Phase Space AI

Trade Construction

Edwards Lifesciences [EW]

Edwards Lifesciences [EW] — Trade Construction & Risk Management

Task 5 · investment-memo v1.6.0 · analysis dated 2026-07-28, migrated onto the Criteria framework 2026-07-29

This document no longer states a position verdict

Under framework v1.6.0 the memo scores every Criteria and blocks on none of them. It emits an analysis, not a position. Whether this analysis justifies owning EW is a question about a particular book, and two books answer it differently — so the answer is not the memo's to give.

The previous version of this document opened with INVESTMENT DECISION: WATCHLIST and named a "binding Criteria". Both have been removed. What follows is the evidence that produced them, which is unchanged and still useful; the conclusion drawn from it is not.

---

THE ANALYSIS SUMMARY (replaces "THE TRADE TICKET")

ARCHETYPE: COMPOUNDER          (reports/scan/EW_analysis.json, as_of 2026-07-28)

CRITERIA                        TYPE       RESULT
  Quality Criteria              BINDING    PASS   — access expansion, evidenced and quantified;
                                                    scan quality flag PASS
  Valuation Criteria            BINDING    FAIL   — implied path requires 21.1% revenue CAGR vs
                                                    10.8% demonstrated.  MARGIN -10.3pp
  Liquidity Criteria            BINDING    PASS equity / MARGINAL options (57%-of-mid bid-ask)
  Catalyst Criteria             MEASURED   PASS   — CMS final NCD decision memo due 2026-09-13
  Momentum Criteria             MEASURED   PASS   — 12-1 +16.2%, 44.7th cross-sectional percentile.
                                                    ENTRY TIMING ONLY. Never a selection veto.
  Peer Spread Criteria          MEASURED   INDETERMINATE — no own-history multiple spread vs a named
                                                    single peer was computed in this memo
  Downside Criteria             MEASURED   PASS (scored) — see s.6.4; permanent-loss cause named
  Short Mechanism Criteria      MEASURED   FAIL   — growth accelerating, margin runway not exhausted
                                                    (i.e. no short mechanism; scored, acted on by nothing)
  Consensus Criteria            MEASURED   BLANK  — Alpha Vantage quota exhausted. Blocks nothing.

VALUATION, TWO HORIZONS (both mandatory)
  Implied-path test (5y reverse DCF):  price requires 21.1% revenue CAGR
                                       demonstrated 10.8%   MARGIN -10.3pp
                                       exit multiple 22.7x EV/EBIT, basis GROWTH_MATCHED (n=31)
                                       implied compression from 35.2x today: -35.5%
  12-month target (own multiple history, P/S percentile):
                                       base $96 (+14.9%), band $89 (+6.1%) to $105 (+25.8%)

TIME HORIZON: 12 months, anchored to the 2026-09-13 CMS decision and the ~Feb-2027 FY2027 guide
REFERENCE LEVELS (retained as analysis, not as instructions):
  prior-document entry $71.50 · invalidation $58.00
VOLATILITY: LOW (24.3% realised, beta 0.56) — the input inverse-volatility sizing consumes
VEHICLE FEASIBILITY: equity is feasible; options are not. The chain is too thin
   (18-Sep-26 $85 call quotes 1.97/3.54, a 57%-of-mid spread) and Jan-27 implied
   vol of ~31-33% sits ~1.3x above 24.3% realised.

---

1. Portfolio context — the book comes first

portfolio_book.json as of 2026-07-27: zero open positions, 100% cash, fourteen watchlist entries. Config: max single name 5%, max sector concentration 25%, pairwise correlation disclosure threshold 0.60, cash hurdle 4.700%.

Capital competition. The book is entirely in cash, so no existing position would be displaced.

