Phase Space AI

05 Trade Construction

Cooper Companies [COO]

The Cooper Companies [COO] — Trade Construction & Risk Management

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 v1.6.0 the memo scores every Criteria and blocks on none of them. It emits an analysis, not a position. The prior version opened with INVESTMENT DECISION: WATCHLIST (long bias) and named a binding Criteria; both are removed. The evidence that produced them is unchanged and retained.

ARCHETYPE: COMPOUNDER          (reports/scan/COO_analysis.json, as_of 2026-07-28)
SPOT: $72.085 (scan) / $72.06 (Alpaca close in the body)

CRITERIA                        TYPE       RESULT
  Quality Criteria              BINDING    PASS (qualified) — evidenced pivot from acquisition-and-leverage
                                                    to cost-discipline-and-cash, sourced from filed 10-Qs
  Valuation Criteria            BINDING    PASS   — implied path requires 6.1% revenue CAGR vs 7.3%
                                                    demonstrated.  MARGIN +1.3pp
                                                    *** BOUNDARY CASE: -2.6pp and FAIL once COO's own 16.7%
                                                        operating margin replaces the scan's 20.0% constant.
                                                        See 03_Valuation_Analysis.md s.0.1. ***
  Liquidity Criteria            BINDING    PASS equity / FAIL options (the chain is not investable at size)
  Catalyst Criteria             MEASURED   PASS   — CooperSurgical strategic-review update; FQ4 + FY2027
                                                    guidance ~4 Dec 2026
  Momentum Criteria             MEASURED   12-1 -2.0%, 31.7th cross-sectional percentile (quintile 2).
                                           ENTRY TIMING ONLY. It does not and cannot block this name.
  Peer Spread Criteria          MEASURED   PASS   — named peers ALC / BLCO / STAA; COO at the 5th percentile
                                                    of its own 5.5-year P/S history
  Downside Criteria             MEASURED   PASS   — permanent-loss case and named cause, s.9
  Short Mechanism Criteria      MEASURED   PARTIAL — growth IS decelerating (organic guidance cut across six
                                                    consecutive quarters) but margin runway is NOT exhausted
                                                    (op margin expanding +220bp). Scored; acted on by nothing.
  Consensus Criteria            MEASURED   PASS   — house within 1% of Street on every line. Blocks nothing.

VALUATION, TWO HORIZONS (both mandatory)
  Implied-path test (5y reverse DCF):  price requires 6.1% revenue CAGR
                                       demonstrated 7.3%   MARGIN +1.3pp
                                       exit multiple 22.7x EV/EBIT, basis GROWTH_MATCHED (n=21)
                                       implied compression from 22.7x today: 0.0% (none)
                                       breakeven exit multiple: 21.4x
  12-month target (own P/S history):   base $77 (+7.1%), band $68 (-5.5%) to $83 (+14.9%)

REFERENCE LEVELS (retained as analysis, not as instructions):
  prior-document conversion level $60.50 · invalidation $54.00
VOLATILITY: MODERATE (31.3% realised) — the input inverse-volatility sizing consumes
VEHICLE FEASIBILITY: outright equity only. The options chain is not investable at size.

---

1. Portfolio context — the book was read first

portfolio_book.json as of 2026-07-27: zero positions, 100% cash, watchlist of MU, ISRG, NBIS, SNDK, GH, NTRA, TXG, TWST, CIEN and others. Config: max single-name weight 5%, max sector concentration 25%, max pairwise correlation without disclosure 0.60, cash hurdle 4.7%, drawdown ladder −5% review / −10% de-gross / −15% stop.

Correlation. With no live positions there is no marginal-correlation test to run. For the record and for future use: COO's 252-day correlation is 0.224 to SPY and 0.498 to XLV; beta 0.558 to SPY and 0.994 to XLV. Against the healthcare cluster already on the watchlist (ISRG, GH, NTRA, TXG, TWST), COO is a medical-device consumables name, not a diagnostics or tools name — the business overlap is low, but the sector-cluster limit would apply if several converted together. Healthcare is already 5 of 16 names underwritten; a sixth healthcare watchlist entry is worth noting against the 25% sector cap even though nothing is live.

