Wealth Signal

Wealth Signal

Edition 053 5 November 2026 Coverage 29 Oct – 5 Nov 2026

Anomaly board

2 breaks

Actions that break the person’s own prior pattern

01

Lip-Bu Tan · CEO and director, Intel

Trade 2 Nov · Form 4 reported 4 Nov

Bought 150,000 shares at a $64.7312 weighted average — $9,709,680 — on the open market, through his family trust, with no Rule 10b5-1 plan. This is the first unforced open-market purchase of his twenty months as CEO. His only prior acquisitions were a $25M purchase in March 2025 that his employment terms required him to make, and a $10.0M subscription to Intel’s own August 2026 share offering. Between those two he neither bought nor sold.

The timing breaks pattern too: he stepped in two sessions after Intel fell 28.58% on 30 October, its largest one-day drop since 1974, and the stock has fallen a further 10% since he bought.

02

Dušan Šenkypl & Jan Barta · Pale Fire Capital, 10% owners of Dole plc

Trades 2–4 Nov · Form 4 accepted 4 Nov

Added 986,557 shares at $13.05–$13.38 in three sessions, taking the reported position from 9,634,264 shares to 11,269,177 — past 10.2% to roughly 11.9% of the class. The break is the price: their own Schedule 13D/A of 14 October puts the aggregate cost of the position at $112,043,544, about $11.63 a share. They are now buying 12–15% above their own blended basis after four years of accumulating below it.

Checked and not anomalous. Travis Boersma’s $106.4M of Dutch Bros sales (2–4 Nov) ran inside 10b5-1 plans and inside his established cadence. Jason Murray’s PACS sales ran inside a plan adopted 18 May. Marc Andreessen’s reported $12.8M Meta disposal was a fund transaction, not a change in his own holding — see the log. No domicile change, no first-time pledge, no trust or holdco formation, and no resignation-plus-disposal surfaced in this window. Larry Ellison’s 413-million-share Oracle pledge and Peter Thiel’s $130M Casa Encantada purchase are both real and both outside it — September and early October.

02

The log

three person-actions

The “$4.8M loss” requires deleting his largest and cheapest purchaseLip-Bu Tan · CEO and director, Intel Corporation (NASDAQ: INTC)

Purchase confirmed · 5 sourcesMarch 2025 tranche · 2 independent sourcesMarch execution price is an estimateForm 4 read via transcription, not EDGAR
What they did
On 2 November 2026 bought 150,000 Intel common shares on the open market at a weighted average $64.7312 (range $64.705–$64.735), for $9,709,680. Held indirectly through a family trust; no Rule 10b5-1 plan. Shares beneficially owned following the transaction: 1,464,669 through the trust, plus 500 in a 401(k).
The document
SEC Form 4, Intel Corporation, reporting person Tan Lip-Bu, transaction date 2 Nov 2026, reported 4 Nov 2026. Fields as transcribed by Investing.com, BigGo Finance, Calcalist and MarketScreener, which agree on every figure. I was not able to open the filing on EDGAR directly from this environment.

Consensus read

“Insiders only buy for one reason.” The CEO put $9.7M of personal money into a stock that has halved in five weeks, with no plan to hide behind, lifting his holding 11.4%. He is already underwater on his August buy and bought anyway.

Read as painful, high-conviction commitment: the man with the best information thinks the Q4 guidance wipeout is overdone, and he is taking real personal loss to say so.

Contrarian read

The purchase is real and genuinely discretionary. The frame around it is arithmetically false. Tan is not underwater on Intel. He is up roughly $31M, about 69%, on a personal position with an average cost near $34.

Every tally in circulation today starts the clock in August 2026 and counts 255,263 shares. His own Form 4 says he holds 1,464,669. The $9.7M is a 22% top-up funded by a very large embedded gain — a cheap option, not a sacrifice. That changes how much the act should move your priors.