CORRECTION (2026-07-29). The prior text competed EW's net E[R] of +9.6% against a 4.7% cash hurdle and reported a +4.94pp excess. The cash hurdle is retired — it is replaced by slot competition inside the strategy, not by an arithmetic test inside the memo. The comparable number under v1.6.0 is the Valuation Criteria margin of -10.3pp (demonstrated 10.8% minus required 21.1%), which is a different quantity measured over a different horizon and is not an expected return.

Correlation — measured, not assumed. The brief specifically flagged that a prior run asserted correlations that measurement contradicted. Measured from Alpaca daily bars over 251 trading days:

ρ(EW, ISRG) = +0.399 (251d), +0.433 (126d), +0.471 (63d) — below the book's 0.60 disclosure threshold on every window. Two medical-device names do not automatically form one bet, and in this case they do not.

Highest measured correlation to anything in the book or peer set is +0.399 (ISRG). EW's mean correlation to the AI-infrastructure complex that dominates the watchlist is materially lower still. On diversification grounds EW would be the best addition available on diversification grounds. That is an input to a book's ranking, and this document records it without drawing a conclusion from it.

Sector cluster. EW would sit with ISRG in a "medical devices" cluster. Both are watchlist-only, so the 25% cap does not bind today.

---

2. Four conclusions, kept separate

2.1 Fundamental conclusion — POSITIVE

Revenue growth reaccelerated from +3.8% (25Q1) to +11-14% over four quarters. TMTT compounds at +47.7% on a $198.6m quarterly base with two newly-approved platforms and a third and fourth indication expected in Q4-2026. Gross margin 78.0%, adjusted operating margin expanding from 28.7% to ~31% over the forecast. Net cash of $1,847m. Accruals −3.81%, cash conversion 1.49x. This is a good business getting modestly better, and nothing in the research suggests otherwise.

2.2 Expectations conclusion — THE MARKET ALREADY EXPECTS THIS

ΔE = house − Street is +0.8% on FY2026 revenue, −0.3% on FY2026 EPS, +1.6% on FY2027 revenue (against a derived, not sourced, Street figure) and +4.2% on FY2027 EPS. Every one of those sits inside a 28-analyst dispersion, and the largest — the FY2027 EPS gap — shrinks to +2.4% once the tax-rate step is held flat.

The Street is at Buy, average target $100.96, +20.7%. Consensus is more bullish than the house. There is no expectations gap in either direction.

2.3 Valuation conclusion — superseded by the two-horizon test

SUPERSEDED (2026-07-29). The figures below are the v1.4.2 outputs and are retained for the record. The authoritative valuation is now the two-horizon test in §0 of EW_Valuation_Analysis.md: implied path 21.1% required vs 10.8% demonstrated, margin -10.3pp, exit multiple 22.7x EV/EBIT (GROWTH_MATCHED, n=31), implied compression -35.5%; 12-month target $96 base (+14.9%). Note the direction of the change: the old five-year DCF marked EW down to $74.92, 10.5% below spot. The new method does not. It says the price embeds an unusually demanding revenue path, and separately that the 12-month multiple sits at the 18th percentile of its own three-year range. Those are different claims.

DCF $74.92 (−10.5%). Comps put EW essentially on the fitted forward-P/E-versus-growth line for medtech. Probability-weighted 12-month target $92.00 (+10.0%). Implied penetration requires ~290,000 devices a year against ~233,000 today — achievable, but ~3.7x more than the NCD alone delivers.

2.4 Portfolio conclusion — removed; this is not the memo's question

The prior version concluded "NO POSITION — Watchlist". That conclusion has been deleted, not softened. Under v1.6.0 the memo does not decide whether a book should own EW; it supplies the analysis a book decides on.

What survives is the observation the conclusion rested on, which is worth keeping on its own terms: a good business, correctly priced, with no identifiable expectations gap. That is a description, and it is the mirror image of the founding NET error — there, a fine business was almost shorted on valuation alone; the risk here is the same shape in reverse, letting one conclusion substitute for another. Whether that description warrants capital depends on what else is competing for the slot, which this document cannot see.

