dollar tranche. That asymmetry is not consistent with both readings — price-insensitivity is diagnostic.
The ratio. 844,130 / 41,648,522 = 2.03%. If this were about expected return, a 2% addition to an already-maximal concentration is close to irrelevant. If it is about what the market infers on 13 October, it is sufficient. Inference — the Form 4 states the share counts; the purpose is my reading of them.
What I excluded: the absence of a 10b5-1 plan, which both readings predict (a conviction buyer and a signaller both want discretion); and the stock’s $27.71 open on 14 October, which is noise at this size.
A sixth purchase at materially lower prices — say a large tranche under $22 after a drawdown — at size comparable to or above these. A signaller buys when attention is on him and the tape is firm; a buyer with a price view buys when it is cheap. One big buy into a decline and the signalling read is wrong.
Also fatal: any disposal at all. A Form 144 or a Form 4 sale from Cohen voids the “no exit hatch” premise the whole claim rests on.
Cohen has never reported a GameStop sale. The novelty here is not that he buys — it is the monotonic price ladder and the compression into 33 days. September alone carried three purchases; the prior comparable stretch had none.
Steven Schuurman · co-founder and Director, Elastic N.V. · record holder CMXI B.V. · sales 5 and 12 October 2026
5 October: 1,500,000 ordinary shares at $90.13 ($135.19m), holdings 6,004,600 → 4,504,600. 12 October: 2,000,000 at $94.15 ($188.30m), holdings 4,504,600 → 2,504,600. Both held of record by CMXI B.V., both brokered by Goldman Sachs, neither under a reported 10b5-1 plan. Against 104,991,280 shares outstanding, he is down to about 2.4%.
FORM 144 · ELASTIC N.V. · filed 10/02/2026 20:21 EDT · for account of CMXI B.V. (Director)
Shares to be sold 1,500,000 · Aggregate market value $140,910,000 · Broker: Goldman Sachs & Co. LLC
SECURITIES SOLD IN THE PAST 3 MONTHS: Nothing to Report
—
FORM 144 · filed 10/13/2026 09:59 EDT · notice date 10/13 · approximate sale date 10/12
Remarks: intended for submission 10/12, not submitted owing to a federal holiday
Shares to be sold 2,000,000 · Aggregate market value $191,040,000
Past 3 months: CMXI B.V. · 10/02/2026 · 1,500,000 shares · gross $138,375,000
—
ACQUISITION OF SECURITIES TO BE SOLD
04/28/2015 · 7,290,000 sh · sale/purchase/transfer by notarial deed from JTG Ventures B.V. · paid EUR 7,290
12/17/2020 · 5,186,000 sh · part of merger with IXII B.V. · paid EUR 0
—
FORM 4 · Schuurman Steven (Dir) · 10/05/2026 · S · 1,500,000 @ $90.13 → 4,504,600 · Form: I (by CMXI)
F1: CMXI holds of record; reporting person is controlling shareholder with sole voting and dispositive power; disclaims beneficial ownership except as to pecuniary interest
A co-founder with a near-zero cost basis is diversifying a legacy stake into strength, with the stock in the mid-$90s. The holdco, the Dutch structure, the bank and the round lots are all the ordinary machinery of a long-planned wind-down. It is an overhang, not information. Nothing to read.
There is no long-planned wind-down. The filings say the decision is recent: the sales are discretionary, they follow a quarter of complete inactivity, and they arrive 7 days apart in round lots at the top of the price range. Whatever caused this — a view, a commitment, a liquidity need — was formed in roughly the last few weeks, not in 2024. The structure is old; the selling is new.
The three-month lookback field on the 2 October Form 144 reads “Nothing to Report.” This is the single fact that separates the readings, because a long-running diversification program is exactly what that field is built to show. Had he been trimming quarterly, the table would carry tranches. It is empty. Eleven days later the next 144 lists one entry: the 1.5m sale that started this.
