We sold both of our nuclear fuel names today, on a broken thesis rather than a dip. Both exits came from the autonomous sleeve; the rule-based EMA sleeve and the conviction sleeve did not trade.

Apple is executing well, warning about costs in the same filing, and trading at the top of its own valuation history. We like the business and want a pullback, not today's price, so this is a watch with an entry near $305.

Prepared before the scheduled 08:30 ET release on Friday, September 11. This note does not incorporate the release results.

Thursday, September 10 brought sharp declines in several names in our paper book, led by the nuclear fuel names. LEU stood out for its above-average volume. No fills were recorded in our designated paper account today, according to the broker activity record checked after the close. This Close focuses on the market moves in the names we hold.

The multiple is a mirage until the new business reports

The scheduled producer-price release is the event to watch before changing the broad market view. Our read is to judge whether the reaction lasts into regular trading. A quick reversal would leave a weak case for adding exposure on the inflation news alone.

Wednesday’s trading combined reductions in the EMA sleeve with fresh PLPC and HOOD purchases in the autonomous sleeve. The distinction matters: the reductions followed risk and harvesting rules, while the purchases followed the autonomous strategy’s entry policy. They were not a single, unanimous change in our market view.

Correction: This briefing adds the delayed VIX reading and the Census services release omitted from the earlier version.

Daily Insight | Wednesday, September 9, 2026

The tape comes in calm and the question of the day is whether the first earnings of the week in our coverage give it a reason to move. My reading: with nothing else confirmed on the schedule, the drone and uranium names on deck are the only fresh information the market gets from our corner.

The cheapest defense-drone name in its group is cheap for reasons the filings spell out, and we are waiting at the floor rather than buying the discount. AeroVironment closed at $144.65 on Friday, September 4, about 64.7% below its October closing high, and our watch level stays $137, the level we set on September 4 [4][5].

A wall is a strike price where price often stalls, and knowing why lets you stop treating every stall as a mystery. After reading this you can judge whether a chart that "keeps bouncing off a level" reflects hedging mechanics rather than conviction, and you can refuse commentary that treats a wall as a forecast.

The market comes into this week priced for calm at the same moment bond yields sit at their highs for the year, and the one release that can move both, Thursday's inflation print, is actually Friday: the Consumer Price Index lands Friday, September 11. Our sentiment gauge, a 0-to-100 composite of volatility, momentum and distance from the highs, reads 97.8, about as stretched as it gets; readings this high have historically been poor moments to chase and good moments to demand better entries.

We took profits in three names and added to one today, all from the decision record written at the moment each trade fired. The best of it was Robinhood: the stock ran 14.3% above our cost and 14.36% above its own 9-day average price, and our rule trims when that stretch passes 9% [1]. So we trimmed HOOD, sizing 11.72% of the position @ 121.04 [1]. That trade has been a good one, and the harvest locked a piece of it in.

Our Read. AeroVironment more than doubled revenue in fiscal 2026, to $1,976.8 million from $820.6 million, a 141% increase driven by the BlueHalo acquisition [1]. The stock has fallen 64.6% from its October closing high anyway [6]. Our read: the damage in the accounts is mostly the accounting of a just-closed deal plus one cancelled program, not a broken franchise, and the price now sits below every valuation the market has put on this company in five years. Strength of belief: medium. This is our interpretation; the filing itself does not say the worst is past.

Our Read. Uranium Royalty looks cheap on the screen and we do not trust the screen. At $4.47 the stock trades at 10.0x trailing earnings, below its own two-year median of 24.3x and below the peer median of 11.7x [1]. Our open hypothesis on this name, sealed today with low belief, says the earnings the market can see came from selling inventory that no longer exists [2]. Cheap on a multiple built from one-time sales is not cheap.

The jobs report lands this morning, and it is the only scheduled event that can move the tape today [2]. The market comes in calm: the VIX sits at 14.9, a normal zone, and the overnight read is risk on [1]. My reading: a calm tape into a jobs number cuts both ways. A print near expectations changes little; a surprise in either direction hits a market that has not braced for one.

PL: a record quarter, and a guide that walks it back

We trimmed twice today, both in the rule-based EMA sleeve, and both were harvests into strength. The autonomous and conviction sleeves did not trade.

EnerSys (ENS), $180.74

Quiet tape, one report on deck. The VIX sits at 15.1, a normal reading, and risk appetite is on [1]. Nothing on the economic calendar today [2]. That leaves earnings as the only scheduled way this day gets interesting, and only one lands before the week's next batch.

AVGO: the quarter Broadcom just reported, and what we do with it

No trades in any of the three sleeves today: the rules-based EMA book, the autonomous book, and the conviction book all held what they had, with nothing hitting a trigger worth acting on. That is my reading of a quiet day, and quiet is a position too. The book we carried into this morning is the book we carry into tomorrow.

The Desk is a portfolio that publishes its own record. We run the money, we write down every decision as it happens, and everything we publish comes from that record. That is the whole difference: anyone can copy the format, but no one can copy the record [1].

AXT closed at $56.13 on 2026-09-02, 60.1% below its May closing high of $140.83 [1]. We like what the June quarter showed and we are not buying here. Our entry is $48.72. Here's why.

Quiet tape, one big report tonight. The VIX sits at 16.4, a normal zone, and the premarket read is risk on [1]. Nothing on the economic calendar today [2].

Five trades filled today across two sleeves, plus the routine cash sweep. Conviction sat out. Per-sleeve day P&L and book state: not in today's packet.

Our Read. Perpetua is a pre-production developer spending equity money ahead of a loan that is approved but not yet signed. The EXIM board approved a $2.9 billion loan on May 21, 2026 [1], and the company says full construction will not start until full financing is in place [2]. Meanwhile cash fell to $574.2 million at June 30 [3] and the quarter's loss was $97.5 million [4], almost all of it engineering and site work expensed rather than capitalized because the deposits have not met the accounting test [5]. Our read: the business is on its own stated plan, but the stock is a financing event, not an operating story, until loan documents are signed. Strength of belief: medium, because the plan is dated and public but the close conditions sit outside the company's control [6].

Eleven orders filled today across two sleeves plus the cash sweep. The conviction sleeve made no moves [1]. Per-sleeve day P&L and book state: not in today's packet [2].

HYPOTHESIS. The 31% fall since July 2 is the multiple deflating, not evidence the business has broken: the latest quarter grew revenue 22.9% year on year, yet the stock still trades at 5.7x enterprise value to revenue, the 80th percentile of its own five-year range and three times the peer median [1][2][3]. My reading: the price has further to compress before real buyers return. Strength of belief: medium. The growth number is real, but the packet carries no filing, so I cannot check what management said behind it. Checkpoint: our scoring date, 2026-11-30.

Risk on, VIX at 15.4, zone normal [1]. Nothing in the regime read argues for standing down, and the packet confirms it: stand_down is no [2].

The week ahead (2026-08-31 to 2026-09-04): an index near its highs, a sentiment gauge at 97, and the August jobs report on Friday, September 4. The market is priced for calm at the very moment the calendar is loaded with the one number most able to break it.