High Risk

Recent Performance — Last 5 Sessions

Today = vs our Oct 19 mark ($1,431.36). Monday’s update was emailed and texted, so this is also the last figure you received; there is no separate emitted basis today. Live marks (Tue, 7:42 MDT, twelve minutes after the open, 9:42 Eastern); the daily move is measured against our Oct 19 (Monday) published mark, which was also emailed and texted, so it is the same figure you received — two bases today rather than three. The direction matters, so read it before the numbers: Monday’s tape kept falling after we marked at 10:01 Eastern, so Monday’s official close landed below our mark. That makes today’s headline, which is measured against our mark, the harsher of the two readings — your own screen measures from that official close and will look better than our number. Both readings are printed, for the book and for every position. October MTD is measured from Wednesday Sep 30’s official month-end close. October MTD vs Sep 30 close ($1,733.06). Monday’s official close was $1,420.40 — 0.77% below our mark — so on the close basis today is -3.1%. Sessions before Jul 14 reflect the prior book; the current holdings were established at the Jul 14 open.

Baseline note — read this one before the numbers. Today’s move is measured against our Oct 19 mark ($1,431.36), which was both published and emailed, so it is the number you actually received. But this book kept falling after we marked at 10:01 Eastern and closed Monday at $1,420.40 — 0.77% below the figure we sent you. Measured from that close the book is -3.1% rather than -3.9%, and the close figure is the one your brokerage screen will show. So our headline is the harsher reading and your screen will look better; we print both and lead with ours. This is the same direction as the orientation Monday’s letter described, so if you read that one the shape of this will be familiar. October MTD is measured against the Sep 30 official month-end close of $1,733 for the current whole-share model. September is locked closed in the monthly table at +19.50%.

Position Scoreboard — Sorted by Portfolio Weight

Today = vs our Oct 19 mark, with the vs-Monday’s-close reading beneath it wherever the two differ. Since Fill = vs SOXL $179.62 / TQQQ $75.19 / MSTR $94.78 / BE $205.15 / NBIS $172.19. Weight = % of unit. Live marks (Tue, 7:42 MDT, twelve minutes after the open, 9:42 Eastern); the daily move is measured against our Oct 19 (Monday) published mark, which was also emailed and texted, so it is the same figure you received — two bases today rather than three. The direction matters, so read it before the numbers: Monday’s tape kept falling after we marked at 10:01 Eastern, so Monday’s official close landed below our mark. That makes today’s headline, which is measured against our mark, the harsher of the two readings — your own screen measures from that official close and will look better than our number. Both readings are printed, for the book and for every position. October MTD is measured from Wednesday Sep 30’s official month-end close.

Portfolio Allocation by Category

Live Track Record — Since January 22, 2026 (Actual)

▲ Current month through today; July, August and September 2026 are locked closed; return measured from the Jan 22, 2026 entry. * Jan–Mar 2026 are estimated partial-period returns carried from our May 2026 updates; they predate our daily archive (begins 15 April 2026) and are shown as estimates, not as-published daily marks. Apr–Sep cells are locked closed from official month-end closes; close-basis restatement pending the anchor review. No backtests, no hypotheticals. Units are price-return; dividends are not accrued.

Calculate Your Position Size

High Risk is marked at $1,376 on the current whole-share model, and the two bases disagree today, so take the worse one first. Against our Oct 19 mark — the figure we emailed and texted you — this book is -3.9%. Against Monday’s official close of $1,420.40 it is -3.1%. The second number is the one your brokerage screen will show and it is the kinder of the two, because this book kept falling after we marked and closed 0.77% below the figure we sent. October is -20.6% month to date and the year is negative at -15.9%. We are leading with our own harsher reading rather than the one that looks better.

The most important thing that happened overnight was not a price. It was an auction that failed. Elice, a South Korean company that builds and operates AI data centres, listed this morning in Seoul. The deal could not have been better subscribed: it priced at 90,500 won, the top of its range, raised about 201 billion won — roughly US$151 million — and the retail tranche was oversubscribed four hundred times, on something like 10.3 trillion won of subscription deposits. It then closed its first day of trading at 61,500 won, down 32%, which is the worst debut on record for a sizable South Korean listing. On the same theme, AirTrunk, an Australian data-centre operator, is delaying its own planned listing. Put those two together and you have the question that has been sitting in CoreWeave’s bonds for a fortnight being asked at the primary market instead: not whether anyone wants AI capacity, but who is going to fund building it and at what price. A deal that is four hundred times covered and still breaks by a third on day one is the market answering that question in public.

