Phase Space AI

05 Trade Construction

Intuitive Surgical [ISRG]

Intuitive Surgical [ISRG] — Task 5: Trade Construction & Risk Management

investment-memo v1.6.0 · analysis dated 27–28 July 2026, migrated onto the Criteria framework 2026-07-29 Price $361.96 (scan) / $361.80 (Alpaca close in the body) · Archetype COMPOUNDER

This document no longer states a position verdict

Under v1.6.0 the memo scores every Criteria and blocks on none of them. It emits an analysis, not a position. This document previously carried INVESTMENT DECISION: WATCHLIST (long-biased) in three separate blocks and named a binding Criteria in each. All three are removed. The evidence that produced them is unchanged and retained.

CRITERIA                        TYPE       RESULT
  Quality Criteria              BINDING    PASS   — 8/9 F-score, clean accruals, rising gross profitability,
                                                    upward estimate revisions
  Valuation Criteria            BINDING    FAIL   — implied path requires 26.0% revenue CAGR vs 17.4%
                                                    demonstrated.  MARGIN -8.6pp
                                                    *** REVERSES to +0.7pp and PASS once ISRG's own 29.3%
                                                        operating margin replaces the scan's 20.0% constant.
                                                        See 03_Valuation_and_Factor_Scorecard.md s.0.1. ***
  Liquidity Criteria            BINDING    PASS equity / FAIL options at current IV
  Catalyst Criteria             MEASURED   PASS (weakened) — Q3-2026 call ~20 Oct 2026 (est.)
  Momentum Criteria             MEASURED   12-1 -17.8%, 18.7th cross-sectional percentile (quintile 1).
                                           ENTRY TIMING ONLY. It does not and cannot block this name.
                                           *** Previously recorded as FAIL and used to support the
                                               classification. That use is retired. ***
  Peer Spread Criteria          MEASURED   PASS   — ISRG at the 0.3rd percentile of its own 6.5-year P/S
                                                    history; peer set MDT / BSX / SYK / EW
  Downside Criteria             MEASURED   PASS   — permanent-loss case and named cause, s.5.4
  Short Mechanism Criteria      MEASURED   FAIL   — clean accruals, 8/9 F-score, rising gross profitability,
                                                    upward revisions. No short mechanism. Scored; acts on nothing.
  Consensus Criteria            MEASURED   PASS   — house FY2027 revenue +0.22% vs consensus. Blocks nothing.

VALUATION, TWO HORIZONS (both mandatory)
  Implied-path test (5y reverse DCF):  price requires 26.0% revenue CAGR
                                       demonstrated 17.4%   MARGIN -8.6pp  (+0.7pp on own margin)
                                       exit multiple 29.9x EV/EBIT, basis GROWTH_MATCHED (n=32)
                                       implied compression from 40.2x today: -25.6%
                                       breakeven exit multiple: 42.5x  (29.0x on own margin)
  12-month target (own P/S history):   base $427 (+17.9%); floor case $397 (+9.6%)
                                       up-case anchor: UNIDENTIFIED - no stable recent multiple regime

REFERENCE LEVELS (retained as analysis, not as instructions):
  prior-document conversion level $310
VOLATILITY: MODERATE (35.8% realised) — the input inverse-volatility sizing consumes
VEHICLE FEASIBILITY: outright equity. NOT options at current IV.

---

⟐ TASK 8 INCREMENTAL REFRESH — 28 July 2026

Baseline: Intuitive_Surgical_Memo_Manifest_2026-07-27.json (commit ee4a135), price $357.16.

The baseline's position verdict and binding-gate designation are recorded as history and are not carried forward. v1.6.0 outputs an analysis, not a position.

Scope discipline, stated up front. This refresh sits one trading day after the initiation and eight sessions after the Q2 2026 print (16 July). Almost nothing fundamental could have changed in that window, and this document does not pretend otherwise. What it does add is genuinely new: (i) the mandatory PEAD check, which the initiation did not run and which turns out to be the single most interesting thing about the name right now; (ii) a verbatim trigger audit; (iii) re-computed Valuation Criteria, 5 and 6 on live data with the E[R] reported as a range with flip points per framework item C4; (iv) a re-check of the free independent corpora that back the Vector C and D triggers; and (v) one new SEC filing the initiation could not have seen.

---

R.1 THE PEAD CHECK (mandatory on any earnings refresh) — and the two signals disagree

Post-earnings-announcement drift (Bernard & Thomas 1989) and estimate-revision momentum (Chan, Jegadeesh & Lakonishok 1996) both say a stock drifts in the direction of the surprise. The standard construction uses two signals — standardised unexpected earnings (SUE) and the earnings-announcement abnormal return. On ISRG they point in opposite directions, and that is the finding.

Leg 1 — the surprise (SUE): clearly POSITIVE

Consensus is taken from the Alpha Vantage EARNINGS_ESTIMATES snapshot's 30-days-prior field (≈ 28 June 2026), which is the pre-print expectation. Actuals are from the Q2 2026 8-K Ex-99.1 (16 July 2026).

Operationally the print was strong on every disclosed driver: worldwide procedures +16%, da Vinci +15%, Ion +36%; 468 da Vinci placements vs 395 (of which 246 dV5 vs 180); installed base 11,710, +12%. Non-GAAP operating income $1,218m vs $947m. The beat is real even stripped of the one-off — an important check, because the initiation had already flagged that the 70.0% Q2 gross margin contained a $36m non-recurring tariff refund (clean margin 68.7%).

Leg 2 — the revision direction: clearly POSITIVE

Up-revision counts exceed down-revisions on every horizon except FY2027 (2 down, 0 up in the trailing 30 days — noted, not hidden). Direction: up.

Leg 3 — the announcement return: strongly NEGATIVE

The call was after the close on 16 July, so 16 July is the clean pre-announcement mark.

A +2.4% revenue beat and a +8.7% clean EPS beat produced a −13.2% abnormal one-day return. The explanation is in the guidance, and it is specific:

The P&L guidance was raised for the second consecutive quarter; the volume guidance was verbally trimmed. Q2 da Vinci procedures ran +15%, at the top half of the range — so "closer to the midpoint" (~14.5%) for the full year implies H2 deceleration to roughly 14%. The market sold a beat because the volume guide implied deceleration. That is coherent, and it is the single most important thing the print said.

