Discovery › watchlist
Watchlist scan
S&P 500 7,705.68 (−0.76% today, after a new closing high of 7,764.70 on 9/21) · VIX 15.33 · Fed funds 3.75–4.00% after the 2026-09-16 hike · 10yr ~4.96%
145 sleeve names swept. 47 sit in or below their own entry zone · 29 sit above their own trim or fair-value ceiling · 8 have moved >15% since their last analysis.
🔴 The headline: the hike landed and the market went up
Yesterday's regime note called the 2026-09-16 hike "the largest flagged risk in this series, now a fact." That was right about the event and incomplete about the reaction. The S&P made a new closing high two sessions after the first Fed hike since 2023, and VIX has fallen from 16.44 (9/10) to 15.33.
Read the distinction carefully, because it decides what to do:
- Rate-duration names did re-price and the pain is real and specific: VICI −27.9% off its high, O −18.3%, MPLX above trim, the whole 7–13% yield screen offering nothing but duration.
- The broad market did not. There is no rate-driven de-rating to buy. The 47 in-zone names are in zone because of company-specific guide-downs and a long sector de-rate, not because a macro event cracked the index.
So the "buy the crash" framing is wrong this month. The 8/27–8/30 scans called ~19 entry zones "a museum, unreachable without a crash." The museum opened — but name by name, on earnings guides, not all at once on rates. Every in-zone name still has to be underwritten individually.
⭐ ACT NOW
| Ticker | Conv | Price | Fair value | Entry | Discount to FV mid | Catalyst | Recommendation |
|---|---|---|---|---|---|---|---|
| AVGO | 8.3 | $354.66 | $425–500 | $360–420 | −23% | Q4 FY26 print ~Dec; AI rev guided $58B FY26 | 🟢 BUY — now BELOW the entry band |
| BR | 8.5 | $164.78 | $190–230 | $150–180 | −21% | 20th straight dividend raise banked; NYSE proxy-fee review is the risk | 🟢 BUY — highest conviction on the list, mid-band |
| LDOS | 8.0 | $125.96 | $148–173 | $115–135 | −22% | 9.9x fwd, −38.8% off high | 🟢 BUY — mid-band, cheapest high-conviction name |
Why AVGO is the single clearest action
Yesterday's KLAC analysis concluded: "hold AVGO at size, not KLAC — purely on price," with the instruction to fund the AVGO top-up from the INTC trim. Since that was written AVGO has fallen out the bottom of its own entry band ($362.93 → $354.66) while INTC still sits 33% above its fair-value ceiling ($120.11 vs a $60–90 FV).
Both legs of that trade improved in one day. The funding source got more overvalued and the destination got cheaper. Target remains 4.5% AVGO / 3.5% KLAC, combined bucket cap 8.5%.
⚠️ The one caution: CNBC reported 9/18 that Anthropic and OpenAI are chasing smaller datacenter deals, and Anthropic reportedly asked Broadcom to slow capacity deployment. Hock Tan denied it publicly. That is the live bear case and it is not resolved — size the tranche accordingly rather than taking full size at once.
Close behind (in zone, high conviction, no fresh catalyst)
INTU [7.0] $288.15 — in its re-derived $280–320 band and cheaper than yesterday's $295.83. The 9/22 re-analysis already did the work: QuickBooks Free is a funnel, zero of seven break triggers fired, but conviction sits at 7.0 because clean FCF grew +13% not +41.7%. Tranche, do not take full size. · GOOGL [7.0] $338.21 top of its $300–345 band · BKNG [7.0] $156.04 bottom of band at 12.6x fwd · DPZ [7.0] $293.88 now below its $305–335 band, print ~Oct 10.
