The Direxion Daily Semiconductor Bear 3X Shares executed a one-for-ten reverse split at today’s open. It touches only the Harvest Shadow, and the accounting is built to absorb it without distortion. The rest of the session was a grind.

Dashboard — July 15, 2026 Settled Close

All dashboard levels are July 15, 2026 New York settled closes from a single source (yfinance batch OHLC); the Yahoo v8 chart API did not confirm for this historical date, so this edition is single-source by exception. The distillate crack spread is front-month ULSD times forty-two, less front-month WTI, in dollars per barrel. Day changes are measured against the July 14 settled close.

I. The Reverse Split, and a Market That Would Not Break

The Direxion Daily Semiconductor Bear 3X Shares (“SOXS”) executed a one-for-ten reverse split effective at today’s open. It is worth being precise about what that does and does not touch. SOXS appears in this book only through the two closed Harvest Shadow pairs, P35 and P37, so the split is an accounting event, not a position event. Because the price series is now back-adjusted for the split, every SOXS mark in the Shadow is restated to the pre-split scale to preserve continuity with the v5 inception fills; the share counts — 16,051 and 40,733 — are unchanged. The reproduction check holds to the cent against the June 30 exit of $3.24. This is housekeeping done in daylight rather than buried.

The rest of the tape was quietly constructive and, for the book, quietly expensive. The Standard & Poor’s 500 added 0.38 percent to 7,572.40 and the Nasdaq Composite added 0.62 percent, a second straight session of the semiconductor complex firming. That is the move that cost the book: the short technology legs that have carried the winners all week gave a little back. The CBOE Volatility Index fell another 5.03 percent to 15.67, the lowest reading of the entire escalation, which tells you the fear bid that spiked on the strait declaration Monday has now fully unwound. A market that will not break on a strait closure, a muted double catalyst, and a semiconductor-bear reverse split in the same week is a market that has decided the war is priced.

II. The Crack Spread Narrows, the Curve Rallies

Two things moved under the surface that matter more than the index prints. First, the distillate crack spread narrowed $3.03 to $86.23 a barrel as crude held near $79.60 and the refined-product premium came in. That is the first real give-back in the margin that has been the book’s central thesis since the escalation began, and it is worth watching rather than dismissing: the refiner pairs are long that spread in all but name, and a sustained compression would take the air out of P40 and P41 faster than any move in the barrel itself.

Second, the curve rallied. The ten-year par yield fell four basis points to 4.54 percent, and gold slipped $17 to $4,044. For once this is the ordinary reaction rather than the vigilante one: a calm tape, a falling volatility index, and a bid to the long bond is the market relaxing, not the market pricing inflation. The register this publication has kept since May is not contradicted by a quiet day — it is simply not being tested. The test comes when the next supply shock lands, and whether the long end sells then is the whole question.

III. The Pair Book

Twenty-six active pairs per the v6 canonical book of record. Day is the July 14 to July 15, 2026 settled move; QTD is measured from the June 30 settled close per the period-baseline rule; pairs marked † (P39, P40, P41) were opened during the quarter and their QTD is inception-to-date. Exact inception share counts throughout; zero notional approximation. Ranked by unrealized profit and loss.

The book lost −$12,951.12 on the session, and it lost it in a way that is worth understanding rather than lamenting. The damage was concentrated in the refiner and software winners giving back, not in the losers getting worse. P40 (long Valero against short Apple) cost $7,278 as Apple rallied 3.9 percent into the firming technology tape and Valero’s crude-driven premium eased; P18 (long GitLab against short Atlassian) gave back $5,481 of its outsized run. Against that, the semiconductor-short structures did their job on the day the complex actually rose: P39 (long Broadcom against short Intel) added $9,576 as Intel fell another 26 percent from entry, and P10 (long BlackRock against short the financials fund) added $6,497. This is the book’s character in one session — the winners and the hedges pulling against each other, the net a small loss on a day the index went up. Quarter-to-date the book stands at +$91,818.51; active unrealized is −$100,045.90 against the frozen realized register of +$487,933.25, for an inception-to-date result of $387,887.35. The book is not being carried by its open positions. It is being carried by what it closed on June 30.

IV. Harvest Shadow — the Edge Stands at $326,564.55

The standing counterfactual: the ten pairs closed at the June 30 settled-close sweep, marked forward as though the harvest had never happened, exits frozen at their actual June 30 prints. Through the July 15 close the harvest decision is +$326,564.55 better than holding would have been — a fresh weekly high for the edge even on a day the active book lost money, because the Corning-short structures that anchored the harvest keep compounding the advantage. P1 alone (long Corning against short Microsoft) realized $245,728 at the close and would be worth only $140,391 on the hold path today, a $105,337 edge from that single decision. The one place the harvest looks worse than holding is P19 (long AMD against short the Korea fund), now within nine dollars of break-even between the two paths — a reminder that the Shadow is an honest ledger, not a victory lap.

V. What to Watch

Whether the semiconductor complex firms now that the leveraged-inverse decay is fully out of the active book. Whether crude settles into a range after the Monday spike. And whether the refiner pairs hold their gains into the back half of the week.