Tan’s three personal Intel acquisitions. March 2025 price is the market price at the time as reported by Tom’s Hardware (9 April 2025), not a Form 4 execution price — see the inference note. Valued at Intel’s 4 November 2026 close of $58.27.
DateMechanismSharesPriceCost
Mar 2025Required by employment terms1,043,406~$23.96$25,000,000
11 Aug 2026Subscription, Intel’s own offering105,263$95.00$9,999,985
2 Nov 2026Open market, no plan150,000$64.7312$9,709,680
TotalAverage cost $34.431,298,669$58.27+$30.96M
As reportedAugust and November only255,263$58.27−$4.84M

Evidence that discriminates

  1. The Form 4’s own holdings field contradicts the commentary. “Shares beneficially owned following transaction: 1,464,669.” The widely circulated loss figures — 24/7 Wall St’s “$20 million,” Calcalist’s “$19.7M at an average $77.20,” FourWeekMBA’s “$19.7M at an average $77” — are computed on 255,263 shares. That is 17% of what the filing says he holds. The number they are all dividing by is not the position.
  2. The reconciliation is exact, which rules out my being the one making an error. 255,263 shares at $58.27 is $14,874,174 against $19,709,665 invested — a loss of $4,835,491, or 24.5%. That reproduces the published figure to the dollar and the decimal. So the published figure is not using different data; it is using the same data with the first tranche removed.
  3. The conclusion is robust to the one number I am estimating. The ~$23.96 March price is an estimate. It does not matter: for Tan to actually be at break-even on the full 1,298,669 shares, the March tranche would have had to cost $53.63 a share. Intel traded in a $19–$27 band through 2025. There is no plausible March price at which the “he is down” framing becomes true.
  4. Calcalist says so itself and then does not do the arithmetic. Its own text concedes the March purchase “remains deeply in the money” — and then excludes it from the totals in the next paragraph, which is the number everyone else picked up.

Inference, not a field I read: the 166,000-share gap between the three purchases (1,298,669) and the reported holding (1,464,669) is almost certainly vested equity from his pay package. I did not read a vesting schedule. It does not affect the argument — shares he was given make the cost-basis cushion larger, not smaller.

What would falsify me

Intel’s next proxy beneficial-ownership table showing Tan holding materially fewer than ~1.3 million shares — which would mean the March 2025 tranche was sold, forfeited or never sat where I think it does. Or a Form 4 or 8-K showing the March 2025 shares were acquired at a price near $77 rather than near $24. Either one and the “down $4.8M” framing is right and this entry is wrong.

Narrower test on the conviction question: if his next open-window purchase is materially larger in dollars than ~$10M, the “cheap top-up” characterisation weakens and the consensus conviction read gains.

Pattern vs his own historyThree acquisitions in twenty months. March 2025: contractually required, roughly 1.04M shares, reportedly subject to a five-year hold. August 2026: a $10,000,000 subscription — 105,263 × $95, a round dollar figure, not a round share count — into the offering Intel itself was placing. November 2026: 150,000 shares, a round share count, the whole order filled inside a three-cent band, in the first open trading window after the 29 October earnings release. The dollar amounts of the last two are within 3% of each other across a 32% difference in price. That is the signature of a size he has decided on, not a price he has judged.

0
1

Šenkypl and Barta already sold this stake once — to a company they controlDušan Šenkypl & Jan Barta · Pale Fire Capital SE / SICAV a.s. · Dole plc (NYSE: DOLE)

Form 4 fields confirmed13D/A fields confirmed2022 transfer: single sourceCost-basis figure derived
Form 4, Dole plc, accepted 4 November 2026. Reporting persons: Pale Fire Capital SE, Pale Fire Capital SICAV a.s., Pale Fire Capital investični společnost a.s., Dušan Šenkypl, Jan Barta. 10% Owner box checked. Ownership form: indirect on every row, held by the SICAV. No 10b5-1 plan.
DateSharesWtd avgRangeOwned after
2 Nov 2026230,294$13.2572$13.2173–13.290610,512,914
3 Nov 2026753,563$13.0477$13.0360–13.077611,266,477
4 Nov 20262,700$13.38—11,269,177
What they did
986,557 shares bought for about $12,921,484, lifting the reported position from 9,634,264 (10.2% of the class at the 14 October 13D/A) to 11,269,177 — roughly 11.9% of the 94,458,807 shares outstanding at 29 July. Up 17.0% in three weeks. All held by Pale Fire Capital SICAV a.s.; both individuals disclaim beneficial ownership except to the extent of their pecuniary interest.