---

3. Criteria-by-Criteria

Quality Criteria — Causal mechanism · PASS

Named mechanism: a step-change in access to TAVR, driven by three separately-evidenced events, plus a second franchise ramping on two new approvals.

Evidenced, not hypothetical:

Quantitative corroboration (long-side mirror): the Factor & Anomaly Scorecard supports the mechanism. Accruals −3.81%, CFO/NI 1.49x, gross profitability 0.346, asset growth +4.9% — the clean-earnings, high-quality profile where long-side fundamental improvement is empirically most likely to be real. The one dissent is the F-score of 5/9, and four of its four failures are year-on-year deterioration tests (ROA, current ratio, gross margin, asset turnover), three of which are explained by the Critical Care divestiture reshaping the asset base and the $325.4m litigation charge. The disagreement is disclosed rather than ignored; it is not enough to fail the Criteria.

Counter-evidence, stated: the single largest driver of the reacceleration may be the Boston Scientific ACURATE exit, which management credited in four consecutive releases and which lapped in the quarter just reported. That is why the bear case carries a 25% weight rather than less.

the dissolved variant-vs-consensus test (old Gate 2) — Variant vs consensus · FAIL (both branches). BINDING.

2A — Estimate variant: FAIL.

The +4.2% is the only candidate, and it does not survive inspection. It is substantially a tax assumption: the model steps the normalised rate from 23.0% (FY2026, elevated by the enacted R&D-credit law change) to 21.5% from FY2027. That step alone is worth ~+1.9% of EPS. Held flat, the gap is +2.4% — inside dispersion. I could not determine from the Q2 release whether the FY2026 tax elevation is a discrete catch-up or a permanent change.

A variant I cannot distinguish from my own modelling choice is not a variant. the dissolved variant-vs-consensus test (old Gate 2A) fails.

2B — Duration / optionality variant: FAIL on leg 4.

Leg 4 fails on the plainest possible reading of the evidence:

• The CMS proposed decision memo has been public since 15 June 2026.
• Trade press covered it (MedTech Dive).
• Named sell-side analysts are already quoted on it — William Blair's Brandon Vazquez on one-third of severe AS patients being asymptomatic; BTIG's Marie Thibault on existing-centre efficiency and new centre openings.
• The company itself has flagged it in every release since 2025Q4, including the September decision date.
• The sized opportunity — ~1.44pp/yr of incremental growth — sits comfortably inside the dispersion of a 28-analyst estimate set.

There is no version of "consensus has not modelled this" that is honest here. The mention-frequency method found the story faster than reading the sell side would have; it did not find something the sell side has not seen. That distinction is the entire Criteria.

The one item consensus genuinely is not looking at, and why it still fails. SAPIEN X4 appears zero times across 19 consecutive quarters of earnings releases, while ALLIANCE (NCT05172960, n=1,234, primary completion 2027-07) and ALLIANCE Mitral (NCT06167213, starting 2027-01) are registered and progressing. That is a genuine information asymmetry sourced from an independent corpus. It fails leg 3 and leg 4 anyway: there is no disclosed ASP, no placement plan, no regulatory timetable and no revenue before 2028–29, so no defensible bottom-up TAM can be built, and no consensus gap can be quantified against a line item that does not exist. Sizing it would be narrative dressed as analysis. It is logged as monitoring item E-U2, not underwritten.

(This is the pattern calibration item B5 describes — "prove consensus does not model it" is close to unfalsifiable when line-item consensus is unobtainable. Logged there, not acted on here.)

Catalyst Criteria — Catalyst · PASS

CMS final NCD decision memo, CAG-00430R2, due 2026-09-13. Dated, binary, and it resolves the specific question the Quality Criteria mechanism turns on: whether US TAVR access actually expands. Secondary: PROGRESS (moderate AS, n=2,250) and CLASP IITR presentations at TCT (~late October 2026), both confirmed by the company in the Q2-2026 release.