Neither Form 4 reports a Rule 10b5-1 plan. A pre-committed program of this size — $323m from a sitting director — is almost always run under a plan, both for the affirmative defence and because that is what the structure is for. Discretionary execution means the dates were chosen with current knowledge.
What I excluded as non-discriminating: the EUR 7,290 basis for 7.29m shares and the EUR 0 merger tranche. A near-zero basis is equally compatible with selling today and with selling five years ago, so it explains the size of the gain and nothing about the timing. Same for Goldman as broker and CMXI as holder, both present for years.
A field discrepancy worth logging. The 2 October 144 gives an approximate sale date of 10/02 and the later 144’s lookback dates the executed sale 10/02, but the Form 4 reports the transaction on 10/05. Three days. Probably notice-versus-execution, but it means the 144 lookback table and the Form 4 do not agree on when the first sale happened.
Any disclosure of a 10b5-1 plan adopted before September 2026 covering CMXI B.V. — a later Form 4 checkbox with an earlier adoption date, or an Elastic proxy or 8-K naming one. That would make these tranches of an old program and the “recent decision” claim simply wrong.
Equally: if the next two or three 144s show steady equal tranches running deep into 2027, the empty lookback table was a gap in an older program rather than its absence, and consensus was right that this is mechanical.
The filings establish what and when to the day. They do not establish why, and I am not going to pretend otherwise. Three live hypotheses:
The lookback field is the comparison: zero sales in the three months to 1 October, then 3,500,000 shares in seven days. Of the 12,476,000 shares the 144 records him acquiring in 2015 and 2020, 2,504,600 remain.
Control Empresarial de Capitales S.A. de C.V. · 10% owner, Talos Energy Inc. · signed by Marco Antonio Slim Domit, attorney-in-fact · sales 6–14 October 2026
| Trade date | Shares | Avg price | Proceeds |
|---|---|---|---|
| 15 Sep | 1,381,000 | $18.52 | $25.58m |
| 6 Oct | 480,000 | $16.9596 | $8.14m |
| 8 Oct | 2,310,000 | $17.03 | $39.34m |
| 9 Oct | 1,730,000 | $17.08 | $29.55m |
| 12 Oct | 2,200,000 | $17.25 | $37.95m |
| 13 Oct | 2,700,000 | $17.51 | $47.28m |
| 14 Oct | 265,446 | $17.30 | $4.59m |
| 6–14 Oct | 9,685,446 | — | $166.85m |
Holdings fell from about 39,079,036 to roughly 29,393,590 shares, held direct, still above the 10% reporting threshold. The same filer also sold 170,000 PBF Energy shares at $88.03 on 8 October, about $14.97m.
FORM 4 · TALOS ENERGY INC (TALO) · Control Empresarial de Capitales S.A. de C.V. — 10% owner
Paseo de las Palmas 781, Piso 3, Lomas de Chapultepec, Ciudad de México
Txn 10/06/2026 · Code S · 480,000 sh disposed · wavg $16.9596
F: sold in multiple transactions at prices ranging from $16.90 to $17.06
Shares owned following transaction: 38,599,036 · Form: D (direct)
10b5-1 plan: not reported
/s/ Marco Antonio Slim Domit, attorney-in-fact · 10/08/2026 · accepted EDGAR 10/08 17:58 EDT
Carlos Slim is backing out of Gulf of Mexico E&P. The largest holder cutting a quarter of its stake in seven sessions is a verdict on Talos, and the persistent discount in the shares is the market pricing in both the verdict and the remaining ~29m-share overhang. Sell-side will frame it as sponsor capitulation.
This is not a verdict on Talos; it is a size-target being worked by an administrator. The direction of the realised price through the sequence is the tell, and “capitulation” does not produce it. The vehicle is executing a quantity, accepting whatever bid clears it, and the only decision embedded in the tape is how much, taken before the window opened — not at what price, which is what a view would look like.