Leg 4 — the drift actually realised so far

The stock bottomed at $328.57 intraday on 23 July — a new 52-week low — and has risen +10.1% off it in three sessions, closing at $361.80 on 28 July.

PEAD VERDICT

The SUE leg and the revision leg are both POSITIVE. The announcement-return leg is strongly NEGATIVE. The realised drift over the seven sessions since the reaction is +5.1% abnormal — i.e. it has so far followed the SUE/revision leg, not the announcement-return leg.

The house stance is long-biased. That stance is WITH the expected drift on the surprise and revision legs, and AGAINST it on the announcement-return leg. Net: the drift evidence is a mild supporting factor for the long bias and a direct contradiction of any short, but it is not clean enough to be leaned on. Seven sessions is not a drift window; PEAD is documented over 60 days, and the measurement window closes ~15 October, just before the Q3 print.

Honesty note on the conflict. Where SUE and announcement return disagree, the empirical literature does not hand you a tiebreak — Chan/Jegadeesh/Lakonishok found both carry independent information. The disagreement itself is the information here: the Street is marking numbers up while marking the multiple down. That is exactly the pattern the initiation already documented for 2026 as a whole (price −36% YTD against FY2026 EPS revised +3.6%), and this print extended it rather than breaking it. This is a de-rating, not a downgrade.

---

R.2 CONVERSION-TRIGGER AUDIT — each quoted verbatim, then marked

Price route

"CONVERSION TRIGGER (long): $310 or below -> E[R] +14.8% [was $305]"

NOT FIRED. The 52-week low was $328.57 intraday / $332.02 close on 23 July 2026 — it came within 5.6% of the trigger and turned. Spot $361.80 is 16.7% above the trigger. Recomputed E[R] at $310 on unchanged scenario inputs: +14.75%. The level is confirmed, not moved.

the dissolved duration-variant test (old Gate 2B) variant route (three conditions, ALL required)

"(a) Management discloses an XiR ASP at or above ~$650k on any call"

NOT FIRED. No call has occurred since the initiation. Management's position on the Q2 call was an explicit refusal ("we haven't said yet what the XiR price range is likely to be… we'll let you model"). The next opportunity is the Q3 call, ~20 October. Management has committed to quantifying the extended-use programme on that call and has NOT committed to XiR pricing — that asymmetry is unchanged and remains the thing to watch.

"(b) ASC placements exceed 27 in a second consecutive quarter"

NOT FIRED — and not observable until ~20 October. Q2 2026 was 27 (20 of them XiR). No interim disclosure exists; ISRG does not publish placements between quarters.

"(c) Sell-side models begin carrying an explicit XiR/ASC line (observable as a consensus FY2028 revenue revision of >2% that is not attributable to FX, M&A or a guidance raise)"

NOT FIRED. The Alpha Vantage EARNINGS_ESTIMATES pull on 28 July returned a response byte-identical to the 27 July cache — FY2027 revenue $13,239.6m, FY2027 EPS $12.0195, FY2026 revenue $11,774.0m, analyst counts 26/31 unchanged. No FY2028 line is carried at all. Consensus has not moved in the last day; the vendor has not refreshed its snapshot. Stated as a limitation, not dressed up as a finding.

Variant route: 0 of 3. the dissolved duration-variant test (old Gate 2B) leg 4 remains unsatisfied.

Invalidation triggers (nine, all checkable)

All nine are keyed to the Q3 print or to a filing. None has fired. Two are re-verified against fresh primary data this pass:

---

R.3 NEW SINCE THE BASELINE — one filing, and it is a governance item

8-K filed 27 July 2026 (accession 0001035267-26-000063), event date 23 July 2026 — Item 5.03, Amended and Restated Bylaws. This filing post-dates the initiation and is the only new company disclosure in the window. The board amended the bylaws to, among other things: allow the company to disregard votes for any nominee from a shareholder that has not solicited holders of 67% of outstanding shares under the universal proxy rules; add synthetic-equity, holding-period and intent disclosure requirements for nominating shareholders; cap the number of candidates a shareholder may nominate at the number of seats up; require a dissident to use a proxy card of a colour other than white; add record-date request mechanics for special meetings; and move non-election corporate actions to a majority of votes cast.

Read, stated with its uncertainty. Most of this is standard post-universal-proxy bylaw modernisation that hundreds of issuers have adopted, and the majority-of-votes-cast change is shareholder-neutral to mildly shareholder-friendly. But the timing is notable and is recorded as such: the board adopted a package of advance-notice and proxy-contest frictions on 23 July 2026 — the exact session the stock printed its 52-week low, 41% below the January high. No activist has filed a 13D. I am not claiming one is present, and this is explicitly not treated as evidence for a thesis. It is logged as (a) a mild negative for any activist-catalysed value-unlock path, and (b) a new dated observable for the calendar. It does not move any Criteria.

---

R.4 CRITERIA RE-RUN ON LIVE DATA — 28 July 2026

CRITERIA                     TYPE       RESULT (28 Jul 2026 live data)
  Quality Criteria           BINDING    PASS            unchanged
  Valuation Criteria         BINDING    FAIL, -8.6pp    see 03_Valuation... s.0.1
                                        (+0.7pp, PASS, on ISRG's own terminal margin)
  Liquidity Criteria         BINDING    PASS            unchanged, but see the vol note
  Catalyst Criteria          MEASURED   PASS (weakened) unchanged - Q3 call ~20 Oct (est.)
  Momentum Criteria          MEASURED   12-1 -17.8%, 18.7th cross-sectional percentile.
                                        ENTRY TIMING ONLY - it does not block this name.
  Consensus Criteria         MEASURED   house FY2027 rev $13,268.5m vs consensus $13,239.6m
                                        = +0.22%. Reported; blocks nothing. The old
                                        variant-vs-consensus test is DISSOLVED, not renamed.