✂️ SELL SIDE — 29 names above their own ceiling
The reliable breaches (trim renders on a sound basis, conviction ≥6.0):
| Ticker | Conv | Price | Trim level | Over by | Note |
|---|---|---|---|---|---|
| VEEV | 8.5 | $270.54 | 44x ttm ≈ $269 | +0.6% | At trim and at the $271 FV top. The highest-conviction name in breach. |
| SAP | 7.0 | $211.17 | 25x ttm ≈ $192 | +10% | Above trim and above the $207 FV top. Held. |
| ADI | 7.0 | $380.13 | $320 | +19% | 25% above the $305 FV top. File is 2026-07-29 and has no frontmatter. |
| V | 7.0 | $362.50 | 30x ttm ≈ $353 | +2.7% | Inside FV but above trim. Held. |
| QLYS | 7.0 | $185.00 | 23x fwd ≈ $194 | at FV top | 9/22 verdict is already TRIM; trim to 2.7–3.0%. |
| ETN | 7.0 | $438.49 | 34x fwd ≈ $549 | FV top +8% | Above the $405 FV ceiling. |
| ISRG | 6.5 | $396.87 | 45x ttm ≈ $393 | +1.0% | |
| AMAT | 6.5 | $468.96 | $420 | +12% | 27% above the $370 FV top. No frontmatter on file. |
| TGT | 6.0 | $156.95 | 16x ttm ≈ $154 | +1.9% | |
| MEDP | 6.0 | $618.17 | 30x fwd ≈ $579 | +6.8% | |
| QCOM | 5.5 | $195.74 | 19x fwd ≈ $194 | +0.9% | 9/10 verdict already said "do not add, do not trim"; now at the line. |
| INTC | 4.0 | $120.11 | 44x fwd ≈ $91 | +33% | 🔴 Standing TRIM from 9/22. Fund the AVGO buy from this. |
⚠️ Do not act on a rendered trim without checking its basis. The scan found rendered trims below the entry band on O, DOCN, RBRK, RKLB, ARCC, NLCP and LEU — REIT/BDC names where the site derives a dollar trim from GAAP EPS when the correct denominator is AFFO or NAV, plus loss-makers where a forward multiple is meaningless (RKLB 1556x fwd, CRWD 163x, CBRS 161x). These were excluded from the sell list rather than reported as signals. → [[pitfall-multiple-trim-inherits-the-broken-vendor-field]]
Held names in breach: INTC, QLYS, SAP, V, META, CRWD, TEAM, DT, TWLO, AMD. Of these only INTC and QLYS carry an actual TRIM verdict; the rest are "above a level" without a sell call, which is a different and weaker statement.
🚩 Thesis break suspected — FICO
FICO [5.5] — the news is real, the "gap in the file" is not. Corrected before publishing.
The sentiment sweep flagged a 2026-09-09 FHFA action opening VantageScore 4.0 to all
approved lenders as new information absent from the analysis. Checking the file refuted that.
Output/Stocks/Technology/FICO/analyze-2026-09-06.md opens with a section titled "The event
that defines the analysis" describing exactly this: the 2026-09-04 Pulte order approving
"VantageScore 4.0 for all lenders, effective immediately", ending the 50-lender pilot and
FICO's decades-long conforming-mortgage monopoly. The 9/09 item is the same order's news
cycle, not a second step.
📌 Two corrections this produced:
- The watchlist entry reads "ANALYZED 9/10" and tags A2026-09-10. The file is dated
2026-09-06. Fixed below.
- No /analyze-from-before is warranted on this news. The file already prices it.
What IS live: the file's own break trigger is "VantageScore GSE securitization share past
the 9% Sept level" — and the file records share as already >9% of Fannie/Freddie
securitizations by early September. That trigger is sitting exactly on its threshold, with
the first post-ruling Scores/mortgage read at the early-November print (recheck 2026-11-06).
Mortgage is 62% of Scores revenue. Watch the print, not the headlines.
A portfolio-specific passage was removed from the public build.
🧭 Thesis drift
Fired rechecks with no refresh (9 — only 3 are held)
| Ticker | Verdict | Age | Held | Recheck |
|---|---|---|---|---|
| MU | TRIM [4.0] | 50d | beta | fires today; FQ4 print 9/30 into memory prices up 5–7x |
| META | HOLD [6.3] | 27d | self+beta | fires today — Meta Connect |
| NU | WATCH [6.5] | 54d | — | 2026-08-13 |
| KT | WATCH [5.5] | 43d | — | 2026-08-18 |
| FIGR | AVOID [3.5] | 43d | — | 2026-08-13 |
| LEN · CMPX · CNM · BSX | WATCH/ACC | 26–35d | — | passed |
Only MU and META matter — both held, both firing today. The rest are unheld watchlist names at middling conviction.
🔬 The structural finding: 27 tickers are invisible to every scanner
A verdict-bearing report with no YAML frontmatter is indistinguishable from no report at all.
kb.py, due.py and /thesis-drift all read frontmatter; a file without it contributes no
verdict, no conviction, no recheck and no triggers.
