HScreen · ORCL

ORCL · Oracle Corporation 5d old

conviction2

coverage tierT2-PARTIAL

horizon12 months

Return distribution probability-weighted scenarios, never collapsed to a point

-22%30%

+16%48%

+70%22%

EV +16.48%

why these probabilities: All legs from USD 143.8875 (2026-08-18), 12 months, built forward growth -> revenue and margin -> EPS -> multiple -> price, and all dividend-inclusive at the current USD 2.00 annualised rate (1.4% yield). Consensus source for every growth figure: fmp:/stable/analyst-estimates?symbol=ORCL&period=annual, as_of 2026-08-19, 86 covering analysts, period-matched to fiscal years ending 31 May. BASE — what the market expects, delivered. Consensus FY2027 revenue growth +33.2% and EPS growth +7.6% (FY2026 reported EPS 5.94 -> FY2027 ~6.39), then FY2028 revenue +45.5% and EPS +35.6% (-> FY2028 ~8.67). In 12 months the stock prices off FY2028. Holding consensus EPS but applying a modest de-rate from today's ~22.5x forward to 19x forward FY2028 EPS of 8.67 gives ~165, i.e. +14.5% price +1.4% dividend = ~16%. The de-rate rather than a flat multiple is the honest reading of a market that has already cut 58%: it is paying less per unit of guided-but-not-delivered AI earnings. This leg is DESCRIPTIVE — I am not competing with consensus on the level. BULL — growth materially faster than consensus: FY2028 revenue growth ~55% versus consensus +45.5%, on RPO converting at or ahead of the guided pace (USD 130bn RPO, 2025Q3; multicloud revenue +404% and bookings +325% y/y, 2026Q4) with blended AI gross margin holding at the top of the guided 30s and applications mix (+21% industry apps, +17% ERP, 202

not built — options-implied distribution on the same axis, with the KL number and where the disagreement sits
needs: desk-options implied bands (v1.4 Part 4a) — activates at 60% coverage; currently 33%

Visibility how far ahead this name can be seen

not built — visibility score
needs: visibility_score.py

Trust how much weight this name's own record can carry (credibility Z)

not built — credibility
needs: credibility.py

Sizing survival-first size (half-Kelly) vs the enforced ladder

ladder (enforced)3%

kelly-derived (report-only)0.3%

divergence2.7pp

inputsEV(ann) 16.48% · σ 32.78 · Z

Falsifiers what would prove this wrong

The contracted RPO book converts into delivered cloud infrastructure revenue at or near the guided pace
Blended AI data-centre gross margin holds in the 30s as guided, so incremental revenue carries incremental profit
Free cash flow inflects positive as capex intensity peaks, so the build is self-funding by FY2028
FY2027 capex is guided at a level consistent with consensus FY2028 EPS of ~8.67
No large AI customer cancels or renegotiates contracted capacity
The dividend at USD 2.00 annualised remains funded without incremental borrowing

Conflicts surfaced, never netted out (T4)

Our probability-weighted expectation is +16.5% over 12 months versus a street implied +67.2% to the dated 240.57 consensus target (86 analysts, 56 buy / 26 hold / 4 sell). We do NOT disagree with the street's arithmetic	high
The revision tape is UNAVAILABLE, not neutral: estimates_query.py returns n_up 0, n_down 0, breadth 0.0, direction flat, median_move null, across only 7 estimate rows for ORCL. On a stock down 58% from its 52-week high,	high
Consensus disagrees violently with itself in the years our thesis is priced off: revenue dispersion 14.8% for FY2028 rising to 29.9% for FY2029, and a target range of 95 to 325 — a 3.4x spread. The point estimate is ther	high
p_bear 30% is 16pp above the pooled anchor and 23pp above the historical anchor. The anchors are computed on five effective windows of a low-beta software compounder and pooled against two early-stage neoclouds (CRWV, NB	medium
Management guides AI data-centre gross margin "in the 30s" while describing legacy autonomous-database/SaaS margins as "stunningly high". Both statements are from the company and both are credible; together they mean rev	medium
Options, crowding, technical and factor desks do not exist, so there is no independent dimension to agree or disagree with our view. Four of the seven cross-checks the system is designed to run are absent for this name.	medium

The case against independent adversary, different model family (P6) weakened

Version history from the research record

archived versions1

archived	hash	bytes
2026-08-19T08:32:01	ad88f893fff1	35,452

One version only — the archive was created today. The diff view fills in on the next re-run.

Rendered from canonical artifacts only (T1) — no terminal database exists. Every figure is read from the system's own files at render time; where an artifact is missing the panel names the dependency rather than showing an empty box. No execution path exists anywhere in this interface (T5), not even a link.