PRICE:  $361.80  (was $357.16, +1.30%)
12-MONTH TARGET: $427 base (+17.9%) - own-multiple construction, see 03_Valuation... s.0.2
REFERENCE LEVEL (not an instruction): prior-document conversion level $310

The INVESTMENT DECISION: WATCHLIST (long-biased) line that headed this block has been deleted.

Valuation Criteria — implied path · FAIL, margin −8.6pp — reversing to PASS on the terminal-margin correction

Fully rewritten. The prior version computed a probability-weighted E[P] of $356.05, netted it against spot to give an E[R] of −1.69%, compared that to a 4.7% cash hurdle, and ranged the result over scenario weights, reporting three "flip points" — on price, on the bear weight, and on the base-case midpoint. All of it is retired. The cash hurdle is gone; the E[R]-versus-hurdle test is gone; and ranging over probability weights while point-estimating the exit multiple is the exact defect the reverse DCF replaces.

Full working in 03_Valuation_and_Factor_Scorecard.md §0.1; inputs from reports/scan/ISRG_analysis.json, not recomputed.

Sensitivity over the exit multiple (never over probabilities):

THE CORRECTION THAT REVERSES THIS RESULT. ISRG_analysis.json carries no terminal_margin field; the published required CAGR inverts to a flat 20.0% applied universe-wide — the hardcoded 0.20 that coverage_scan.py records as a since-fixed defect. ISRG already earns a 29.3% operating margin. Substituting it: required 16.7%, margin +0.7pp, breakeven exit multiple 29.0x — PASS.

Terminal EBIT margin	Required CAGR	Margin	Result
20.0% (scan constant)	26.0%	−8.6pp	FAIL
29.3% (ISRG's own)	16.7%	+0.7pp	PASS

The scan value is the headline because the spec instructs use of the pre-computed inputs. It should not be read as the settled answer on this name.

Liquidity Criteria — feasibility. PASS, but the volatility tier deserved a second look

The vol read I nearly got wrong, stated because the process caught it. Raw realised volatility has apparently exploded: 21-day 71.1%, 63-day 48.3%, against the 35.9% the initiation used. Taken at face value that pushes ISRG from the Moderate (25–45%) tier into High (>45%), which under the sizing grid turns Low-conviction × High-vol into "below minimum → no position" — a tidy, thesis-confirming result. (v1.6.0: the conviction grid is retired as a gating device; volatility remains the live sizing input.)

It is an artifact of one day. Excluding the single 17 July earnings gap:

This is structurally the same error as the "Ion ramp" the initiation falsified — a single anomalous observation driving a series that looks like a regime change and is not. On the 252-day and ex-event measures the name remains squarely Moderate, and the volatility tier and 1.0% conversion size are therefore UNCHANGED.

Vehicle re-check, and the same artifact nearly flipped it. IV ~41% now sits below 63-day realised (48.3%), which would mean options had become cheap and the initiation's "NOT options at current IV" ruling should reverse. It has not: against 252-day realised (36.1%) and 126-day ex-event (33.6%), IV is still 5–7 points rich, and the variance risk premium is intact. The 20-Nov ATM straddle ($360C $37.19 + $360P $29.00 = $66.19) implies an 18.3% move by 20 November against a house base case of roughly +2.8% over twelve months — the implied move still materially exceeds the house expected move, which is precisely the condition under which buying premium is wrong. Vehicle ruling unchanged: outright equity on conversion; no long-premium structure. The pre-authorised 20-Nov $320 cash-secured put re-quotes at bid $13.02 (was $13.92), IV 41.5%, delta −0.242 — effective basis $306.98 if assigned. Still not to be used before the trigger fires; selling a put is long exposure.

Momentum Criteria — MEASURED. Entry timing only. It cannot veto this name.

This section is where the type discipline matters most in the whole batch. The prior version recorded Momentum as a FAIL, described the "mitigation" as "unsatisfiable", and counted it among "three of six gates [that] still fail on their own terms" supporting the classification. Every one of those uses is deleted. Momentum Criteria is MEASURED: it governs when to enter a position the thesis already justifies, never whether to own it. It was a blocking gate, was demoted, and is the change most likely to be silently reversed under pressure — so it is stated in the strongest available terms here.

Assessed cross-sectionally, not by absolute rules such as "above the 200-day".

Read as timing, which is all this Criteria may say: ISRG is a bottom-quintile cross-sectional momentum name (18.7th percentile) 25% below its 200-day and 12% below its 50-day, 10% off a 52-week low set five days ago. That is an unfavourable entry window — an argument for staging entry, or for waiting for the 50-day to flatten, not an argument about whether the business is worth owning. A book that concluded from §0 that ISRG is worth owning would use this section to decide how and when to build, and nothing more.

CORRECTION (2026-07-29) — 12-1 momentum. The scan reports −17.8%; the initiation reported −18.4%; independent recomputation on this project's Alpaca bars gives −23.6%. The spread is the widest of the four names in this batch and is driven by lookback convention (skip-month placement) and end date (the project bars end 2026-07-27 for ISRG). The scan's figure is used because it is what the cross-sectional percentile was computed against. Flagged, not averaged away.

The PEAD read, retained as a genuinely new input. A +5.1% abnormal drift over seven sessions off a 52-week low, with estimate revisions moving up, is the first tentative evidence that the down-trend is decelerating. It is seven sessions, and it is recorded as a watch-item, not a conclusion. Under v1.6.0 it feeds entry timing.