And that is why Monday’s rebound has not survived the morning. We measured it rather than describing it. On Monday the Nasdaq Composite rose 1.11% to 26,430.89 and the semiconductor index 1.6% — the first up session after eight consecutive declines. At our 9:42 Eastern mark today the Nasdaq Composite is 26,201, down 0.87%, the semiconductor index is down 2.1%, and the semiconductor ETF at 490.92 is below where it closed last Friday. The entire bounce is gone and a little more with it, in one morning. Korea tells the same story harder: the Kospi closed 5,552.74, which is below the 5,554.22 it printed on the day last week that Seoul called Black Thursday, and we checked it against five years of daily closes — the last session to finish lower than this was April 7. SK Hynix fell about 4%. Against that backdrop SOXL is -7.3% against our Monday mark, -6.5% against Monday’s close and -48.6% since entry, with TQQQ -2.7% against our Monday mark, -3.0% against Monday’s close. The leverage arithmetic is the part to hold onto rather than any single session: a 3x daily-reset fund does not lose three times the index over a stretch like this one. It resets its exposure every afternoon and re-buys at each new lower level, so through a sustained choppy decline it loses considerably more than three times, and it needs a far larger rally than the one it fell on to recover.

The credit side, which is where this story actually lives, has not healed at all. CoreWeave is at $57.90 at our mark, down 5.3% from Monday’s $61.12. Its 2032 notes, issued this year with a 9.625% coupon, still change hands to yield about 14.9% — off the peak above 15% last week, which is a long way from repaired — its five-year credit default swaps are at distressed levels, and it carries more than $51 billion of debt and obligations against a business that has to keep borrowing to grow. The framing worth repeating, because it was exactly right: through Monday the reporting was that equity investors were ready to move on from the AI rout and credit investors were not. Today equity came back to credit’s view rather than the other way round. That is the read-across to NBIS, which is -3.9% against our Monday mark, -5.1% against Monday’s close and -6.2% from our fill: it funds GPU capacity with capital-markets money against forward contracts, which is the same shape of business, so it reprices on another company’s bonds and another country’s IPO whether or not anything has happened at Nebius. Nothing company-specific has.

Thursday is the date that matters most to this page, and the two halves of it are not equally settled. BE is -5.2% against our Monday mark, -4.9% against Monday’s close, and Bloom Energy reports third-quarter results after the close on Thursday, October 22 — the company confirmed that date itself in a press release on October 12, with the call at 5:00pm Eastern, so its reaction lands in Friday’s session. Together BE and NBIS are 27% of this book, and they are counterparties to one another, since one sells the other power for AI data centres. The Nebius half we still cannot confirm, and we would rather say so again than let Bloom’s certainty cover for it. Market-data providers carry Nebius for the same Thursday before the open, which matches its own standing pattern — last quarter was an 8:00am Eastern webcast — but Nebius has not announced the date itself: there is no filing, its investor-events page lists nothing upcoming, and at least one provider still carries an estimate of November 10. We re-checked all of that this morning and none of it has changed since Monday. So if both land as expected this is roughly a quarter of the portfolio repricing across two sessions rather than one afternoon, which is the less concentrated version of the risk — but the Nebius leg is an expectation, not a confirmation, and you should hold it that way.

One genuinely quiet thing, and it is a change from last week. Bitcoin is at about $81,529 at our mark against $81,835 at Monday’s equity close, down 0.4%, and MSTR is -2.7% against our Monday mark, -0.6% against Monday’s close. Those two numbers are close to each other, which is the point: on Friday MSTR fell 9.6% on a day bitcoin fell 0.15%, and we told you then that the market was repricing the premium it would pay for bitcoin held inside a company that may lose index sponsorship. Today it is not doing that. MSTR is simply moving with the asset it holds, and on the worst morning the rest of this book has had in a week it is the position that has barely moved. Its weight is 39.3% of this unit against 38.8% on Monday — weight moving through price, not through any trade. The index position is unchanged and still dated: the affected-company list publishes November 11, any exclusion takes effect only after the close on November 30, and an existing member is removed only if it fails the screen in each of its two most recent annual filing periods. We hold four shares, the review of trimming to three is closed, and no trade was made.

Bottom line: High Risk is at $1,376 a unit — -3.9% against our Oct 19 mark, -3.1% against Monday’s official close, -20.6% for October and -15.9% for the year. On the performance chain this is the lowest this book has marked since August 19; we checked that against every session of 2026 rather than reaching for a superlative, and it is not a low for the year — the book traded below here on 37 sessions in 2026 and its actual low is well under today. Nothing here is a change of view and nothing was traded. The structure that produced a 19.5% September and this October is the same structure, and it is worth seeing written down: 33% in two 3x daily-reset funds, 39% in a bitcoin treasury company carrying a dated index risk, and 27% in two AI-infrastructure counterparties that both report this Thursday, one before the open and one after the close. What we are watching, in order of how much it can move this page: Bloom Energy after Thursday’s close on October 22, with Nebius expected the same morning though still not confirmed by the company; CoreWeave’s bond yield and credit default swaps as the financing-side read on the entire AI complex, and whether any new AI issuer can price a deal at all after this morning in Seoul; the MSCI affected-company list on November 11 and the November 30 effective date, with MSTR at 39.3%; the Federal Reserve on October 27–28; and the midterms on November 3.

Portfolio Builder | Market Update | Tuesday, October 20, 2026

Trade Ledger — Entry & Exit History

Every position we open and close, newest first. Realized P/L on closes is marked against the original entry price.

Entry prices verified against the Jan 22, 2026 cost basis. “Open” = still held today. An entry marked “Closed <date>” was exited in that rotation — its realized result is shown on the CLOSE row for that date. This ledger builds our public track record going forward.