The weighted-average price rose on every single October print: $16.9596 → $17.03 → $17.08 → $17.25 → $17.51, and the largest block of the campaign (2.7m shares, $47.28m) executed at the highest average of the series. A holder dumping on a changed view hits bids and drags its own realised price down across the sequence; 9.69m shares left over seven sessions with the print improving throughout. That separates the readings — forced or conviction-driven selling has a characteristic downward realised-price slope and this has the opposite one.
The 15 September counter-fact, stated against myself. He sold 1,381,000 shares at $18.52 — higher than every October print. So he is not an opportunist waiting for strength either; he sold less at a better price and more at worse ones. That kills the flattering version of my own read and leaves the narrower one: the schedule is driven by quantity and capacity, and price is an output. Both the September and October data fit “working an order”; neither fits “timing the stock.”
Signature block. Marco Antonio Slim Domit signs as attorney-in-fact. The disposals are being administered by the family office, not executed by a principal reacting to the oil tape. Inference — the signature is a fact, what it implies about who decided is my reading.
Continued heavy selling through a material decline — sustained multi-million-share prints with the realised average falling, say into the low $15s — would show the vehicle is indifferent to execution quality and is in fact exiting a view.
Or the cleanest refutation: Control Empresarial taking the stake below 10% and filing an exit 13D/G amendment. A quantity-target seller stops at a level; a view-driven seller leaves.
Control Empresarial has historically been an accumulator in US energy names, including Talos and PBF. Nine of the last ten filings on this security are disposals, and 11.07m shares have gone since 15 September. The sign has flipped; the question this section answers is whether the magnitude implies a view, and the price slope says it does not.
Tor Olav Trøim · Director, Borr Drilling Ltd. · through Drew Holdings Ltd. / Drew Trust · buys 9, 13 and 14 October 2026
9 October: 2,500,000 shares at $4.74 ($11.84m). 13 October: 1,500,000 at $4.93 ($7.40m). 14 October: 750,000 at $4.7234 ($3.54m), deemed execution 15 October, taking the line to 39,617,808 shares. All indirect, held by Drew Holdings Ltd. On 13 October a Schedule 13D disclosed 12.2%.
FORM 4 · BORR DRILLING LTD (BORR) · Troim Tor Olav — Director
Txn 10/14/2026 (deemed execution 10/15) · Code P · 750,000 @ $4.7234
Shares owned following transaction: 39,617,808 · Form: I (indirect)
F1: Held by Drew Holdings Ltd., wholly owned by Drew Trust, a non-discretionary trust in which
the reporting person is the beneficiary
—
SCHEDULE 13D · filed 10/13/2026 · 12.2% via Drew Holdings
“The Reporting Persons have no plans or proposals as of the date of this filing”
Intends to remain a director · may buy, sell, hedge or distribute to other entities
A billionaire director moves from passive to activist disclosure while buying $22.8m of his own company in six days. The 13D is the escalation: he is signalling that he intends to drive outcomes at Borr, and the buying funds the influence. Coverage will lead with “goes active.”
The 13D carries no new intent and the word “active” is a filing category, not a behaviour. He was already a director; board influence was never the constraint. He crossed the 2%-in-twelve-months line that ends passive-filer eligibility, and the filing is the consequence. The purchases are the real content, and the 13D is being reported as though it were the event.
The Item 4 language. “The Reporting Persons have no plans or proposals as of the date of this filing.” An activist 13D is written to preserve optionality and typically enumerates board, capital-structure or strategic proposals. This one disclaims them and states he will continue as a director — the status quo. An intent change and a threshold crossing predict different Item 4 text, and this is the threshold version.
He already holds the board seat. A 13G-to-13D switch matters when it tells you an outsider now wants control. For a sitting director it conveys almost nothing about influence, because the influence predates the filing. The informational content of the switch is therefore close to zero — which is precisely what makes the “passive to active” framing a measurement artifact rather than a finding.