---

R.5 WHAT MOVED THE ANALYSIS, AND WHAT DID NOT

Stated explicitly so the refresh cannot be accused of being a rubber stamp:

1. Price path. The stock traded to $328.57 intraday on 23 July, five days before this refresh, and has since rallied ~10%. Under the retired E[R]-versus-hurdle arithmetic that level sat below the stated flip point and would have turned the old Gate 4 to PASS. That arithmetic is retired, so the near-miss no longer means what it was recorded as meaning — but the dated fact is kept, because the record is the record. The v1.6.0 equivalent statement is that the implied-path margin scales with EV: at $328.57 the required CAGR falls and the −8.6pp margin narrows, without changing the Criteria's structure.
2. Any XiR ASP disclosure ≥$650k would have advanced the variant route. There has been no call.
3. A GI/endoluminal ClinicalTrials.gov registration would have revived Vector C. Re-checked this pass: still zero. All 88 Intuitive-sponsored studies were re-pulled; the sixteen colorectal/colon hits are every one of them transabdominal robotic surgery (da Vinci SP colectomy, staplers, imaging), not endoluminal. Note explicitly: NCT07346066 (IS-001) remains registered under condition "Ureter Injury" and remains NOT_YET_RECRUITING with a stated start of 31 March 2026 — four months past, still not started. The initiation's falsification of that "corroboration" stands and is not resurrected here.
4. A GI/endoluminal FDA 510(k) would have revived Vector C. Re-checked: still none. The most recent Intuitive clearance is K261249 (Universal Seal, 14 July 2026), a sealing accessory. openFDA's index was last updated 20 July 2026, so the last eight days are not covered — stated as a lag, not a null result.
5. Any consensus movement would have moved the Consensus Criteria (MEASURED). The vendor snapshot is byte-identical to yesterday's.

---

R.6 MENTION-FREQUENCY — re-run, source pinned, NO new quarter exists

Source pinned: Alpha Vantage EARNINGS_CALL_TRANSCRIPT, the identical vendor used for the initiation series. No vendor was mixed; the 0-vs-78 discrepancy found on SMR is exactly what that rule exists to prevent.

There is no new quarter to append. The newest call in the series is 2026Q2 (16 July 2026); ISRG's fiscal Q3 ends 30 September and the call is ~20 October. The window is therefore 2023Q1–2026Q2, 14 quarters, identical to the initiation. The series was re-derived from the cached corpus and reproduces exactly. Saying so is the honest output; manufacturing a new data point would not be.

Selected series, per 10,000 words (never raw counts — that rule exists because of this name):

* 2025Q1 (2,400 words) and 2023Q4 (4,033 words) are materially truncated; this is why every figure is per-10k.

The three falsified claims stay falsified — re-verified against the data this pass, not assumed:

1. endoluminal was NOT first-ever in 2026Q2. It appears in 2024Q3 at 2.5/10k full and 4.7/10k in prepared remarks, and again in 2024Q4 — describing Ion. What is new in 2026Q2 is endoluminal applied to the GI tract, and even that is Q&A-only: prepared-remarks gi is 0.0 in every one of the 14 quarters, 2026Q2 included.
2. There is no "Ion ramp." Per-10k across 14 quarters: mean 23.5, σ 6.0, no trend, 2026Q2 (22.9) below the mean. The 4→10→25 series in the first pass was a transcript-length artifact.
3. The ClinicalTrials "corroboration" was and remains a misattribution. Re-pulled and re-read this pass (see R.5 item 3). Not resurrected.

The only defensible new reading is a null one: nothing in the transcript series changed, because no new transcript exists. The XiR/value-segmentation cluster (xir 14.0/10k in prepared remarks — its highest ever; cost 23.3/10k in prepared remarks — its highest in the window; affordability and eup appearing for the first time) is the same finding the initiation made, at the same strength. Next data point: ~20 October 2026.

---

R.7 CALIBRATION — item B8 repeats, and the numbers are unambiguous

CALIBRATION_WATCH.md item B8 ("we reject names on the multiple even when our own forecast is at or above the Street") is at HIGH with four healthcare instances: ISRG (+0.2%), NTRA (+3.9%), TXG (+0.5%), TWST (+0.4%).

This refresh is a fifth data point on the same name, and the pattern is not just repeating — it is strengthening. Stated plainly, as required:

• House FY2027 revenue is +0.22% ABOVE consensus. Unchanged.
• The company then beat the quarter on revenue (+2.4%) and clean EPS (+8.7%).
• The Street raised FY2026 EPS +3.6% and FY2027 EPS +2.0% after the print.
• And the framework's answer is still no — because the multiple implied by the E[P] does not clear a 4.7% cash hurdle.

The B8 pattern in one sentence: our forecast is above the Street, the company beat, the Street revised up, and we are rejecting it anyway on the exit multiple. That is the exact shape B8 describes.

But the counter-argument is strong and is recorded with equal weight, because B8's stated discriminator is whether the rejection survives independent of the multiple:

• The flip point on the bear weight is REACHABLE. 12.3% vs the 25% used. Unlike TXG and TWST — which fail by ~50pp even at a 0% bear weight and are therefore not probability-weighting artifacts — ISRG's Valuation Criteria result genuinely hinges on one unverifiable judgement input. This is the discriminating case B8 asked for.
• ISRG has clean earnings quality. F-score 8/9, accruals −0.9%, GP/A 0.352, asset growth +3.5%, zero debt, $8.63bn net cash. There is no TXG-style non-recurring-revenue problem and no TWST-style retired-disclosure problem. B8's own text says the pattern "needs a healthcare name with clean earnings quality to discriminate." ISRG is that name, and it is now failing on the multiple for the second consecutive pass while beating and being revised up.
• Against that: the price is 1.3% higher than at initiation, so Valuation Criteria mechanically got worse by 1.3pp. Rejecting a name because it went up is not a calibration fault; it is the Criteria working. And the implied- penetration cross-check — the pipeline carries ~35% of the growth the price already requires — is a multiple-independent objection that stands on its own.

Item logged for the user's call. Nothing loosened. See PENDING_ISRG.md.

---

R.8 WHAT THE REFRESH DID NOT CHANGE

Company research (Task 1), the financial model (Task 2), the DCF and comps (Task 3's valuation core), the TAM build, and the chart pack are unchanged and not rebuilt. No new quarter, no new guidance, no new consensus, no new transcript. The one new filing is a governance item that moves no Criteria. Re-deriving any of it would have been activity, not analysis.

---

End of Task 8 refresh. The 27 July 2026 initiation follows, unaltered.

