Phase Space AI

Trade Construction

SentinelOne [S]

SentinelOne [S] — Trade Construction & Risk

As of 2026-07-29. Spot $18.53. No position verdict is issued here — the book decides.

---

1. Liquidity Criteria — BINDING — PASS

PASS. A $6.2bn NYSE-listed name at ~$9.4m of daily notional supports the size this book would take. Per the brief, size constrains position sizing, never admission — and the constraint here is real but modest.

---

2. Vehicle ladder — EQUITY (rung 1)

The measured spread

+14.7pp is squarely inside the 11–15pp band criteria.md documents across five large-software names. Paying a ~15-point vol premium for 17.8 months on a name whose 12-month base target is +9.9% is straightforwardly negative expected value: the $20 call at $5.66 mid needs S at $25.66 (+38.5%) to break even at expiry, which is above even the bull case ($24.23).

Horizon conflict, named rather than resolved silently. S's listed expiries are 2026-12-18, 2027-01-15, 2027-03-19, 2028-01-21. There is no expiry between March 2027 and January 2028 — so nothing matches a 12-month horizon. The nearest longer-dated contract is 17.8 months out and carries the full premium; the nearest shorter is 7.7 months and is inside the thesis horizon, which per rung 3 would require a catalyst before expiry. Where no expiry is both liquid and correctly dated, the answer is equity (the CRDO precedent). It is equity.

Note the depth asymmetry, because it is the WDAY lesson: the $20 strike quotes 104 bid × 1,669 ask. Ample to buy, thin to sell. Open interest is not the constraint; the bid is.

---

3. Sizing

Inverse-volatility sizing at 49.5% realised vol places S mid-tier — above PANW (41.6%) and CRWD (47.6%), below ZS (64.4%) and NET (60.4%). So S sizes larger than ZS on the volatility rule despite being a much smaller and less mature company. That is a limitation of inverse-vol sizing worth flagging to the book, not a recommendation: S has the thinner cash buffer, the weaker disclosure, the unresolved share count and the higher bear probability (0.30 vs 0.20), yet screens as the safer size. This is the HALO lesson — inverse-vol sizing misprices a fat left tail — recurring with a different shape.

Two additional sizing constraints: 1. Correlation. The book already carries PANW, CRWD and NET. S and ZS would make five cybersecurity names. S's direct competitor CRWD is already held. portfolio_book.json limits govern; this memo flags the exposure. 2. Share count. Per-share figures are on a basic 337.0m basis; the diluted count is not determinable (S is dual-class and all three EDGAR share-count concepts 404) and with SBC at 28.96% of revenue it is materially higher. Position sizing should carry a haircut for this, not the valuation — per valuation.md rule 5, uncertainty reduces position size, never the operating assumption.

---

4. Entry, exit and invalidation

The single most useful level: the implied-path PASS flips to FAIL at an exit multiple of 3.55x EV/Sales, the ~13th percentile of the 2026 regime. Above that the price requires less than the business demonstrates; below it, more.

Invalidation gap — specific and falsifiable

The thesis is refuted, and the position exited regardless of price, if any of these prints:

1. Revenue growth falls below 16%. The implied path requires 15.8%. S is at 20.8% and decelerating monotonically from 45.8% over eight quarters. This is the invalidation, and it is a matter of arithmetic on the current trend: continuing the recent pace reaches 16% in roughly two years. The PASS is real and time-limited.
2. GAAP gross margin falls below 72%. Currently 74.0% TTM (74.7% prior year). Below 72% and the "not discounting to hold share" finding — the only clean answer available to the brief's CRWD question — is overturned.
3. Operating margin improvement stalls. +9.22pp TTM GAAP, +7.38pp ex-SBC. Below +3pp in a year and the Quality PASS, which rests entirely on the change, is at risk.
4. SBC as a percentage of revenue stops falling. 41.6% → 28.96% over three years is the only demonstrated evidence for the 12.5% terminal assumption. Flat for two consecutive quarters and the terminal margin is unsupported. (This is the test ZS already fails: its SBC ratio moved −0.20pp in a year.)
5. Deferred revenue growth falls below 5%. Already at +7.5% against revenue at +21.4% — the weakest signal in the name. Below 5% and the unbilled-RPO composition (60.9% of total, +47.9% YoY) becomes the dominant concern rather than a flag.
6. Free cash flow margin fails to recover above 6%. It went 8.77% → 4.24% while management guided improvement. Two more quarters below 6% and the operating-margin improvement is a working-capital effect.
7. The RPO disclosure degrades further — e.g. rounded to $0.5bn, or dropped. It is already rounded to $100m, making every growth rate a ±4.3pp band.

Monitoring that is now the analyst's job because the company stopped doing it

• Net revenue retention. RETIRED — last numeric figure 129% in FQ1 FY23, now "well into expansionary territory." Not reconstructible from any filed statement. Every call must be searched for a conceded number, as ZS conceded 115% under Q&A pressure. Until one appears, NRR direction is INDETERMINATE and must be reported as such rather than assumed stable.
• Calculated billings. S has never reported it. Reconstruct quarterly: revenue + Δ(ContractWithCustomerLiabilityCurrent + Noncurrent) from EDGAR XBRL. Do not use AV's deferredRevenue — it returns None for every S quarter. Current TTM: $1,089.8m, +18.9% — below revenue at +21.4%.
• Unbilled RPO share. Total RPO − total contract liabilities. Currently $913.8m = 60.9% of RPO, up from 53.1%. Rising share means duration is still extending; if it rises while total RPO growth falls, the backlog is deteriorating in composition and level at once.
• Duration. Asked by an analyst and not answered by the CFO. Ask again at every opportunity; it is the single disclosure that would resolve the RPO/billings divergence.

---

5. Downside Criteria (MEASURED — logged, blocks nothing)

Named cause: CrowdStrike and Microsoft compress S's growth below 15% before its ex-SBC margin reaches a level that supports the equity, while the $913.8m unbilled-RPO tranche renegotiates. Full argument in S_Research.md §6.

P(bear) of 0.30 is higher than ZS's 0.20, and the reasons are specific: (i) the multiple has de-rated in every one of five years since IPO with no plateau; (ii) NRR is unverifiable, so the expansion engine cannot be monitored; (iii) FCF margin is going the wrong way; (iv) CRWD is outgrowing S at 4.9x the scale.

Going concern: not applicable, and not argued. $812.5m net cash, zero debt, positive TTM free cash flow ($44.5m), five consecutive quarters of positive non-GAAP net income margin, and the company guides its first full year of non-GAAP operating profit in FY26. The bear case is a growth-and-multiple case.

The one financing observation, for the record and not as a going-concern claim: annual SBC of $303.8m against $812.5m of net cash and $44.5m of FCF means the economic cost of compensation is roughly 6.8x free cash flow. That cost is settled in shares, so it dilutes rather than drains — which is exactly why the unresolved share count (§3) is the most consequential open data item on this name.