Counting against myself: the purchase sequence is not price-insensitive the way Cohen’s is — $4.74, then $4.93, then back to $4.7234 — and he has been buying since at least June. That is a genuine accumulation programme, and it is the thing worth watching. I am arguing the 13D is noise, not that the buying is.
A 13D/A within the next few months adding concrete proposals — board changes, a strategic review, a take-private approach, or a stated intent to seek control. That would mean the 13 October filing was the opening move of a campaign and the “mechanical” read was wrong.
Also falsifying: evidence that Drew Holdings had not crossed 2% in twelve months, which would remove the mechanical trigger and leave a deliberate choice to file 13D.
Not an anomaly. Drew Holdings bought roughly 1.06m shares in June 2026 and added again on 20 September before this week’s 4.75m. The vehicle has been accumulating for four months; the October buys are continuation at larger size, and the 12.2% is the cumulative result.
The 13D percentage and the Item 4 wording are drawn from a single summary of the filing rather than the filing text itself — hence the single-source chip. The Form 4 figures are from the filings.
Mandatory section. The candidate with the most contrarian-looking surface on this tape was Martine Rothblatt’s daily United Therapeutics selling. It does not survive contact with the plan terms, and the arithmetic says why.
Martine A. Rothblatt · Chairperson & CEO, United Therapeutics · family trusts · sales 5, 6, 8, 9, 12, 13, 14 October 2026
A founder-CEO liquidating about $5.1m a day, every trading day, with the stock near $545 and the proceeds routed through four separate family trusts, looks like a quiet, structured exit dressed as routine. The “10b5-1” label is the cover story; the relentlessness is the signal.
FORM 4 · UNITED THERAPEUTICS CORP (UTHR) · Rothblatt Martine A — Dir, Chairperson & CEO
Txn 10/06/2026 · Code M · 9,500 options exercised @ $117.76 strike
Options vested 1/3 each on 03/15/2021, 03/15/2022, 03/15/2023 · expire 03/15/2027
Code S · 9,500 sh sold across 17 transactions, $527.90–$545.50 · gross ~$5.12m
Rule 10b5-1 checkbox: marked · plan adopted 11/07/2025
Plan runs until the earlier of (a) exercise of 1,734,410 options, all expiring 03/15/2027,
or (b) 12/31/2026
Holdings after: options 290,410 (indirect by trust) · common 628,049 indirect by trusts
+ 40,513 direct + 166 by spouse
Three facts dismantle the quiet-exit reading, and each of them is in the filing rather than in my head:
The real mechanic is mundane: options struck at $117.76 expire in March 2027 against a $545 share price. Letting them lapse would forfeit roughly $124m of intrinsic value. Exercise-and-sell at a fixed daily rate is the cheapest way to convert them without moving the stock. The ordinary reading survives because the plan document explains the behaviour completely, and the division comes out right.
A daily quantity that varied with price, a plan adopted within weeks of the selling, or a declining underlying position. None of those are present. Had the 290,410 remaining options not divided cleanly into the sessions left before 31 December, the calendar explanation would have been incomplete and the view explanation would have had room.
This is not incidental plumbing. Read the ownership-form column across the week and the same structure appears in four unrelated jurisdictions:
| Person | Vehicle | Form | 10b5-1 |
|---|---|---|---|
| Steven Schuurman | CMXI B.V. (Dutch holdco) | Indirect | Not reported |
| Tor Olav Trøim | Drew Holdings Ltd. / Drew Trust | Indirect | Not reported |
| Slim family | Control Empresarial de Capitales S.A. de C.V. | Direct (by entity) | Not reported |
| Martine Rothblatt | Four family trusts + an options trust | Indirect | Adopted 07 Nov 2025 |
| Ryan Cohen | — none — | Direct | Not reported |
Two things follow. First, the vehicle does work the person cannot do: Trøim’s Drew Trust is described in the filing as non-discretionary with him as beneficiary — a deliberate choice to separate legal title from control. Schuurman’s footnote states he holds sole voting and dispositive power over CMXI’s shares while disclaiming beneficial ownership beyond his pecuniary interest. Rothblatt’s four trusts differ from one another in exactly who is trustee, who is settlor and who can revoke — which is the point of having four.