Formally the Criteria passes. It should be read with the honest caveat that a catalyst cannot rescue a variant that does not exist — an event that resolves a disagreement nobody is having is an event, not an edge.

Valuation Criteria — implied path · FAIL, margin -10.3pp (BINDING for a long-only absolute-return book)

This section is fully rewritten. The v1.4.2 version tested a scenario-weighted E[R] of +9.6% against a 4.7% cash hurdle and reported a +4.94pp margin, with the sensitivity run across bear-case probability weights. Both halves of that are retired: the cash hurdle is replaced by slot competition in the strategy, and running the range over probabilities while point-estimating the exit multiple is the exact defect the reverse DCF replaces.

Inputs are taken from reports/scan/EW_analysis.json (as_of 2026-07-28) rather than recomputed.

Sensitivity — over the exit multiple, not over probabilities:

The margin does not turn positive anywhere in a defensible multiple range. EW's demonstrated 10.8% CAGR only justifies today's EV at a 35.4x exit multiple — i.e. at no multiple compression at all from today's 35.2x. That is the honest statement of what the price embeds: five more years at the current growth rate and a permanently premium multiple.

DISCLOSED DEFECT IN THE INPUT (2026-07-29). EW_analysis.json carries no terminal_margin field. The required-CAGR figure inverts to a flat 20.0% terminal EBIT margin applied across the whole scan universe — which coverage_scan.py's own comments identify as a defect fixed after these files were written. EW's actual operating margin is 20.8% (GAAP FY2025, depressed by a $325.4m litigation charge; ~28.7% adjusted). Using 20.8% instead of 20.0% moves required CAGR to 20.1% and the margin to −9.3pp. The headline number is reported from the scan as instructed; this is the correction, stated rather than silently applied. On adjusted margins the gap would narrow further, and that calculation is not run here because the scan's exit-multiple peer set is built on GAAP EV/EBIT and mixing the two would not be like-for-like.

Liquidity Criteria — Implementation feasibility · PASS (equity) / MARGINAL (options)

Equity implementation is trivially feasible. The options chain is thin for a $48bn company and any structure pays a large crossing cost before the thesis can work. This distinction matters, and it is the one the CIEN post-mortem flagged: "this options trade is bad" is not the same finding as "this position is uninvestable." Here the equity is clean and only the options are poor.

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

The type discipline is the point of this section. Momentum was a blocking gate, was demoted to timing, and is the change most likely to be silently reversed. Any language in an earlier version of this document where the tape blocks, vetoes or "fails" EW is deleted. Momentum governs when to enter a position a thesis already justifies — never whether to own it.

EW sits in the middle of the cross-section — 44.7th percentile, third quintile. Not a runner to lean on, not a falling knife to time around. Read as timing: the 50-day above spot with the price on the 200-day is a short-term pullback inside a flat twelve months, which is a mildly favourable entry window rather than an unfavourable one.

CORRECTION (2026-07-29) — RSI-14 convention. The scan reports 16.7, which reads as an extreme oversold print and sits oddly against a name 12% off its 52-week high and above its 200-day. Recomputing from this project's own Alpaca bars: the simple 14-day average form gives 16.7 (reproducing the scan), while Wilder's smoothed RSI-14 — the standard definition — gives 37.5. The scan's figure is the non-standard variant. Both are reported; the standard reading is mildly weak, not extreme oversold.

CORRECTION — 12-1 momentum. The scan's +16.2% and this document's earlier +17.9% differ by lookback convention (skip-month placement). Recomputation on project bars gives +17.9%. The scan's +16.2% is used above because it is the figure the cross-sectional percentile was computed against, and the percentile is the number the framework consumes. The 1.7pp discrepancy is noted, not averaged away.

On calibration item B1: momentum could not have bound here even under the retired rule — EW is neither a beaten-up value candidate nor a top-decile runner. B1 remains untested by EW. Under v1.6.0 the question is moot: a MEASURED Criteria has no binding power to test.

