Phase Space AI

Trade Construction

JFrog [FROG]

JFrog Ltd [FROG] — Trade Construction & Risk

The memo issues no position verdict. It scores Criteria and hands the book an analysis. What follows is construction detail for whichever side the book takes.

1. Vehicle: EQUITY — ladder step 1

Required disclosure, per the vehicle ladder:

Why not the LEAP — measured, not conventional

Steps 2, 3 and 4 each fail their own test.

Price. IV exceeds realised vol by +5.4 points. Smaller than the 11–15 points on the five large software names checked on 2026-07-29, but a LEAP pays it for 17 months running.

Constructability. Open interest across the entire Dec-2027 $70–$90 call strip is 3, 2, 13, 11, 6, null, 9, null, 6 contracts. This is the HCA (18 across a whole March-2027 chain) and GMED (13) pattern. The quote at the $75 strike looks deeper than the open interest, but a 17.3% of mid spread costs more in a single round trip than the +5.4pp vol premium costs over a year. A vehicle that cannot be filled is not a vehicle, and one filled at 17% of mid is worse — it looks like execution.

Multi-leg. Not attempted. Constructability must be checked at the actual strikes, and the strikes here quote in single-digit-to-low-double-digit size. The FCEL failure (short leg 12 contracts, unconstructable) and the TE failure (constructable in thousands and capped exactly where the thesis pays) both apply.

Horizon. The thesis horizon is the 12-month target plus the 5-year implied path. The liquid expiry (Jan-2027, ~5.6 months) is too short for either; the horizon-matched expiry (Dec-2027) is illiquid. This is the CRDO conflict verbatim. Where no expiry is both liquid and long enough, the answer is equity.

2. Liquidity

Median 60-day dollar volume $187.5m. Market cap $8,878m; float 107.5m shares (88.7% of outstanding); institutional ownership 89.3%, insiders 10.8%.

A $50m position is 26.7% of one day's median volume — one to two days to build or exit at normal participation. Liquidity is not a constraint at any size this book would take. Size is governed by the risk protocol, not by tradability.

3. Position sizing inputs

A caution on inverse-vol sizing here. HALO demonstrated that inverse-vol sizing misprices a fat left tail at only 30% vol. FROG's downside case (§4) is a −45% to −55% growth-and-multiple de-rating, a fatter left tail than a symmetric 71% vol implies, because the multiple sits at the 93.8th percentile of its own applicable history and has 8.96x (the median) to fall to before any fundamental deterioration. Inverse-vol will not fully capture that. Flagged for the book.

4. Downside case — named cause

Scenario: −45% to −55%, to $33–40 per share. Probability 25%.

Named cause: the unbilled-RPO mechanism reversing.

40.6% of total RPO — $233.4m — is contracted but unbilled, up from 19.3% two years ago. That balance is the accounting expression of large multi-year enterprise commitments signed across 2024–2026. The mechanism is not fragile because the contracts are cancellable; they are disclosed as non-cancellable. It is fragile because it requires continuous replacement.

The trigger chain:

1. New large-deal signings pause for two to three quarters. Most likely proximate causes: enterprise DevOps budget consolidation onto a hyperscaler-native artifact registry (the structural competitive risk — the buyer already pays AWS/Azure/GCP), or a security incident in the software-supply-chain product that is currently the upsell engine.
2. Billings, which already grow BELOW revenue at +24.0% against +25.8%, fall further.
3. Deferred revenue stops building. The ex-Δdeferred-revenue operating cash margin — currently 15.86% — drops toward single digits, and the reported 26.9% FCF margin goes with it.
4. Revenue growth converges down toward the self-managed rate of +7.6%, since SaaS growth is what the large multi-year commitments fund.

The arithmetic of the de-rating. At 15% growth and a flat GAAP margin on ~$680m of revenue:

A growth scare in a 71%-realised-vol name overshoots its own p10 — which is how −45% to −55% is reached. The de-rating does most of the work; the fundamental deterioration only has to be credible, not large.

Going concern: NOT APPLICABLE, argued explicitly. $725.5m of lease-inclusive net cash, zero borrowings, $151.4m of TTM reported free cash flow, $239.7m of indefinite Israeli NOL carryforwards, and a $300m buyback authorisation. There is no financing path to permanent impairment. The bear case is a multiple and growth de-rating, not solvency.

5. Invalidation triggers — what would refute the thesis

Monitor quarterly. Each is a disclosed number, so none requires estimation:

6. Entry timing — MEASURED, blocks nothing

12-1 momentum of +107.8% is very strong and cross-sectionally likely top-decile. But the trailing month is −18.7% and spot sits 11.6% below the 50-day average and 26.1% below the 52-week high of $99.22, while remaining 21.5% above the 200-day.

Momentum is decelerating sharply from a very high base. Momentum Criteria is MEASURED — it governs when to enter a position the thesis already justifies, never whether. Here the thesis does not justify a long on the Valuation Criteria (−28.7pp), so the momentum reading is recorded and not acted on.

The Q2 2026 print sits within days (see the catalyst calendar, date confidence INFERRED). Anyone entering before it is taking event risk into a guide that looks like a floor — asymmetric in the long's favour on the print, but into a name that has already fallen 18.7% in a month.