Second, and more useful: the plan box is empty almost everywhere. Only Rothblatt — the one case where the behaviour turned out to be fully mechanical — reports a 10b5-1 plan. The three largest discretionary actions of the week were all made by people who kept their discretion. The folk belief that big insider trades are pre-scheduled housekeeping is the opposite of what this tape shows, and it is why the plan checkbox is the first field worth reading on any Form 4.
Most of what is above needs scale, a US domicile or a liquid listed stock. Three mechanics survive the trip down; three do not, and saying so is the more useful half.
This is the week’s one genuinely scale-free mechanic. Schuurman’s CMXI B.V. and your Ontario holdco do the same structural job: the operating shares sit under an entity you control, so proceeds, voting and succession are separable from each other and from you personally. In the Canadian version a holdco lets you move surplus out of the operating company by inter-corporate dividend, hold investments corporately, and — with a family trust in the structure — give several people a claim on future growth without giving any of them your present control. That is exactly the Drew Trust pattern: beneficiary without discretion.
First step this week: write down, in one page, who holds the shares of Rocky Software today and in what proportion, and whether there is any class of share other than the one you hold. You cannot evaluate a freeze, a trust or a sale without that page, and most founders cannot produce it from memory.
The hard constraint to know before you like this too much: a Canadian family trust faces a deemed disposition on its 21st anniversary, which can trigger tax on accrued gains unless assets are rolled out to beneficiaries first. Structures built at 40 create a problem at 61. That is a planning input, not a reason to avoid it.
The Form 144 shows Schuurman acquired 7,290,000 shares for EUR 7,290 — a tenth of a cent each — and 5,186,000 for nothing. Nearly the entire proceeds are gain. Your position is structurally identical: founder shares issued at nominal value, with all the value accrued since.
The Canadian analogue is the lifetime capital gains exemption on qualified small business corporation shares, which shelters a substantial amount of gain on a sale of shares (not assets) if the company meets the QSBC tests — broadly, a Canadian-controlled private corporation with the large majority of its assets used in an active business in Canada, tested both at the moment of sale and over the preceding 24 months. The exemption amount is indexed and has been increased in recent years; confirm the current figure with your accountant rather than relying on a number from a report.
What does not transfer is Schuurman’s liquidity. He can sell 2m shares on a Monday. You cannot, which means the QSBC tests have to be satisfied long before a buyer appears, not during diligence.
First step this week: ask your accountant one question — “if I sold the shares tomorrow, would they qualify as QSBC shares, and if not, what is disqualifying them?” The common answers are accumulated cash, passive investments or non-active assets sitting in the operating company, and all of them are fixable with lead time and awkward without it.
Rothblatt’s plan was adopted 7 November 2025 and fixed a quantity per session; October’s price was irrelevant to October’s behaviour. You have no 10b5-1 regime and no public stock, so the instrument does not transfer — but the discipline is the whole content and it costs nothing.
The private-company version: pre-commit the rule, in writing, for the decisions you will otherwise make emotionally. How much of a windfall or a large contract payment goes to the corporation versus out to you. At what revenue level you take a salary increase. What share of surplus is invested rather than held in cash. Written in advance, these are policy; written in the moment, they are mood.
First step this week: one rule, one sentence, with a number in it, dated. Rothblatt’s was “9,500 shares per session until the options are gone.” Yours might be “every payment above $X goes 60/40 to retained surplus and salary.” The form matters less than the fact that the price is unknown when you write it.