Receipt Thursday, closes August 1 through August 7, 2026
I closed a SpaceX bear put spread at -19.88% of premium the same fortnight the company's numbers confirmed my AI-capex call. Both are true and I am not netting them.
One name gave me a correct read and a losing trade in the same fortnight, and both are on the tape.
I closed a bear put spread on SpaceX on 8/6 for -19.88% of the premium I paid. The setup was that a lockup releasing 911.5 million shares, over 140% of the float, would find no bid. The shares went up anyway. I had already written on 8/7 that it looked like a short squeeze in the making, and I was on the wrong side of the one I was describing. I cut it two days before expiry with premium still in it rather than hold the lottery ticket.
The same company’s Q2 numbers confirmed the AI-capex funding call I published on 7/22. That call is about how this buildout is being paid for, and it was right. The trade was an expression of a different question and it was wrong. Both are true. I am not netting them against each other and neither should you.
The energy book had a bad week. Nearly everything else had a good one. Crude stopped out for -8.46%, Brent for about -5.2%, wheat cut early for -1.87%, and a box long closed for -3.49%. Against that: a bank position took half off at +31.2%, a biotech took half off at roughly +45.6% on the original call, and the last two covered calls in the retired program both closed at target.
Cutoff is Friday August 7. Anything closed after that reports next week.
Setup-by-setup review
SpaceX bear put spread: -19.88% on premium
Opened 8/5, closed 8/6, cut ahead of its expiry. -19.88% of the premium spent. That percentage is of the debit, not of the account. A defined-risk spread’s percentage is computed on what you paid for it, so an unqualified -19.88% would read as a drawdown roughly two orders of magnitude larger than it was.
The read was that a lockup releasing 911.5 million shares, over 140% of float, would overwhelm the bid. It did not. The stock rose into it. My own note on 8/7 called it a short squeeze in the making, which is an uncomfortable thing to write while holding the other side, so I will say it plainly: I described the move that was beating me and stayed in it for another session before cutting.
Cutting a losing defined-risk position before expiry is a management decision, not a stop being hit. The spread had two days left and residual value, and I took what was left rather than hold for the tail. Strikes, structure and expiry stay in the members’ channel where they were posted live.
The AI-capex funding call: confirmed, and it is a separate thing entirely
Published 7/22. The argument is that the AI buildout has moved from being financed out of cash flow to being financed with paper, and that the risk concentrates in the names doing the most issuing. SpaceX’s Q2 numbers confirmed it.
The supporting evidence has kept arriving since. Amazon, Alphabet and Meta have sold nearly $170 billion of investment-grade bonds in 2026, dwarfing last year’s total tech issuance, with Alphabet adding $25 billion. On 8/11 Nvidia announced a financing partnership with Apollo, Blackstone, a BlackRock unit, Brookfield, Goldman and KKR to fund up to $500 billion of customer purchases, with its CEO explicitly framing the structure as an answer to concerns about circular financing. Two of the six banks involved said publicly they were unconvinced.
There is no position behind this call and there was not one at any point. It is a published read with a date on it.
Crude: -8.46%, stopped 8/4
Long from 7/29, stopped 8/4, -8.46%. The stop was set at entry time and honoured where it was set. This was a routine daily-timeframe entry in a range, sized for a wide stop, not a regime call, and I said so at the time.
A wide stop means a smaller position for the same risk budget. An 8.46% move against a wide-stop trade is not an 8.46% dent in the account, and anyone reading a percentage without knowing the sizing convention behind it is reading the wrong number.
Brent: about -5.2%, stopped 8/2
A tactical chart trade opened 7/31 and stopped 8/2, on a session that ran roughly 8% peak to trough. Entry sat just above the day’s high and the low ran well below the stop. Pre-set, honoured, small.
Wheat: -1.87%, and this one is a process receipt
Short from 7/31, closed 8/5 at -1.87%. The published stop was 686.2 and it never traded there.
In my own words at the time: I took it at about a third of the risk I had advertised, well before the stop, because I would rather cut a thesis that stops behaving than wait to be proven wrong at the full stop.