---

What changed in this revision. The old Gate 2 admitted a duration/optionality variant against a four-leg evidence bar. Four vectors were tested against it — Ion/lung, da Vinci cardiac, endoluminal GI, and a fourth (XiR/ASC) found by the generative mention-frequency pass. (v1.6.0 note: the variant-vs-consensus test is now dissolved, and the position verdict this passage refers to has been removed from this document. The vector work below is retained because it is evidence about the business, which is still useful.) Two errors in the prior build were also corrected: the model's DCF spine was running a superseded FY2027 growth rate, and the prior memo's GI corroboration was a misattributed trial.

---

5.0 The four conclusions, kept separate

The framework requires these be stated as distinct conclusions before any decision, because blending them into one narrative ("great company, cheaper than it was, therefore buy") is the specific failure mode this process exists to prevent.

1. FUNDAMENTAL CONCLUSION — the business is compounding and its quality is not in question. Revenue +19% in Q2 2026, installed base +12% to 11,710, placements +18.5%, non-GAAP operating margin 36.5% → 40.5% in H1, gross profitability 0.352, accruals negative, F-score 8/9, $8.63bn net cash, zero debt. But procedure growth is decelerating on a defined path — 18% (FY2025) → ~14.5% guided (FY2026) → ~13% (my FY2027) — US procedure growth is already only 12%, US utilisation is pinned at +3%, bariatric is in high-single-digit decline on GLP-1s, China is deteriorating, and the company has pre-announced a consumable-reducing extended-use programme for H1 2027. This is a high-quality business with a genuinely slowing top line, not a broken one.

Added in v1.4.0, and it is a genuine improvement to the fundamental read: the prior build treated the H1-2027 extended-use programme as an unexplained, self-inflicted revenue headwind. The generative transcript pass shows it is not unexplained — cost, access, affordability, reimbursement and ecosystem all rise together into 2026Q2, with cost at its highest level in fourteen quarters and two thirds of it in prepared remarks. EUP (cheaper consumables) and XiR (cheaper hardware) are the two instruments of one deliberate value-segmentation strategy, announced in consecutive quarters. That cuts both ways and both are stated: it opens a real new channel (§5.1, the dissolved duration-variant test (old Gate 2B), Vector D) and it is evidence management sees enough price pressure to act pre-emptively on hardware and consumables simultaneously.

2. EXPECTATIONS CONCLUSION — the market and the Street disagree with each other, and the house agrees with the Street's numbers. Consensus FY2027 revenue $13,240m (+12.4%) and non-GAAP EPS $12.02 (+11.4%), with estimates revised up over 30 days. The house build lands at $13,270m and $12.89 — a 0.2% revenue gap. The stock meanwhile fell 37% year to date on 100% multiple compression. So the market is pricing something well below what the Street publishes, and the house has no informational reason to take either side.

3. VALUATION CONCLUSION — no edge at $357, and slightly less than the prior build claimed. DCF base $326.53 (−8.6%), corrected — see the box below. Probability-weighted target $356 (−0.3%) on revised scenario weights. Spot sits inside the base-case range of $330–390. The stock is neither cheap nor expensive on the house numbers.

Resolved in v1.4.0, and stated carefully because the prior build did not hide this. The published DCF base of $334.38 ran Assumptions!C26 = 14.5% while the dissolved variant-vs-consensus test (old Gate 2) used the revised house build of +11.5% ($13,270m). The prior model notes flagged that openly as "a deliberate, disclosed inconsistency," arguing the higher number kept the DCF from looking tuned to fit the decision. That argument has merit and the disclosure was honest. But it left the headline DCF value describing a forecast the memo did not hold. The whole spine now runs the house forecast: FY2027E revenue computes to $13,268m, +0.22% versus consensus — what the dissolved variant-vs-consensus test (old Gate 2) always claimed — and the DCF base falls to $326.53, i.e. the change moves against the position, which is how you can tell it was not tuned. The prior build's own estimate of the effect ("roughly $310–315") was too low by $13–16. Verified in Excel after recalculation; Bear and Bull runs are unaffected.

4. PORTFOLIO CONCLUSION — no position. Detailed below. Even a correct bullish research view does not produce a position when the expected return does not clear the cash hurdle and the tape is against it.

---

5.1 The Criteria

Quality Criteria — Causal mechanism: PASS

For a long, the required mechanism is a specific, evidenced driver of improving or durably sustained fundamentals — not "it's a good company."

The mechanism: mechanical installed-base compounding into a 76%-recurring revenue base. Every system placed generates I&A consumption and a service contract for roughly a decade. The installed base grew 12% to 11,710 with a calibrated 5.2% annual retirement rate, so the fleet is genuinely accreting, not churning. Placements grew 18.5% in Q2 and 18.0% in H1. I&A revenue is 60% of the P&L and grew 18% on 15% procedure growth.

Quantitative corroboration (required standard). The Task 3 scorecard corroborates on every axis a long-side mechanism should: gross profitability 0.352 (Novy-Marx — high and rising), accruals −0.9% of assets (Sloan — CFO exceeds net income every year since 2021; no earnings-quality issue), asset growth +3.5% (Cooper/Gulen/Schill — low, favourable), Piotroski 8/9. There is no disagreement between the narrative and the numbers on this Criteria.

PASS.

the dissolved variant-vs-consensus test (old Gate 2) — Variant vs. consensus: FAIL ← still the binding constraint, but now for evidenced reasons

Under v1.4.0 this Criteria accepts either kind of variant. Both were tested.

2A — Estimate variant (near-term): FAIL

The reasoning is unchanged from the prior build and remains correct: an initial build had FY2027 revenue at $13,626m by extrapolating H1-2026 operating leverage, and that variant evaporated once the H1-2027 extended-use programme, US utilisation pinned at +3%, China, and the Italy/Spain/Portugal distributor anniversaries were modelled properly. The residual EPS gap is a 50bp gross-margin assumption and a lower share count — second-order modelling choices, not a differentiated view. (The model now actually computes the $13,268m it always claimed; see the correction box above.)

2B — Duration / optionality variant: FAIL, but this is where the work is

The Criteria requires all four evidence legs. Four vectors were tested. The full build is in 01b_Unpublished_Scoping_and_TAM.md and on the model's TAM & Vectors tab.

Vector D is the near miss, and leg 4 is why it fails. This deserves to be stated precisely, because it is a close call and the temptation to wave it through is exactly how the original NET error happened.

