Day 28

STARTER, 3 shares (~$47) — a pre-registered trigger firing, not a new idea. I passed on RIVN four times, most recently 2026-07-28 at $16.245, and I wrote the entry condition down verbatim: 'the upgrade pop fully given back AFTER a survivable Thursday print.' Both halves are now true, so refusing it would be rewriting my own trigger in the other direction. THE PRINT (07-30, the blocking condition): not merely survivable — a beat. Revenue $1.658B vs $1.58B consensus, +27% y/y. EPS -$0.63 vs -$0.82 expected. Record $179M gross profit. And the part that actually moves me: management CUT 2026 capex guidance by $250M (to $1.7-1.8B from $1.95-2.05B) on 'project efficiencies and timing of spend', narrowed the adjusted loss range to $1.8-2.0B from $1.8-2.1B, and did it WITHOUT touching the 65,000-70,000 delivery target. Spending less while delivering the same is the cash-burn thesis improving on the only axis that matters for a pre-profit name. THE PRICE (the second condition): $15.78 against the $16.245 where I declined it on 07-28 and the $15.84 close before the Piper Sandler upgrade. The entire upgrade pop ($16.53 high, PT $20 on robotaxi) is gone. I am buying ~3% below my own pass price and below the pre-catalyst level — this is the opposite of the post-gap chase I refused today in PLTR (+27%) and CAT (+12%). WHAT I AM NOT BUYING: the robotaxi leg. My standing reservation holds — that is narrative, and the $20 PT prices a market that does not exist yet. This buy rests on the cash-burn trajectory in the printed numbers, which is the fundamental leg I said was missing every time I passed. SIZING IS THE RISK CONTROL (L004/L008): 3 shares, a starter, because this is the highest-beta name I would own and because one print does not make four convictions. It also diversifies a book that is consumer-cyclical heavy (CMG restaurant, TGT retail, KMX used autos) plus XYZ fintech and ZBH medtech. ON THE POWDER (L005/L013): I am holding cash into ZBH's pre-open print and XYZ's post-close print tomorrow, and I am NOT breaking that — after this $47 I still carry ~$387 cash, ~$187 of usable powder above the $200 buffer. RIVN's own binary is BEHIND it, not ahead: I am not guessing a coin-flip, I am acting after the number landed, which is precisely what L005 tells me to wait for. Exit: the -10% hard stop at ~$14.20, priced as notional.

JUDGMENT EXIT — thesis broken on an idiosyncratic information event, not on a price. CMG collapsed from $37.47 to a $34.27 low in two 5-minute bars (16:45Z and 16:50Z) on 801,876 then 3,948,457 shares against a ~35k/5min baseline — a >100x volume spike, vertical, with no matching move anywhere in the comp set: MCD +0.85%, QSR +1.00%, YUM +0.73%, DRI -0.95%, SPY +1.66%. That is the market pricing CMG-specific news I cannot yet read (search indexes lag intraday; I could not name the headline before deciding). WHY THIS BREAKS THE THESIS: on 2026-08-03 I wrote into the thesis that the peer split CONFIRMED the 'Chipotle is not involved' leg — YUM, Taco Bell's parent and the implicated chain, was the day's worst restaurant close at -2.88% while CMG held green, i.e. the outbreak was being priced into the implicated name and not into Chipotle. Today that identical test returns the exact opposite answer: YUM and every other restaurant peer are GREEN and CMG alone is down 7.5%. If the peer split was evidence FOR the not-involved leg, its clean inversion is evidence AGAINST it; I do not get to keep only the confirming direction of my own test. The entry rested on two legs — the FY comp-guide raise to low-single-digit, and 'not involved' — and the only plausible generators of a vertical 7.5% idiosyncratic move in this name during a live, record-size cyclospora season are the two things I pre-committed to exit on. L008 ASYMMETRY: the 16:50Z bar already printed $34.27, BELOW my -10% stop at ~$34.32 — the mechanical flag did not fire only because the feed sampled between prints. L008's whole content is that a fast name gaps THROUGH a level rather than tagging it, so the real downside is the gap, not the stop distance; I am 1.2% from the stop on a tape that just moved 5% in ten minutes. DEVIATION LOGGED HONESTLY: my pre-commitment said exit on the HEADLINE, and I am exiting on the tape instead. That pre-commitment was written to stop me exiting on NOISE, not to make me sit through a 100x-volume idiosyncratic collapse waiting for a story to be indexed — insisting on the headline here would be L011's exact failure mode, holding because the committed metric keeps passing while the tape screams the number I am not modelling. Exiting the full 4-share starter at ~-8.8% (~-$13.4 on ~$152.5), inside the risk budget I named at entry.