Consensus read

Benzinga’s headline on this exact filing: “Insider Confidence On Display.” A 10% owner bought $12.9M more of a cheap food business, including 753,563 shares in a single session.

Read as two founders of a successful Czech investment house pressing an undervalued position harder — the kind of concentrated conviction buying that tends to precede a bid or a re-rating.

Contrarian read

Neither man bought anything. A fund structure they control bought, and the specific thing that makes this interesting is the reverse trade: in December 2022 Pale Fire Capital SE acquired a 7.8% Dole stake from Šenkypl and Barta for $76.2M.

The individuals converted direct ownership into cash and indirect ownership years ago. What you are watching now is a balance sheet that is no longer theirs in the direct sense adding to a position at prices above the blended cost of the position itself. Call it corporate capital allocation; “insider confidence” is the wrong label.

Evidence that discriminates

  1. The ownership column again. Every row of the Form 4 is indirect, held by Pale Fire Capital SICAV a.s. The individuals appear because of control, and the filing records that each “disclaims beneficial ownership except to the extent of” pecuniary interest. That field is what separates personal conviction from corporate allocation, and it says corporate.
  2. They are averaging up, not down. Item 3 of the 14 October 13D/A gives an aggregate purchase price of $112,043,544 for the 9,634,264 shares then held — $11.63 a share. The November buying went off at $13.05–$13.38. An investor pressing a mispricing wants a lower average; this raises it. That is consistent with a mandate or a target weight and inconsistent with the “cheap, so buy more” story the headline implies.
  3. The 2022 direction of travel is the fact that discriminates hardest. If these two men personally believed Dole was the asset to own, the moment to own it personally was at $10.31 a share in December 2022. They sold it to their own vehicle instead. Everything since has been the vehicle’s money.

Where I am inferring: the $11.63 figure assumes Item 3’s aggregate price covers the whole 9,634,264-share position. Some 13D amendments restate Item 3 only for shares acquired since the prior amendment, which would make the true blended basis different. I could not open Exhibit 2, the 60-day transaction schedule, to check. The 2022 transfer is also described in only one source, which does not confirm the stake was held personally rather than through another entity.

What would falsify me

If the next 13D/A restates the aggregate purchase price in a way that puts the blended basis above $13.38, the averaging-up point dies and the ordinary “adding to a cheap position” reading is correct. If the original December 2022 13D shows the 7.8% stake was sold out of another Pale Fire entity rather than by the two men personally, the “they already took their cash out” framing is wrong and should be struck.

Pattern vs their own historyPale Fire has been a Dole holder since 2021 and has filed seven 13D amendments. The consistent shape is accumulation through a Czech SICAV, with the individuals named only as controllers. The change this week is tempo and price — 1.63 million shares in three weeks, at a level above the stated blended cost — and the crossing of roughly 11.9% without a stated control intention.

0
1

Andreessen did not sell $12.8M of Meta. Fund VIII did.Marc L. Andreessen · Director, Meta Platforms Inc. (NASDAQ: META)

Form 4 fields confirmed10b5-1 status not establishedPecuniary share is inference
What they did
On 2 November 2026, Andreessen Horowitz Fund VIII, L.P. sold 20,951 Meta Class A shares at $606.85–$613.88, about $12.77M. Andreessen is a managing member of AH Equity Partners VIII, L.L.C., the fund’s general partner; he may be deemed to share voting and dispositive power and disclaims beneficial ownership except to the extent of his pecuniary interest. The fund no longer holds shares attributable to the reported sales. His own indirect holding — 49,448 shares through the LAMA Community Trust, where he and his spouse are trustees — did not move.