Capital competition.

CORRECTION (2026-07-29). The prior text competed COO's central E[R] of +7.5% against a 4.7% cash hurdle. The cash hurdle is retired — replaced by slot competition inside the strategy, not an arithmetic test inside the memo. The comparable v1.6.0 quantity is the Valuation Criteria margin: +1.3pp on the scan's inputs, −2.6pp on COO's own terminal margin. It is a different quantity over a different horizon and is not an expected return. Nothing is displaced because nothing is held.

---

2. Quality Criteria — Causal mechanism · PASS (qualified)

The mechanism, named specifically: a deliberate, evidenced pivot from acquisition-and-leverage to cost-discipline-and-cash, producing a step-change in free cash flow and deleveraging, with a potential CooperSurgical divestiture on top.

Evidence that it is underway, not hypothetical:

Quantitative corroboration (the required standard). The long-side mirror asks for strong profitability + low accruals + high F-score. Result: accruals −3.40% (strongly supportive, best in four years), asset growth +0.6% (strongly supportive), GP/A 0.216 and rising four years running (supportive), F-score 6/9 (not supportive — it fails on ROA improvement, current ratio and gross margin).

The Criteria passes, and the qualification is stated rather than buried: the quantitative evidence corroborates a cash and capital-allocation mechanism cleanly, and does not corroborate a growth or returns mechanism at all — ROA and gross margin are both going the wrong way. This is a cost story with an asset-sale option, not an operating-improvement story. That distinction matters directly for the dissolved variant-vs-consensus test (old Gate 2), because a cost story has a ceiling and the market can see it as easily as I can.

Related calibration point (S5). CALIBRATION_WATCH S5 records that the scorecard was structurally blind to CIEN's forward-disclosure defect. Here the opposite happened: the scorecard led to the correct mechanism (the accruals and asset-growth readings are what identified the M&A stop before the cash-flow statement was opened). Worth logging as counter-evidence to S5's generality — the scorecard is blind to forward-disclosure quality, not blind in general.

---

3. Consensus Criteria — house versus Street · MEASURED, blocks nothing

SUPERSEDED (2026-07-29). This section was "Gate 2 — Variant vs. consensus · FAIL. THIS IS THE BINDING GATE." The variant-vs-consensus test is dissolved, not renamed. v1.6.0 requires no variant versus consensus — that requirement was the contrarian-only trap, and Path B is dissolved with it. The measurement below is retained and reported as the Consensus Criteria (MEASURED).

2A — Estimate variant: FAIL

The house is fractionally below consensus on all four. There is no variant — and a −0.9% gap in the bearish direction cannot support a long. This fails cleanly and in the direction opposite to the thesis.

Why the house cannot get away from consensus here, stated honestly: FY2026 is more than half filed actual, and the remainder is company guidance that consensus has simply adopted (the guidance midpoint of $4,303m versus consensus $4,306m — a 0.07% difference; EPS midpoint $4.62 versus consensus $4.629). To produce a FY2027 variant one would have to disagree with the FY2027 growth rate. I do not, materially: my 4.5% sits inside management's own framing of a market growing at "the low end of the historical 4% to 6% range," and the base-rate literature says nothing about this company that would justify a large deviation.

2B — Duration / optionality variant: FAIL, on two of the four legs

Two candidate 2B theses were tested against the four-leg bar.

Candidate A — myopia management as an under-modelled duration story.

Verdict: 2B fails on legs 3 and 4, and legs 1 and 2 return evidence pointing the other way. This is not a case of "the corpora were silent." They spoke, and they said the growth engine is losing category share. That is the most valuable output of this memo and it is a negative finding on the consensus story.

