Phase Space AI

Trade Construction

Booking Holdings [BKNG]

Booking Holdings [BKNG] — Trade Construction & Risk

The memo issues NO position verdict. Criteria are scored; the book decides.

Vehicle: EQUITY

Default. No LEAP proposed: measured option IV was not retrieved inside the time box, so the required IV-minus-realised-vol spread is unquantified. Trailing 252-day realised vol 35.6%. Explicit non-computation rather than a reason the document cannot support.

One vehicle note specific to this name. BKNG split 25-for-1 on 2026-04-06, so option chains are post-split and strikes are dense around a ~$190 spot rather than sparse around ~$4,800. Multi-leg construction that was impractical at the old contract size is now constructible — but per the standing rule, multi-leg requires demonstrated constructability at the strike with quoted size, which was not measured. Recorded as an available avenue, not a recommendation.

Entry / exit
Spot $191.565
12-month target (17.5x EV/EBIT, ~40th pctile of the 3-yr window, on NTM EBIT) $226, +17.9%
High anchor (3-yr median 18.83x) $243, +27.0%
Low anchor (3-yr p25 16.11x) $208, +8.4%
Downside (named: SEO decline forces marketing +80bp of gross bookings) $170, −11.3%
Reward : risk vs the named downside 1.58 : 1

The best reward:risk of the three names in this run — MELI is 1.0:1 and RBA is 0.67:1 — and it comes from the tightest distribution, not the biggest upside.

The parameter the book must decide on

BKNG's implied-path verdict is multiple-dependent and the break-even is 12.7x:

This is not a hedge. It is the actual state of the evidence, and the book should size it as a name whose Criteria verdict turns on whether a 58.4%-measured / 25%-assumed ROIC survives five years. Nothing else in the analysis moves the answer: across an 8.8pp band of FCF margin the required CAGR moves 1.6pp, and the terminal margin is within 0.4pp of the trailing actual.

Invalidation — three falsifiable tests, each with a print date

  1. Room-night growth below 5% on any quarterly print. It has gone 8.8% → 5.9%, with ~2pp of the Q1'26 deceleration attributed to the Middle East. Below 5% ex that drag, the volume leg is gone and 3.76% required CAGR stops being comfortable.
  2. Marketing above 4.7% of gross bookings on a full-year basis (FY2024 and FY2025 were both 4.40%). That is the named downside mechanism, mid-way to the 5.20% modelled, and it is the metric the company itself reports on this exact denominator.
  3. Direct-booking mix described as "a low-fifties percentage" in any 10-Q. It has been "mid-fifties" in both TTM periods. The verbal-bucket disclosure means a step down a whole bucket is the smallest detectable deterioration — which is itself why this test is coarse, and why it is listed third rather than first.

Sizing constraint

size_bucket: mega ($152bn), 35.6% realised vol, no liquidity constraint. Evidence grade A− — the highest of the three names. BKNG decomposes its own gross-bookings growth into FX, volume and price in the filing, which is more than either of the other two names does. The deduction is for the verbal direct-mix buckets and for traffic-source concentration that is discussed qualitatively and never quantified.

Correlation note for the book: BKNG, MELI and RBA share the label "network-effect marketplaces" and almost nothing else. BKNG is European travel demand and Google's traffic policy; MELI is LatAm consumer credit and FX; RBA is North American used-equipment supply and catastrophic weather. Treating them as one exposure would be a mistake the research framing invites.