What can be proven: consensus is not explicitly modelling XiR. Management withheld the ASP outright — "we haven't said yet what the XiR price range is likely to be... there's a set of offsetting mix dynamics there that, frankly, we'll let you model" (Q4 2025). You cannot model a revenue line whose price the company refuses to give. A year after launch, a sell-side analyst was still asking for basic unit counts and mix outlook on the Q2 2026 call. Every post-Q2 target change was a cut on US procedure growth; not one note raised on channel expansion.

What cannot be proven, and therefore fails the Criteria: a quantified gap. Vector D contributes ~$170m in FY2026E (1.4% of revenue) and ~$250–300m in FY2027E (~2%) — smaller than the dispersion in the consensus revenue estimate itself, so no gap is demonstrable at any horizon the Street actually publishes. And the sign is genuinely ambiguous: a lower-ASP system is dilutive to near-term systems revenue while being accretive to installed base, procedures, I&A and service. Consensus could be simultaneously too high on one line and too low on another. "Consensus has not thought about it" is not the same as "consensus is wrong by X%", and the Criteria requires the second.

The honest summary: legs 1, 2 and 3 pass for Vector D; leg 4 is a partial. The standard is all four. the dissolved variant-vs-consensus test (old Gate 2) FAILS.

What the TAM work says that no Criteria captures

The implied-penetration statement (01b, Part 5) is the most decision-relevant output of this pass:

At $357.16, ISRG's enterprise value requires roughly 7.7 million annual procedures — 2.4x the 3.24 million performed in FY2025, and 85% of the entire 9-million-procedure pool management itself described as being "in direct line of sight."

And the cross-check:

The four vectors, sized generously and summed, are worth ~1.56 million incremental annual procedures. The price requires ~4.4 million. The pipeline carries about ONE THIRD of the growth embedded in today's valuation. The other two thirds must come from deeper penetration of the core.

This is the direct, quantified answer to the reviewer's thesis. He is right that the pipeline is where the differentiated work lives, and this pass found a fourth vector he did not name and materially downgraded one he did. But sized honestly, the pipeline is the option, not the thesis — and a memo that upgraded to Long on pipeline optionality alone would be substituting a story for the arithmetic.

Catalyst Criteria — Catalyst: PASS, but weakened

Dated and decisive events exist: - Q3 2026 earnings, ~20 October 2026 — carries the promised quantification of the extended-use programme, plus Q3 procedure growth against what management flagged as a "tougher comp in the US." - FY2027 initial guidance, ~late January 2027 — the single most decisive scheduled event.

Both are real, dated, and will move the stock. The Criteria is still weakened but materially less so than in the prior build, because v1.4.0's the dissolved duration-variant test (old Gate 2B) work produced catalysts that resolve a specific disagreement rather than generic volatility events:

• Q3 2026 (~20 October) — the XiR ASP. This is the single disclosure that converts Vector D from "not modelled" into "modellable", and therefore the one that could flip the dissolved duration-variant test (old Gate 2B) leg 4. Management has committed to quantifying EUP on that call; it has not committed to XiR pricing. That asymmetry is itself the thing to watch.
• Q3 2026 — ASC placements. 27 in Q2 2026 was "significantly higher than our history." A second consecutive quarter above 27 establishes a channel; a fall back below it kills Vector D outright.
• Q4 2026 – Q2 2027 — the GI 510(k) decision, checkable at openFDA against applicant "Intuitive Surgical".
• FY2027 initial guidance, ~late January 2027 — still the single most decisive scheduled event.

A catalyst that can flip a Criteria is a real catalyst. PASS, and the weakening is now specific rather than general.

Valuation Criteria — SUPERSEDED IN FULL BY §0.1 OF 03_Valuation_and_Factor_Scorecard.md

SUPERSEDED (2026-07-29). The scenario-weighted expected-return construction below is retired: the cash hurdle is gone, and the sensitivity ran over scenario probabilities while the exit multiple carried a point estimate. The v1.6.0 Valuation Criteria is the implied-path test — required 26.0% vs demonstrated 17.4%, margin −8.6pp (or +0.7pp and PASS on ISRG's own 29.3% terminal margin), exit multiple 29.9x EV/EBIT GROWTH_MATCHED (n=32), implied compression −25.6%. The 12-month target is $427 (+17.9%).

The scenario work below is retained because the business analysis inside it — the bear-driver decomposition and the Vector D evidence — is unchanged and independently useful.

(Retained v1.4.2 construction, for the record)

Computed on the Scenarios tab and verified by reading back the calculated cells in Excel.

Scenario weights were revised, and the sole justification is stated so it can be challenged. The prior build weighted Bear at 30%, and one of the named bear drivers was that benign, high-volume general surgery would migrate to ASCs — a channel ISRG's $1.6m capital system does not serve (§1.8), with Distalmotion having raised $150m specifically to attack it. Vector D is unit-level evidence that this specific gap is being closed: 64 refurbished Xi placed in Q2 2026 against 10 a year earlier, and 27 ASC placements against a negligible history. One named bear driver out of several is materially de-risked, so the bear weight moves 30% → 25%, with the release split to Base (45% → 48%) and Bull (25% → 27%). No other input changed. A reader who rejects this revision gets the prior −2.1% E[R]; under v1.6.0 neither figure is a Criteria output.

(The Base range and midpoint moved down from $340–395 / $367, forced by the DCF correction from $334.38 to $326.53. This is a consequence of fixing the model, not a change of view.)

Less ~10bp round-trip slippage: net E[R] = −0.41%.

Cash hurdle (portfolio_book.json): 4.70%. Shortfall: −5.11 percentage points.

Holding cash beats this position by over five points on a probability-weighted basis. FAIL.