---

4. Required answers for any directional call

Answered because they are the substance of the analysis. They are not answered in order to reach a position verdict — v1.6.0 does not produce one.

• What is the market wrong about? On the evidence assembled here: nothing identifiable. Consensus is at Buy with a $100.96 target, house at $92.00; the fundamental forecasts converge to within a few percent. The disagreement is 2.3 turns of multiple on an identical earnings base.
• What evidence supports the differentiated view? There is no differentiated view on the numbers. There is one genuine information asymmetry (SAPIEN X4, zero mentions in 19 quarters, three registered trials) that cannot be sized.
• Why would the disagreement resolve now? It would not; a 26.0x-versus-23.7x multiple opinion has no resolution date.
• What prevents the company growing into its valuation? Nothing obvious — which is exactly the problem for a thesis in either direction.
• What would invalidate the view? §6.
• What is the adverse scenario? A favourable NCD plus continued TMTT momentum re-rates the stock toward 28x and it runs to $120 without ever offering the entry. That is the cost of waiting, and it is real.
• What does the price require? A 21.1% revenue CAGR for five years against 10.8% demonstrated — a -10.3pp margin. That is the Valuation Criteria result, and it replaces the retired "is passing superior to acting" framing, which presupposed a verdict the memo no longer issues.

---

5. Entry, exit, sizing, vehicle

5.1 Conversion entry

$71.50. Derived from method output, not an arbitrary discount:

• It is roughly the DCF base value of $74.92 less a 4.5% margin of safety.
• At $71.50 the base-case target of $93.36 is +30.6% and the probability-weighted target of $92.00 is +28.7% — a return that no longer depends on where inside the range the bear weight sits.
• At $71.50 the E[R] stays above the 4.7% hurdle even at a 60% bear weight, versus 37.5% at spot. The decision stops being determined by a judgement input, which is precisely what the C4 category asks for.
• It sits inside the DCF sensitivity band ($65.13–$81.49 at a 7.9% WACC).

Invalidation: $58.00 — below the bear-case target of $60.30, i.e. a level at which the bear scenario is not merely being priced but exceeded, implying something is wrong with the franchise rather than with the multiple.

5.2 Sizing, if it converts

the dissolved variant-vs-consensus test (old Gate 2) scoring near zero caps conviction at LOW under the stated rule, regardless of how clean the rest is.

Conviction LOW × Volatility LOW (24.3% realised and beta 0.56 — both criteria, the first name in coverage to qualify for the Low tier) → grid output 1.5% of book, against a 5% single-name cap.

Consensus positioning: the position would be with a bullish and recently-strengthening Street (Buy, $100.96 average, targets refreshed after two guidance raises), but at a lower target than consensus. That is neither a crowded contrarian bet nor a momentum-chasing one. Named next to the grid output rather than absorbed into it; it does not change the 1.5%.

At 1.5% sizing with a $58.00 invalidation from a $71.50 entry, maximum book loss is −0.28%, inside the book's −2.0% per-position hard stop.

5.3 Vehicle — equity, not options

Recommended on conversion: outright equity.

Two structures were priced against real quotes rather than described in the abstract:

A. 18-Sep-2026 $85 / $95 call spread — sized to the CMS decision.

IV 25.0% (long leg) / 27.5% (short leg); delta 0.475 / 0.132; open interest 114 / 275.

B. 15-Jan-2027 $85 / $100 call spread — sized to the structural view.

IV 31.3% / 30.9%; delta 0.548 / 0.258; open interest 744 / 765.