9 names whose newest analysis is invisible:
| Ticker | Newest analysis | Weight |
|---|---|---|
| NFLX | 2026-06-04 | 3.26% (self+beta) |
| SNOW | 2026-04-28 | 1.34% |
| PLAB | 2026-07-29 | 1.14% |
| U | 2026-06-11 | 1.11% |
| BABA · PATH · CVLT · GLW · EPAM | Jul 2026 | 0.26–0.85% each |
Plus 18 watchlisted names: TDC · QBTS · UEC · VRRM · CTSH · NVTS · EFX · FDS · OBDC · TSLX · BMY · CDW · PTC · TRI · SNPS · ADI · AMAT · NXPI.
This is the same root cause as three defects already fixed this week —
kb.py findnot searching reports (fixed Aug), a drift report read as current inventory (fixed 9/22), anddue.pysurfacing superseded triggers (fixed today). In every case the tooling reported absence when the truth was invisibility. It is also why yesterday's health report listed 12 held names as "never analysed" — they were analysed; the analyses just cannot be seen.This is a backfill job, not an analysis job, and it is bounded at 27 files. It should run before the next scan, or the next scan will make the same mistake.
🌱 Sleeve vacancies — 8 nominations, and one instructive blank
These are nominations, not finds. The screener sees TTM only; each must clear the
analysis_notes.md §1 bar via /analyze before it earns a watchlist line.
| Ticker | Sleeve | Case | Flag |
|---|---|---|---|
| UI | Evergreen | Revenue $1.93B→$3.27B, FCF $909M, debt $1.14B→$80M, shares flat, −47% off high | ⚠️ float is 4.22M of 60.5M shares — 93% insider-held. Liquidity risk is the whole story |
| YOU | Evergreen | FCF CAGR 36%, net cash, 14.7x fwd, −45% off high | ⚠️ diluted shares +6.0%/yr |
| AMP | Yield-Tomorrow | 9.9x fwd, FCF +24.5% CAGR, shares −4.7%/yr, 16% payout (room to compound = the sleeve test). Does not add rate duration — its sweep spread widens on the hike | ⚠️ correlates with HLNE |
| APTV | Re-Rating | 6.9x fwd, FCF +54% CAGR, shares −6.6%/yr; Graham IV $45.70 ≈ price $44.85 — the only nominee where Graham doesn't laugh | ⚠️ net income −34.8% |
| QNST | Speculative | FCF −$4.7M → $117M, revenue $613M→$1.29B, net cash, 7.7x fwd | small-cap risk |
| PSN | Re-Rating | Revenue +14.9%, NI +35.6%, FCF +25.6% CAGR, 13x fwd, −45% | ⚠️ revenue fell $6.75B→$6.36B latest year, and ACM [6.5] already occupies this slot in the same cycle |
| BRO | Yield-Tomorrow (weak) | Revenue +17.4%, FCF +18.6% CAGR, −33% | ⚠️ debt $4.06B→$7.92B and shares +3.9%/yr, both from one acquisition |
| OPRA | Structural-Risk, not Speculative | FCF +30% CAGR, shares −6.2%/yr, ~zero debt | ⚠️ Kunlun-controlled; 87% payout |
Top 3 to route to /analyze: AMP, APTV, UI.
🎯 Yield-Today produced zero survivors, and that is the correct answer
25 names screened at 7–13% yield; none survived. The screen at a ~5% risk-free surfaces only four preferreds (pure duration), five tankers at a cycle top (FRO +102%), and EM currency risk. The three real businesses all fail §1: ARLP (FCF −9.7% CAGR, payout 1.17) · APAM (FCF −16.4%, payout 0.81) · OWL (shares +15.2%/yr, payout 7.5x, debt doubled — private credit, one week after the first hike since 2023).
The book is already ~13.9% rate-duration and does not need more. A yield screen that returns nothing in a hiking cycle is the screen working, not failing.
Already analysed and passed on — not re-nominated
AMSC (8/17, AVOID [3.5]) · Z/ZG (8/26, WATCH [4.5]) · ARLO (9/10, WATCH [4.5]) ·
NKE (9/10, WATCH [5.0]) · BSX (8/26, WATCH [5.0]) · SNDK (Graveyard 🌱Revisit [3.5]) ·
HUBS (7/27, no verdict) · APP (9/10, ACCUMULATE [6.5] — live; an /analyze-from-before
candidate at most, not a new find).
Also rejected on §1 grounds: POOL (revenue/FCF/NI down three straight years) · AFL (FCF −13% CAGR) · IA (FCF $6.8M against NI $15.6M, cash down to $2.7M) · LECO and FTNT on price alone.