Candidate B — the CooperSurgical separation as an unmodelled optionality variant.

Verdict: 2B fails on legs 1 and 4. A publicly announced, publicly discussed corporate action is not a variant view. It is a shared expectation with an uncertain outcome — which is a catalyst, and it is credited as such under Catalyst Criteria, not smuggled in under the dissolved variant-vs-consensus test (old Gate 2).

the dissolved variant-vs-consensus test (old Gate 2) therefore fails on both branches, and the dissolved variant-vs-consensus test (old Gate 2) is the binding Criteria. Per the framework, valuation does not by itself determine anything under v1.6.0: the Consensus Criteria is MEASURED, reported, and blocks nothing.

---

4. Catalyst Criteria — Catalyst · PASS (but moot)

These are genuine, dated, and specific. The Criteria passes on its own terms. But Catalyst Criteria requires a catalyst that resolves the the dissolved variant-vs-consensus test (old Gate 2) disagreement, and there is no the dissolved variant-vs-consensus test (old Gate 2) disagreement — so the pass is technically correct and substantively moot. Saying so is more useful than claiming a clean 5-of-6.

---

5. 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 earlier language here about a long being "against" momentum, or about momentum "binding", is deleted. Momentum governs when to enter a position the thesis already justifies — never whether to own it. Assessed cross-sectionally, not by absolute rules such as "above the 200-day".

Read as timing: COO sits in the second quintile of the cross-section on 12-1 (31.7th percentile) — mildly weak, not distressed — while short-horizon trend has turned up (+12.0% over three months, 7.0% above the 50-day, RSI-14 58.5) with the price 1.3% below the 200-day. That is a constructive entry window, which is the only thing this Criteria is entitled to say.

CORRECTION (2026-07-29) — 12-1 momentum. The scan reports −2.0%; this document previously reported −4.8%; independent recomputation on this project's Alpaca bars gives +2.1%. The three differ by lookback convention (skip-month placement and end date), and the sign changes across them. The scan's figure is used because it is what the cross-sectional percentile was computed against, and the percentile is what the framework consumes. The spread is material and is flagged rather than averaged away — on a name whose 12-month momentum is close to zero, the convention determines the sign.

On calibration item B1. B1 argues a blocking momentum rule systematically excludes value entries into fallen names. COO is genuinely below its 200-day and the old rule did not bind here. Under v1.6.0 the question is closed by construction rather than by outcome: Momentum Criteria is MEASURED and has no blocking power to misuse. The B1 record is left as written — the record is the record.

---

6. Valuation Criteria — implied path · PASS, margin +1.3pp — but a boundary case

Fully rewritten. The prior version tested a scenario-weighted E[R] against a 4.7% cash hurdle, reported the result as INDETERMINATE because the hurdle sat inside the E[R] range, and ran its sensitivity over bear-case probability weights. All of that is retired: the cash hurdle is gone, and ranging over probabilities while point-estimating the exit multiple is the exact defect the reverse DCF replaces. Full working in 03_Valuation_Analysis.md §0.1; inputs from reports/scan/COO_analysis.json, not recomputed.

Sensitivity over the exit multiple (never over probabilities):

BOUNDARY WARNING — the terminal-margin correction flips this result. COO_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). COO's own operating margin is 16.7%. Substituting it: required 10.0%, margin −2.6pp, breakeven exit multiple 25.6x — the Criteria reads FAIL.

Terminal EBIT margin	Required CAGR	Margin	Result
20.0% (scan constant)	6.1%	+1.3pp	PASS
16.7% (COO's own)	10.0%	−2.6pp	FAIL

The scan value is the headline because the spec instructs use of the pre-computed inputs. Treat COO as sitting on the PASS/FAIL boundary.