Consensus read

Headline across at least four syndicated outlets and three languages: “Meta director Marc Andreessen sells $12.8M in stock.” It appeared as the second-largest insider sale on the 4 November tape.

Read as a long-standing Meta board member reducing exposure near $610 — a director’s view on the stock after a year of heavy AI capital spending.

Contrarian read

This is a venture fund disposing of a lot, filed on a director’s form because he sits on the board. Andreessen’s personal position is the 49,448 shares in the LAMA Community Trust — roughly $30M at the sale price — and it is untouched.

A fund exiting a position it holds for limited partners is governed by fund life, distribution obligations and portfolio construction. It is not a sentence about Meta’s price, and the person named in the headline kept every share that is actually his.

Evidence that discriminates

  1. Two holdings, one moved. The filing reports the fund’s lot as fully disposed and the LAMA Community Trust holding as unchanged at 49,448. If this were Andreessen forming a view on Meta at $610, the trust is the account that would move. It is the account that did not. A sale that leaves the seller’s own money entirely in place is not a sale by the seller.
  2. The disclaimer is a legal statement about economics, not boilerplate to skip. “Disclaims beneficial ownership except to the extent of his pecuniary interest” means his economic share of the $12.77M is his general-partner interest in the fund, not the $12.77M. I do not know AH VIII’s carry and commitment terms, so I will not put a number on it — but a typical GP economic share would make his share of this sale a low single-digit percentage of the headline.
  3. Size sanity check. 20,951 shares is a rounding error against Meta’s float. It is, however, exactly the size of a fund lot being cleared. The transaction looks like what it is on the face of the form.

What would falsify me

A Form 4 in the next ninety days showing the LAMA Community Trust selling Meta shares. That would make the fund disposal the leading edge of a personal reduction and the headline closer to right than I am giving it credit for. I also could not establish whether a 10b5-1 plan covered the fund sale; if one exists and was adopted recently, the “mechanical fund life” reading needs a second look.

Pattern vs his own historyAndreessen has been a Meta director since 2008 and his Form 4s have for years been a mixture of small personal trust activity and larger a16z fund transactions reported under his name. A $250,907 personal sale and a $12.77M fund sale both arrive labelled “Marc Andreessen sells.” Nothing about this week is new — which is precisely the point. The recurring error is in the reporting, not the behaviour.

0
2
03

Where consensus is right

tested and could not overturn

Murray’s $11.1M PACS sale is exactly as boring as it looksJason Hulse Murray · Co-founder, CEO, chairman, director and 10% owner, PACS Group, Inc. (NYSE: PACS)

Consensus survivesPer-day split not confirmed

What they did. Sold 225,296 shares between 2 and 4 November 2026 at $45.94–$51.3492, for about $11,087,760, under a Rule 10b5-1 plan adopted 18 May 2026.

The claim I tested. This is the most tempting item on the tape. A co-founder-CEO who is also a 10% owner sells eleven million dollars of stock in the three sessions straddling his own earnings release, in a company that has spent two years under short-seller allegations and regulatory scrutiny. The obvious contrarian move is to argue the plan is cover and the timing is the signal.

It does not survive contact with the dates, and the ordinary reading wins. Three things kill it:

Evidence that discriminates

  1. The plan predates the window by five and a half months. Adopted 18 May 2026, sales executed 2–4 November. Whatever he knew on 2 November, he did not choose that date in possession of it.
  2. The news was good, and he sold into it. PACS reported Q3 on 4 November with revenue growth and raised full-year guidance; the stock set a new twelve-month high, and Stephens raised its price target the following morning. The insider-trading worry is a seller who knows something bad. Here the seller gave up upside — the $45.94 end of his range is roughly 11% below where the stock finished the window. An informed seller does not do that.
  3. The price range is the market’s, not his. An 11.8% spread across three days is what a stock does around a beat. It is not evidence of discretion; it is evidence of a mechanical schedule running through a volatile print.