📅 Catalyst calendar — next 30 days
| Date | Ticker | Event |
|---|---|---|
| 09-23 | META | Meta Connect — first Meta Compute pricing/customer detail. Recheck fires today. |
| 09-24 | TSLA | Semi factory inauguration |
| 09-24 | UEC | |
| 09-25 | PDD | EU DSA Board opinion on the Temu remediation plan |
| 09-30 | MU | FQ4 FY26 print — into memory prices up 5–7x and a high bar |
| 09-30 | NKE | Q1 FY27 — gross margin ex the one-time tariff refund |
| 09-30 | CBRS | Lockup tranche (~68M shares Sep 30–Oct 2) |
| 10-01 | LULU | Chip Wilson's third board seat due |
| 10-04 | NU | Brazilian general election |
| 10-05 | INTC | PC CPU price increase — does it hold into Q4 ASPs |
| 10-16 | V | DOJ debit-suit fact discovery closes |
| 10-21 | PEGA | Q3 FY26 print |
| 10-22 | INTC | Q3 FY26 — the dated test of a TRIM sitting 33% above fair value |
| 10-22 | NOW | Q3 FY26 — subscription gross margin vs 80.5% |
| 10-28 | FOMC | ~60% odds of a second hike |
🔧 due.py was fixed during this scan. It was surfacing triggers from superseded reports,
so the calendar showed INTC twice under two different convictions (TRIM [4.5] and [4.0]) and
listed AMZN's FTC trial on 2026-10-13 when a newer file says it was delayed to 2027-02-09.
Now filtered to each ticker's newest report; 23 rows → 19. Selftest added.
Recommendations
Top 3 act-now
- AVGO — buy, below its entry band, funded from the INTC trim. Both legs improved in a day. Tranche it; the Anthropic capacity-pace story is unresolved.
- BR — buy. Highest conviction on the entire list [8.5], mid-band, 14.2x fwd, −31% off high, with the 20th consecutive dividend raise already banked.
- LDOS — buy. [8.0] at 9.9x forward, −38.8% off high, mid-band.
Re-analyse
- ~~FICO~~ — withdrawn. The 9/06 file already covers the FHFA ruling; the “new” 9/09 item is the same order’s news cycle. Watch the early-Nov print instead — its break trigger (VantageScore GSE share past 9%) is already at threshold.
- MU and META — both held, both rechecks firing today, both with prints/events landing now.
- ORCL — not a break, but Q1 FY27 showed capex $28.5B against $8.5B a year ago on RPO of
$664B. A clean instance of
pattern-ai-build-inflates-earnings-while-destroying-fcf.
Route to /analyze (new)
AMP (Yield-Tomorrow, and it does not add rate duration) · APTV (Re-Rating) · UI (Evergreen, but resolve the 93%-insider float first).
Remove from the watchlist
None this pass. No name has both a broken thesis and no remaining optionality. FICO is the closest and it earns a re-analysis, not a removal.
Housekeeping
Backfill frontmatter on the 27 invisible reports — 9 held, 18 watchlisted — before the next scan. Until that runs, every drift and due scan will keep reporting those names as unanalysed.
Methodology
- Knowledge check first.
Knowledge/INDEX.mdread;kb.py findused per ticker to take the newest verdict-bearing report. The six names re-analysed 2026-09-22 (INTC, KLAC, QLYS, INTU, NVO, VICI) were picked up at their current verdicts. - Three parallel scouts — valuation (145 quotes, batched), sentiment (batched WebSearch by
sector), sleeve vacancy (5 screens + the
Output/INDEX.mdcoverage check). Raw output inOutput/_archive/2026-09/_scan-{valuation,sentiment,sleeves}-2026-09-23.md(archived as raws — provenance, excluded from the index and the site). - Drift computed locally from each ticker's newest verdict-bearing report, ranked by age × conviction × held.
- The coverage grep in
.claude/commands/watchlist-scan.mdwas verified working — a scout reported it malformed; direct testing showedgrep -E "^\| $t \|" Output/INDEX.mdcorrectly returns coverage for NKE and AMSC and NOT COVERED for the eight nominations. No change made. - Prices: live at generation, 2026-09-23. All 145 tickers returned a price; no ETF gap this run.
- Not tested: the 27 frontmatter-less reports' verdicts were read from
Positions.mdprose rather than frontmatter, which is lower confidence — flagged per name above.