What survives from the old section, because it is about structure rather than arithmetic. The old analysis found a second flip point on the Base-case exit multiple: a 6% change in one unverifiable multiple assumption moved the answer. That observation was correct, and it is the reason the sensitivity now runs over the multiple by construction rather than by exception.

---

7. Liquidity Criteria — Implementation feasibility · PASS for equity, FAIL for options

Equity: clean. - 60-day average dollar volume $208.6m. A 2% book position is a rounding error against that. - Short interest 3.65% of float, 2.8 days to cover — uncrowded. - Realised volatility 31.5% (252d), 28.6% (126d), 32.3% (63d) → Moderate tier. - Maximum plausible 12-month loss on the bear case: −23.4%, or −0.35% of book at a 1.5% position. - No borrow required (long). No dividend liability. No acquisition risk to a long.

Options: not investable. Alpaca chain snapshot, 2026-07-28, four expiries Aug–Dec 2026, 134 contracts:

Maximum open interest anywhere in the chain is ~704 contracts. Implied vol (33–38%) is only modestly above realised (28.6–32.3%) — the variance risk premium is not extreme — but bid/ask spreads of 33–123% of mid destroy any edge before the thesis is tested. An ATM December straddle implies roughly ±18% by 18 December 2026, which already exceeds the Base-case move of +10%; only the Bear (−23%) and Bull (+40%) cases beat it, and both are minority-weight outcomes.

Applying the CIEN precedent recorded in CALIBRATION_WATCH: the Criteria is scored on the position, not on the worst available vehicle. A bad options market is not an uninvestable position when the equity is liquid. Liquidity Criteria passes. No options recommendation is made — structural or tactical — and consequently no entry is written to trade_recommendations.jsonl, because no specific options trade is recommended.

---

8. Sizing inputs and reference levels (replaces "Decision, sizing, and the conversion level")

The position verdict has been removed, not softened. This section previously read INVESTMENT DECISION: WATCHLIST (long bias). Binding gate: 2. v1.6.0 outputs an analysis, not a position. What remains below are the inputs a book would use — the conviction decomposition, the volatility tier, and the price levels — none of which is an instruction.

The six questions below are answered because they are the substance of the analysis, not in order to reach a verdict:

• What is the market wrong about? Nothing demonstrable. The house model lands within 1% of consensus on every metric that matters.
• What evidence supports the differentiated view? There isn't a differentiated view. There is a differentiated research finding — that the myopia-control category is migrating to spectacle lenses, evidenced across four independent records — but it is a negative finding on a line worth 1.1% of revenue, and consensus already models MiSight at its current rate.
• Why now rather than years from now? The CSI review is genuinely near-term. That argues for attention, not for a position taken without an expectations edge.
• What stops the company growing into the valuation? Nothing — that is the Base case, and it returns +6.7% at an unchanged multiple, which is roughly the cash rate plus 2pp for equity risk.
• What would invalidate it? Section 9.
• What does the price require? A 6.1% revenue CAGR for five years against 7.3% demonstrated — a +1.3pp margin, the only PASS in this batch — or −2.6pp and a FAIL on COO's own terminal margin (§6). That is the Valuation Criteria output, and it replaces the retired "is this superior to not owning it" framing, which presupposed a verdict the memo no longer issues.

Conviction composite:

SUPERSEDED (2026-07-29). The rule that "a near-zero Gate 2 caps conviction at Low regardless of anything else" is retired with Gate 2 itself — the Consensus Criteria is MEASURED and constrains nothing. The composite above is retained as a record of how the old framework scored this name, not as a live input. Volatility MODERATE (31.3% realised) is the live sizing input, via inverse-volatility sizing, which is the framework's interim protection while the Downside Criteria remains MEASURED pending Brier-score calibration.

Reference price level, retained as analysis. The prior document derived $60.50 (−16.0% from spot) as the level at which its E[R] arithmetic became robust. That arithmetic is retired, so the level no longer carries the meaning it was given. It is retained because two facts about it are independently checkable and useful: $60.50 is 6.5% clear of the FY2025 low of $58.98 and sits at 12.2x FY2027E non-GAAP EPS, inside the Bear-case multiple. It is a reference level, not a trigger, and this document authorises nothing.