The honest caveat

I did not get the per-day breakdown from the primary Form 4, so I cannot say how much of the 225,296 executed before the 4 November release versus after it. If the bulk landed after the beat at the top of the range, the “gave up upside” point weakens considerably — though the May plan adoption date would still carry the argument. Worth pulling the filing if PACS comes up again.

0
1
04

What repeats

one mechanic, two shapes

First: the tally that starts on a convenient date. This is the third edition in four running days where the strongest available finding was an error in the measurement rather than an error in the market’s judgement.

The structural reason is the same each time: these numbers are assembled from filings automatically, and the assembly step is where the denominator, the date range or the entity set gets chosen. Nobody is lying. The choice is just never shown.

Second, the one that will not go away: the person almost never transacts as themselves. Three of the four items in this edition are entity transactions reported under an individual’s name — Pale Fire Capital SICAV, Andreessen Horowitz Fund VIII, and the DM Trust / DM Individual Aggregator vehicles behind the Dutch Bros sales. Each one carries a disclaimer of beneficial ownership on the face of the form, and in each case the secondary coverage drops it. This repeat has now appeared in four consecutive editions. The ownership column, D or I, is the single highest-value field on a Form 4 and the one most reliably discarded in transit.

05

Transferable to you

Ontario · private company · honest verdicts

Keeping a complete, dated tranche record

Transfers
0
1

The Tan item is a cost-basis failure, and cost basis is the one thing on this page that works identically at every scale. Four financial outlets got the sign of a position wrong because they could not see the oldest tranche.

In Canada the trap bites differently and harder. Identical shares of the same class are pooled into a single averaged adjusted cost base — you cannot pick which lot you sold, the way a US holder can with specific identification. That means the oldest, cheapest tranche is dragged into every disposition whether you remember it or not, and a missing early purchase record does not just mislead you, it misstates your return on an actual filing. The same averaging applies across your non-registered portfolio, to reinvested distributions, and to shares of your own company acquired at different times.

THIS WEEK → One spreadsheet, one row per acquisition: date, instrument, shares, price, total cost, vehicle (personal / holdco / TFSA / RRSP), and running averaged ACB. Include the founder shares in the operating company at their original subscription price. If you cannot produce an ACB for those shares today, that is the gap to close before anything else structural — an LCGE claim at exit is computed off it.

Selling your personal shares to a company you control

Does not transfer
0
1

The Pale Fire mechanic — Šenkypl and Barta selling a 7.8% stake to their own holding company for $76.2M of cash — is the single most attractive-looking move on this page, and in Canada it runs straight into section 84.1 of the Income Tax Act. That provision exists specifically to stop a shareholder from extracting corporate cash as a capital gain by selling shares to a non-arm’s-length corporation; it can recharacterise the proceeds as a deemed dividend and grind the paid-up capital of what you receive.

The Czech structure does not map and the arithmetic that makes it work for them does not survive the border. There are legitimate Canadian routes to a holdco layer — a section 85 rollover, an estate freeze, a crystallisation of the exemption — but they are built the opposite way round from a cash sale, and every one of them is a planned, documented transaction, not a trade. The useful takeaway is the negative one: if a structure’s appeal is that it turns corporate cash into a personal capital gain, assume it has already been legislated against here.

Buying your own company’s stock to signal

Does not transfer
0
1

Tan’s purchase needs a public market, a quoted price he did not set, a blackout calendar that makes the timing legible, and an audience that reads Form 4s. You have none of the four. There is no window, no print, nobody to signal to, and no mechanism by which buying more of your own private company changes anyone’s estimate of it.