The E[R] is a direct function of entry price, and this is the useful output — all values read back from Excel:

Liquidity Criteria — Implementation feasibility: PASS

• Liquidity: ~3.1m shares/day recently, ~$1.1bn daily dollar volume. A 5% book position is executable in minutes with negligible impact.
• Options: deep, liquid, monthly and weekly expiries, tight-ish spreads on ATM strikes, live Greeks and IV available.
• Borrow: irrelevant for a long; short interest is ~1% of float, so there is no crowding on either side.
• Event risk: quarterly earnings with a demonstrated ±14% one-day move capacity.
• Maximum plausible loss: the bear case is −38%; the 2022 drawdown reached −49.9%, so a −50% path is precedented for this name.

PASS.

Momentum Criteria — MEASURED. Entry timing only.

SUPERSEDED (2026-07-29). This section previously returned FAIL and reasoned that "a long here fights the most robust anomaly in the empirical record", that the Criteria "permits fighting the tape only with…", and concluded "FAIL." Every one of those constructions is retired. Momentum Criteria is MEASURED: it governs when to enter a position the thesis already justifies, never whether to own it. It has no permission to grant and no veto to exercise. See the refreshed treatment in §R.4 above for the cross-sectional reading (12-1 −17.8%, 18.7th percentile, quintile 1).

Read as timing: bottom-quintile cross-sectional momentum, 25–26% below the 200-day, days off a 52-week low. That is an unfavourable entry window and an argument for staging — not an argument about ownership.

---

5.2 CRITERIA SUMMARY (replaces "INVESTMENT DECISION")

The position verdict has been removed, not softened. This block previously opened with INVESTMENT DECISION: WATCHLIST (long-biased) and named a binding gate. v1.6.0 outputs an analysis, not a position. The observable content — the price levels, the horizon, the trigger lists — is retained below as analysis. None of it is an instruction.

CRITERIA                     TYPE       RESULT (27 Jul 2026 initiation data)
  Quality Criteria           BINDING    PASS
  Valuation Criteria         BINDING    FAIL, -8.6pp   (+0.7pp, PASS, on ISRG's own terminal margin)
  Liquidity Criteria         BINDING    PASS
  Catalyst Criteria          MEASURED   PASS (weakened, but now Criteria-specific)
  Momentum Criteria          MEASURED   12-1 -18.4%; 60% of 52wk high; -26% vs 200dma.
                                        ENTRY TIMING ONLY - previously recorded as a FAIL
                                        supporting the classification; that use is retired.
  Consensus Criteria         MEASURED   house FY2027 revenue +0.2% vs consensus. Blocks nothing.

CURRENT: $357.16
12-MONTH TARGET: $427 base (+17.9%)   floor case $397 (+9.6%)
                 up-case anchor UNIDENTIFIED - no stable recent multiple regime
REFERENCE LEVEL (not an instruction): prior-document conversion level $310

TIME HORIZON: to FY2027 initial guidance, ~late January 2027 (2 quarters)

DOWNSIDE CRITERIA (MEASURED) - named permanent-impairment cause:
  worldwide da Vinci procedure growth guided below 10% for 2027

SIZING INPUTS: Volatility MODERATE (35.0-35.9% realised) - the live input to
               inverse-volatility sizing, which is the framework's interim protection
               while the Downside Criteria remains MEASURED pending Brier calibration.

VEHICLE FEASIBILITY: outright equity. NOT options at current IV.

EVIDENCE TRIGGERS (independent of price) - if ALL THREE occur, the XiR/ASC vector
  becomes measurable against consensus rather than un-underwritable:
    (a) Management discloses an XiR ASP at or above ~$650k on any call; AND
    (b) ASC placements exceed 27 in a second consecutive quarter; AND
    (c) Sell-side models begin carrying an explicit XiR/ASC line
        (observable as a consensus FY2028 revenue revision of >2% that is not
         attributable to FX, M&A or a guidance raise)

INVALIDATION TRIGGERS (checkable):
- Q3 2026 (~20 Oct, est.): extended-use programme quantified at worse than ~2% of I&A revenue
- Q3 2026: worldwide da Vinci procedure growth prints below 12%
- Q3 2026: US system utilisation growth prints <= +1%
- Q3 2026: ASC placements fall back below 27, or XiR placements below 64
- Any quarter: da Vinci placements grow below +8% YoY
- FY2027 guidance (~late Jan, est.): procedure growth guided below 11%
- GLP-1 impact described by management in any SECOND procedure category beyond bariatrics
- Adverse Ninth Circuit ruling in the SIS instrument-servicing antitrust appeal
- Ion placements fail to exceed FY2025's 195 in FY2026  [tracking 107 through H1]

What the old block reasoned, and what survives of it. The v1.4.0 duration-variant amendment was written because a Watchlist produced on a 0.2% revenue gap was a miscalibration. That criticism was correct, and this pass took it seriously enough to find a fourth vector neither the reviewer nor the prior memo had named, size it at $1.44bn/yr — the largest of the four — and confirm that consensus is not explicitly modelling it.

The durable finding is the implied-penetration cross-check, and it is retained: all four vectors together carry only about a third of the growth today's price already requires. That is a statement about the gap between price and business, and it survives the framework change intact — it is the same thing §0.1 measures as a −8.6pp margin (or +0.7pp on ISRG's own terminal margin).

SUPERSEDED (2026-07-29). The old block continued: "three of six gates still fail on their own terms — Gate 4 is negative against a 4.7% cash hurdle and Gate 6 is bottom-decile momentum. Even a passing Gate 2 would not have produced a Long at $357", and then explained "why Watchlist and not Avoid" and "why Watchlist and not Short". All of that is retired. The cash hurdle is retired; Momentum Criteria is MEASURED and cannot contribute to any such count; and the memo issues no Long/Watchlist/Avoid/Short verdict to justify.

One line from it is worth keeping verbatim, because it is true and it is the point of the whole reorganisation: "A stock can fall 41% and be neither a long nor a short."