Pre-authorised entry mechanism (documented so conversion is executable, NOT to be used before the trigger fires — selling a put IS long exposure): COO261218P00065000, 18-Dec-2026 $65 put, bid $2.46 / ask $4.03, IV 38.3%, delta −0.270, open interest 68. Effective basis if assigned ≈ $62.54 at the bid. Flagged: with a 48%-of-mid spread and 68 contracts of open interest, this is documented for completeness and is not recommended even on conversion — the outright equity is the vehicle.

Consensus positioning. This stance is with the Street on the numbers (0.9% below on FY2027E EPS) and below the Street on price ($77.45 vs $80.57). It is a mildly more cautious version of a Moderate Buy consensus that has been drifting sideways, not strengthening. No crowding risk is being taken on either side.

---

9. Invalidation and conversion triggers

Converts to LONG (1.5% of book) if: 1. Price reaches $60.50 with the FY2026 organic guide of 3.5–4.5% intact and no seventh consecutive cut; or 2. CooperSurgical is agreed for sale at ≥15x EBITDA (≥$3.8bn) with proceeds committed to debt retirement and buyback — this creates the expectations gap the dissolved variant-vs-consensus test (old Gate 2) currently lacks, because the realised multiple becomes observable rather than assumed; or 3. FY2027 guidance (≈3 Dec 2026) comes in at ≥5% organic and FY2027 FCF guidance ≥$750m — which would break the six-quarter deceleration and constitute a genuine positive variant.

Downside Criteria (MEASURED) — permanent-impairment causes: 1. A seventh consecutive organic-growth guidance cut at the FQ3-26 or FQ4-26 print; or 2. MiSight growth prints below 15% in any quarter — the modality-substitution thesis converting from category-share loss into absolute deceleration; or 3. CooperVision Asia Pacific declines >10% in any quarter, or the weakness spreads to EMEA; or 4. The strategic review is abandoned with no transaction and no explanation; or 5. Gross margin falls below 66% on a non-GAAP basis for two consecutive quarters (tariffs no longer offset by FX).

Falsifies the myopia research finding (would force a re-underwrite of the dissolved duration-variant test (old Gate 2B)): 1. An approved "Labelling Change — Indications" supplement appears on PMA P180035 in openFDA; 2. A US PMA record for MyDay MiSight appears; 3. MiSight growth exceeds 30% for two consecutive quarters; 4. DIMS publication share of "myopia control" rolls over in the 2027 PubMed data.

Hard risk parameters if converted: position hard stop at −2.0% of book (book rule); invalidation-gap exit at $54.00 (−10.7% from the $60.50 conversion, below the Bear target of $55.20); drawdown ladder as per portfolio_book.json.

---

10. The strongest case against this analysis

Stated plainly, because the cost of a false negative is invisible and therefore easy to ignore. If CooperSurgical is agreed within the next eight weeks at 16–17x EBITDA, debt is retired, the buyback runs at $450m+/yr against a 195m share count, and CooperVision re-rates to Alcon's multiple, the stock is $90–100 within twelve months, at a 15.5x-earnings entry with two clusters of CEO/CFO/COO insider buying as the tell.

That is not a remote scenario, and the insider record is the single strongest argument for it.

SUPERSEDED (2026-07-29). The original continued: "the reason this is still a Watchlist is … that the upside is entirely shared with consensus, and the framework's whole purpose is to refuse positions that rest on agreeing with the market more confidently than the market does." That reasoning is retired. No variant versus consensus is required by v1.6.0 — the requirement was the contrarian-only trap, and it is dissolved. Agreeing with consensus is no longer a reason to decline anything; it is recorded under the Consensus Criteria and blocks nothing.