That is the part worth publishing. The loss is small enough to be uninteresting on its own. What it shows is a position being reduced on the thesis degrading rather than on the price reaching a number, which is a decision you can only see if someone shows you the losses.
HMH: -3.49%, and the same discipline again
Closed 8/6 at 19.06 from a 19.75 entry, -3.49%, on a box long opened 7/10. Cut well above the stop, discretionary exit, second time in two days I took a loss early rather than defend it to the level.
BSP: +31.2% on the half
Half came off 8/5 at 44.40 from a 33.84 entry, +31.2%, with the stop moved to entry on the remainder. Roughly a week’s hold on the half.
The entry and stop were posted to the members’ channel when the trade opened, not after it worked.
ELOX: roughly +45.6% on the half, scored on the original call
Half came off 8/3 with the stock trading at 18.20, against the original 7/15 entry at 12.50. That is +45.6% scored on the original call, a roughly three-week hold, with the stop moved to entry on the remainder.
I added on 7/31, so the whole position made less than that. The 45.6% is what the original call was worth.
Covered calls: the last two closes of a retired program
The BTC contract closed 8/1 at +26.1% on premium. The ETH contract closed 8/3 at +38.3% on premium, which on the underlying works out to a 7.28% APR pace, net of fees, across a 3.33-day hold.
That APR is net. I mention it because the gross version of the same number is about 46% higher, and quoting the gross figure would overstate what the program actually earned. The honest number is the one after fees.
I retired the covered-call signal program on 8/5, inside this window. The reasoning was a call on the regime ahead rather than on the results: selling calls into a bullish stretch caps you exactly where you want to be uncapped, and the premium does not pay for what you give up. The results stand exactly as posted and I keep citing them. The replacement went live on 8/6.
ARXS: +13.1%, closed 8/3
Long from 7/30 at 48.71 with the stop published at 40.6228 in the same alert, and out 8/3 with the stock trading at 55.11. +13.1% on a four-day hold, full position, no scaling.
A bot signal rather than a discretionary call. Entry, stop, size and exit rule all went out before the trade did anything.
The stop sat 16.6% below entry, which sounds enormous until you put it against the 3% position size: the whole trade risked about half a percent of the account. Wide stop, small size, same risk budget as anything else.
Other closes
Sugar: +2.85% on the half. Half banked 8/6 at 15.175 from a 14.755 entry, stop moved to entry on the runner.
Cotton: +2.74% on the half. Half banked 8/6 at 83.713 from an 81.481 entry, same treatment.
Both are chart trades. I am long cotton on the chart, not on the weather, and those are two different reasons that I do not want merged.
What’s still live
No P&L marked, no booking history.
· Copper. Long from 8/6 on a monthly and weekly signal together, the biggest timeframe combination I have called in this book. What would change my mind about the character of the move: Chinese demand data confirming broadly, which would make it a growth story rather than a cost story, or the funding channel for AI capex closing, which takes the bid away.
· EURUSD. Long from 8/7 on the daily uptrend, behind the weekly call I published publicly on 8/2. The invalidation at 1.13706 was published in advance and is still the invalidation.
· Sugar and cotton runners. Both riding at break-even risk after the 8/6 half exits. Risk-free is not the same as closed and neither is banked.
· ELOX runner. Still on at a break-even stop. It reported on 8/11, after this window: $66 million raised, a Nasdaq uplisting, $62 million of cash and a runway into mid-2028, with the net loss wider year over year.
· Energy equity. XLE and XOM as the crack-spread expression. Worth flagging against the book: on 8/10, XLE closed up 0.97% on a day WTI ran 5%, after being up nearly 3% intraday. Second session running where the equity leg did not track the barrel. Two sessions is not a regime, but it is the leg I actually hold.
· AAPL. Margin long carried, stop raised to 289.36, which I posted publicly.
· PMCSP. Wound down as of 8/5. No new positions; the system exits the open spreads as its own signals fire, so it keeps producing closes.
Members got the exact levels while these trades were live, not a week later.
Crypto accepted. Pay annually in crypto and save: Notes $260/yr · Laboratory $800/yr.
Or try Laboratory free for a month
Signals only, not investment advice.