5.3 Position sizing (for conversion)

Weighted conviction composite, scored honestly (revised in v1.4.0):

Changes from the prior build, each justified: Quality Criteria 0.85→0.88 (the XiR/EUP value-segmentation strategy is a coherent, evidenced management response to the ASC channel threat). the dissolved variant-vs-consensus test (old Gate 2) 0.05→0.20 (2B legs 1–3 pass for Vector D; only leg 4 fails — a genuine near miss rather than an absence of any variant). Catalyst Criteria 0.60→0.70 (the Q3 2026 XiR ASP disclosure resolves a specific disagreement, not just generic volatility). Valuation Criteria 0.00→0.05 (E[R] is −0.4% rather than −2.1%, still negative). The composite remains firmly LOW and the grid output is unchanged at 1.0%.*

Per the framework, a strong Quality Criteria and a pristine balance sheet cannot compensate for a zero on the dissolved variant-vs-consensus test (old Gate 2) and Valuation Criteria. Conviction is capped at Low.

Volatility: realised 35.0–35.9% annualised → Moderate (25–45% band). Beta 0.88 vs SPY.

Sizing grid (house max single-name weight 5.0%, per portfolio_book.json):

Resulting size on conversion: 1.0% of book, ceiling 1.5%.

Consensus positioning: a long would be with a strengthening Street (79% Buy; FY26 and FY27 estimates revised up) and against the market's price action. That is not contrarian in the useful sense — it means the position depends on other people re-rating a multiple, with no informational advantage. Named here rather than absorbed silently into the volatility tier.

---

5.4 Vehicle selection

Option A — outright equity (recommended on conversion)

Simple, no time decay, no IV exposure, indefinite holding period. The thesis is a multi-quarter re-rating/compounding story with no single binary event, which is exactly the profile that suits equity. A 1.0% book position at $310 is trivially executable.

Option B — defined-risk call spread (rejected at current IV)

The framework's default options structure is a defined-risk vertical, so it is priced rather than dismissed. Live quotes, Alpaca, 27 July 2026:

15-Jan-2027 $380 / $440 call spread | Leg | Bid | Ask | IV | Delta | |---|---|---|---|---| | Buy ISRG270115C00380000 | 32.16 | 34.60 | 41.0% | +0.494 | | Sell ISRG270115C00440000 | 14.60 | 16.57 | 40.1% | +0.290 |

Net debit $20.00 (worst-case fill) · width $60 · max gain $40.00 · max loss $20.00 · breakeven $400.00 (+12.0%)

Why this is rejected: the breakeven of $400 is above the entire house base-case range of $330–390. The structure only pays if the Bull case (25% probability) arrives inside six months. Expected value is clearly negative. And the premium is expensive on its own terms — IV 41–42% against 35.9% realised is a 5–6 point variance risk premium, and the Nov-20 skew shows calls bid over puts, so one would be buying the richest part of the surface.

Explicit check required by the framework: option-implied move vs house expected move. The Nov-20 ATM straddle (C360 $33.81 + P360 $30.86 = $64.67) implies an 18.1% move by 20 November. The house base case is +2.8% over twelve months. The implied move materially exceeds the house expected move — which is the precise condition under which buying premium is wrong. No long-premium structure is recommended.

Option C — cash-secured put as the entry mechanism (noted, not recommended now)

Since the price-based conversion trigger is a lower price, selling a put is the natural way to get paid to wait, and it monetises rather than pays the variance risk premium.

ISRG261120P00320000 — 20-Nov-2026 $320 put · bid $13.92 · IV 40.8% · delta −0.259 - Premium yield: 4.35% on the $320 strike over 116 days ≈ 13.7% annualised - Effective basis if assigned: $306.08 — 14.3% below spot, and essentially at the $310 conversion trigger

It is NOT recommended today, and the reason matters: selling a put is taking long exposure. With the dissolved variant-vs-consensus test (old Gate 2), 4 and 6 failing, taking long exposure by any vehicle is inconsistent with the decision. It is documented here as the pre-authorised mechanism to be used when the trigger fires, so the conversion is executable without re-deriving it. The risk if used is standard and stated: assignment at $306 in a scenario where the bear case is playing out and fair value is $220 — i.e. the put seller is short exactly the tail the bear case describes.

No trade is recommended today, and therefore no entry is written to trade_recommendations.jsonl. The ledger records recommendations, and there is none. The analysis and its trigger lists are written to the book instead.

---

5.5 Portfolio context — the portfolio-book contract

Read from /Users/oahmady/Desktop/Projects/Investing Hub/portfolio_book.json (as_of 2026-07-27).

Book state: zero positions. Gross 0%, net 0%, 100% cash. Watchlist: NET (short-biased), MU (short-biased). Config: max single name 5.0%, max sector concentration 25%, cash hurdle 4.70%, drawdown ladder −5% review / −10% de-gross 40% / −15% stop, position hard stop −2% of book.

Capital competition test. With an empty book, the competitor is cash. Net E[R] −0.4% versus a +4.7% risk-free hurdle: cash wins by 5.1 points. No existing position would be displaced because there are none. The test fails cleanly — by less than it did, and still cleanly.

Correlation and diversification — the genuine portfolio point, quantified. MU, NBIS and SMR are being underwritten concurrently. Trailing ~1-year daily-return correlations (Alpaca, 268 observations):

ISRG is statistically uncorrelated with the entire AI-infrastructure cluster — correlations of +0.02, −0.01 and +0.00 are indistinguishable from zero — while those three names correlate +0.31 to +0.42 with each other and would function as one factor bet. ISRG is the only non-semiconductor, non-AI-infrastructure name in the batch, and its +0.318 correlation to SPY with 35% realised vol makes it a genuine diversifier.

This is a real and material portfolio argument, and it still does not change the answer. Diversification improves a portfolio's risk-adjusted return by combining assets with positive expected returns. It cannot rescue a position whose standalone E[R] is negative — adding an uncorrelated −0.4% to a book earning +4.7% in cash lowers the expected return. Low correlation is a reason to prioritise ISRG for conversion the moment its E[R] clears, and it is recorded on the watchlist entry for exactly that reason. It is not a reason to own it at $357.

Risk protocol on conversion (inherited from the book): 1.0% position, drawdown ladder applies at the book level; position hard stop at −2% of book, which for a 1.0% position means the name would have to fall ~67% before hitting it — so the practical constraint is the thesis-invalidation triggers in §5.2, not the stop.

Book write-back: the analysis with its trigger lists is provided in PENDING_book_and_ledger_updates.json for the parent session to merge. No position, no ledger entry.