Why neither is recommended, even on conversion:

1. The implied move already prices the catalyst. The 18-Sep-2026 ATM straddle at $85 costs ~$6.74 at mid, an implied move of ±8.0% over the 52 days to expiry. The house has no basis for a September move materially larger than that: a favourable NCD is worth perhaps +4-6% and an adverse one −5-8% on this franchise's revenue mix. Buying that event is paying full price for an edge the memo does not have — exactly the test references/alpaca-options.md requires and exactly the answer it is designed to produce.
2. The bid-ask destroys the structure. At quoted worst-case prices both spreads pay ~1:1 to 2:1. The attractive ratios exist only at mid, and mid is not where a 57%-of-mid spread fills.
3. IV crush. Any September structure carries elevated IV into a scheduled decision that collapses immediately afterwards, regardless of direction. A correct call on the NCD can still lose money.
4. Equity has no expiry. The thesis here is a valuation-entry thesis, and the conversion trigger is a price level with no date attached. Options impose a clock on a view that does not have one.

No entry is pre-authorised at spot. Unlike ISRG, where a cash-secured put was documented as an executable conversion mechanism, no such structure is recommended here: selling a put at these premiums into a 57%-of-mid spread is a poor way to acquire stock, and selling a put is long exposure.

Nothing is written to trade_recommendations.jsonl. That ledger records actual options recommendations, and this memo makes none.

---

6. Invalidation and conversion triggers

6.1 Converts to LONG if — DISJUNCTIVE

Joint probability of at least one trigger, 12 months: naive union ~55%; honest effective ~35–40% after allowing for correlation (T2, T3 and T4 all improve the fundamental case without necessarily closing a gap that is a multiple). Unlike NTRA, the re-rating mechanism here is not purely price — T4 in particular is a genuine evidence route. These are listed as events that would move the analysis, not as conversion triggers for a classification the memo no longer assigns.

6.2 Short Mechanism Criteria (MEASURED) — no mechanism present

Would require all of: (i) TAVR growth below 5% for two consecutive quarters and blended ASP falling y/y; (ii) FY2027 consensus revisions net negative for two consecutive months; (iii) a weekly close below the 200-day MA (~$83.50). Joint probability ~3%. More decisively, the Factor & Anomaly Scorecard argues against a short at any price — accruals −3.81%, CFO/NI 1.49x, gross profitability 0.346 and net cash of $1,847m are the opposite of the profile where shorts work. Classified honestly as NO SHORT AT ANY PRICE, not left on the book as a two-sided costume.

6.3 Invalidation of the cautious view (i.e. reasons this memo would be wrong to wait)

1. CMS finalises the NCD as proposed on 2026-09-13 and the stock gaps through $95. The most likely way waiting costs money, and it is roughly a 30% outcome.
2. TAVR growth prints above 11% in both Q3 and Q4 2026 with the ACURATE comparison fully lapped.
3. TMTT FY2027 guidance above $1.05bn (vs the model's $1,024m).
4. Management quantifies the centre-expansion opportunity in units or dollars.
5. Any disclosure of SAPIEN X4 economics.

6.4 Downside Criteria (MEASURED) — the permanent-impairment case and its named causes

1. TAVR growth below 5% y/y for two consecutive quarters with ASPs falling.
2. CMS retains hospital volume floors and the dual-operator requirement in the final memo — the mechanism in Quality Criteria is then substantially dead.
3. PROGRESS reads out negative for moderate AS at TCT — this would cap the disease-stage expansion runway, not just delay it.
4. TMTT growth below 25% y/y in any quarter before FY2028.
5. FY2027 trial starts below six, extending the thinning-pipeline trend a third year.
6. Adjusted gross margin below 76%.

---

7. What would have to change for the Valuation Criteria to pass

Stated plainly, because it is the most useful thing a FAIL can produce. This is a statement about the Criteria, not about a position:

One of two things. Either (a) the price is ~15% lower, at which point Valuation Criteria stops being decided by the bear weight and the dissolved variant-vs-consensus test (old Gate 2)'s absence matters less because the margin of safety does the work; or (b) a quantified consensus gap emerges — most plausibly management naming SAPIEN X4 with a launch window, or a centre-expansion target in units.

Neither is available on 2026-07-28. The Valuation Criteria therefore reads FAIL, margin -10.3pp. What a book does with that is outside this document.