Phase Space AI

Trade Construction

Lumentum Holdings [LITE]

Lumentum Holdings [LITE] — Trade Construction

The memo issues no position verdict. This states how a position would be built if the book chose to take one, and what would invalidate it.

Vehicle: EQUITY

Ladder rung Vehicle Argument required Argument given
1 Equity none — default, adopted
2 LEAP IV − RV, plus why convexity justifies the spread not offered — fails on depth, see below
3 Sub-horizon option catalyst before expiry not offered
4 Multi-leg constructability at the actual strike not offered

Required disclosure

Trailing 252-day realised volatility 91.2%
ATM implied volatility, 12–24m calls 98.3% – 101.6%
IV − RV +7.1 to +10.4 points — the narrowest of the three names in this cluster
Quoted size at the intended strike $600C 2027-07-16: 10 × 13 contracts, bid $228.81 / ask $232.84, OI 29
Next expiry $600C 2028-01-21: 6 × 10, OI 22
Deepest nearby strike $610C 2028-01-21: 21 × 30, OI 46
Equity liquidity ~$46.9bn common market cap — ample for any size this book would take

Why not the LEAP, explicitly

On price, LITE is the most defensible option candidate of the three: a +7.1 to +10.4 point implied-over-realised premium against COHR's +12.5 to +14.0, and FN's chain does not exist at all.

It fails on depth. The $600 July-2027 strike quotes 10 contracts bid against 13 offered, on 29 of open interest. That is the thinnest quote of any strike checked across the three names, and it sits squarely in the range the framework's own record calls uninvestable — HCA at 18 contracts across an entire chain, GMED 13, CRDO 78. Chain depth is not strike depth, and here the aggregate chain looks serviceable while every near-the-money strike is 6–30 contracts.

Second, name-specific reason. LITE's equity already embeds substantial optionality, in the wrong direction for a holder. The diluted share count is a function of the share price — it rose from a guided 83.5m in November 2025 to ~102m in May 2026 — so upside is damped by issuance, while the $3,198.4m of cash-settled convert principal is a fixed claim regardless of price. Buying a call on a capital structure that is already option-laden compounds a risk the inverse-volatility sizing rule does not capture. Equity.

Position construction

Parameter Value Reasoning
Instrument LITE common
Entry Scale over 3 tranches, wider spacing than the other two names 91.2% realised volatility on a $602 stock is roughly a $35 daily standard deviation
Entry zone $500–600 Below spot. Note the reference point: NVIDIA paid $695.31 in March 2026 and is 15.4% under water, so recent strategic-buyer pricing is above the current quote, not below it — that is a fact about price, not a valuation argument
Sizing input 91.2% realised vol — the highest of the three names Inverse-vol sizing therefore gives LITE the smallest weight of the three. That is the correct mechanical answer and it agrees with the evidence grade (C, the lowest here)
Evidence-grade overlay C valuation.md rule 5: uncertainty reduces position size, never the operating assumption. The 28.0% terminal margin is not haircut for the retired segment disclosure or the reconstructed share count — the position is
Bear-case overlay −45% base, −72% deep The framework's own HALO finding is that inverse-vol sizing misprices a fat left tail. LITE's left tail is fatter than 91.2% vol implies, so the vol-derived size is a ceiling, not a target
Single-name cap 20% book Not remotely approached at evidence grade C
Correlation constraint — binding Six names in this book express one driver: COHR, FN, LITE, AAOI, CRDO, CIEN And LITE is doubly entangled: NVIDIA is 27.6% of Fabrinet's revenue and, since 2026-03-02, the holder of $2.0bn of LITE convertible preferred. A book holding FN and LITE holds one counterparty twice, in two capacities. Size the cluster, not the name

Exit and invalidation

Thesis horizon: 12 months, to the Q4 FY2027 print, with the 5-year implied path as the ownership test.

Trigger Level Action
Invalidation — the terminal margin was not real Non-GAAP operating margin below 30% for two consecutive quarters The 28.0% terminal margin and therefore the entire PASS rest on the guided 35–36%. Exit
Invalidation — the bear case confirms Cloud transceiver revenue growing slower than unit shipments for two consecutive quarters ASP deflation has overtaken volume. The 10-Q already discloses this qualitatively ("partially offset by lower average selling prices"). Exit
Invalidation — the product is late LITE's own 1.6T transceivers do not ship by end-CY2026, or the 102.4T switch-silicon gate slips past CY2026 LITE is already a year behind COHR; a further slip means an 800G ASP-deflation trough with no 1.6T revenue to offset it. Exit
Invalidation — concentration turns into concealment Any customer's receivable share exceeds its revenue share by more than ~3pp Currently −1pp. This is the AAOI signature and it is the single most important comparison in the filing
Invalidation — growth decelerates below the argument FY2027 guide below ~+45% The PASS WITH ARGUMENT rests on ~+64%; below +45% the argument does not bridge to the required 49.61%
Invalidation — dilution overruns Diluted count in any 10-Q EPS note materially above the price-implied iteration (~103m at $827) The count is reflexive; a surprise above it means the capped calls do less than assumed
Target $827 (+37.3%) Trim, do not exit
Valuation stop EV/Sales above ~30x on the fully diluted count Currently 24.76x at ≥91st percentile of own history
Drawdown ladder Per the book protocol; nothing name-specific overrides it

The two numbers a reader must hold together

  1. LITE is the only name of the three to PASS the Valuation Criteria on the primary measure, by +19.39pp — the widest margin in the cluster.
  2. That margin becomes −11.02pp the moment the assumed growth rate is set to 40% rather than the trailing 69%, and the required path implies $18.7bn of FY2031 revenue against a company ambition of $8bn annualised.

Both are true. The first is what the criterion measures; the second is what the criterion cannot see. The position should be sized for the second and entered on the first.

What would move this to a clean PASS

Named, so it is falsifiable:

  1. A quantified backlog or multiyear-commitment figure. LITE asserts "multiyear commitments" with no value, term or counterparty. A number would convert the required 49.61% CAGR from an extrapolated twelve-month rate into contracted revenue — the single highest-value disclosure LITE could make.
  2. Reinstated segment or product-line disclosure covering the declining business. The 92.1% segment profit decline in Industrial Tech is now invisible; making it visible again would lift evidence grade C toward B and increase the position size directly.
  3. Two consecutive quarters at the guided 35–36% non-GAAP operating margin. That converts the unprecedented-and-verified terminal margin into an unremarkable one.