The honest analogue is uncomfortable rather than actionable: you are already at Tan’s 100% concentration without his liquidity, his ability to size the position, or his exit. The thing worth borrowing is not the trade but the question behind it — whether the blended cost of everything you own in one company is something you could state from memory right now.

This is information, not financial, tax or legal advice. Tax treatment depends on facts this page does not have, and section 84.1, the ACB averaging rules and the QSBC/LCGE tests all turn on specifics. Confirm anything structural or tax-related with a licensed Canadian accountant or tax advisor before acting on it.

06

Sources

  1. Investing.com — Wednesday’s top insider buys and sells: Intel CEO adds $9.7M stake5 Nov 2026 · Tan Form 4 fields, Boersma and Murray detail, 10b5-1 adoption dates
  2. BigGo Finance — Intel stock halves in a month as CEO buys $9.71M of shares4–5 Nov 2026 · $58.27 close, −28.58% on 30 Oct, Aug 2026 tranche, trust holding
  3. Calcalist — Intel CEO Lip-Bu Tan buys the dip, investing $9.7M despite losses on earlier purchase5 Nov 2026 · the $4.8M / 24.5% computation and its stated exclusion of the March 2025 tranche
  4. 24/7 Wall St — Lip-Bu Tan put $20 million into Intel and is down $4.8 million5 Nov 2026 · the headline framing this entry disputes
  5. Tom’s Hardware — Tan’s March 2025 purchase: 1,043,406 shares at about $23.969 Apr 2025 · primary numeric source for the March 2025 tranche
  6. Tom’s Hardware — Intel CEO puts more of his own money in; March 2025 buy was contractually required, five-year hold13 Aug 2026 · corroborates the requirement and “neither bought nor sold” between March 2025 and Aug 2026
  7. CNBC — New Intel CEO Lip-Bu Tan to receive $69 million compensation package14 Mar 2025 · employment terms context
  8. MarketBeat — Intel Q3 2026 earnings, 29 October 202629 Oct 2026 · establishes the earnings date behind the open-window timing
  9. StockTitan — Dole plc Form 4, Pale Fire Capital, 2–4 November purchasesAccepted 4 Nov 2026 · all three rows, weighted averages, ranges, holdings after
  10. StockTitan — Dole plc Schedule 13D/A No. 7, Pale Fire Capital 10.2% stakeFiled 14 Oct 2026 · reporting persons, 9,634,264 shares, $112,043,544 aggregate cost, 94,458,807 shares outstanding
  11. MarketScreener — Pale Fire Capital SE completes acquisition of 7.8% Dole stake from Messrs Šenkypl and Barta for $76.2MAgreed 13 Dec 2022, completed 22 Dec 2022 · single source for the 2022 transfer
  12. Benzinga — “Insider Confidence On Display”: Pale Fire Capital SE acquires $12.92M in Dole stock4 Nov 2026 · the consensus framing this entry disputes
  13. Investing.com — Meta director Marc Andreessen sells $12.8M in stock4 Nov 2026 · Form 4 fields: AH Fund VIII as seller, disclaimer, LAMA Community Trust 49,448 shares
  14. Finwire — insider buying and selling on 4 November4 Nov 2026 · daily tape; also the source of the headline/table contradiction noted in “What repeats”
  15. Finwire — insider trades of the day, 3 November3 Nov 2026 · comparative daily tape
  16. Seeking Alpha — PACS Group Q3 2026 results presentation4 Nov 2026 · earnings date and content
  17. MarketBeat — PACS Group reaches new 12-month high after strong earnings4 Nov 2026 · price path through Murray’s selling window
  18. GuruFocus — PACS Group reports Q3 2026 revenue growth, raises full-year guidance4 Nov 2026
  19. Family Capital — Larry Ellison’s pledged Oracle shares and the Warner dealOracle proxy filed 25 Sept 2026, pledges as of 21 Sept · checked, outside this window
  20. Billionaires.Africa — Peter Thiel revealed as buyer of Casa Encantada at $130M29 Sept 2026 · checked, outside this window