McDonald's Corporation (MCD) — Investment Tree v1

Stage 7 final essay. Bilingual companion: tree_v1_zh.md. Date: 2026-09-06 · Anchor price: $255.69 (2026-09-04 close) · Market cap: ~$180.9B · Derived TTM P/E: ~20.8x · Shares outstanding: 707.6M Archetype: a land-backed rent annuity with a fading royalty kicker — the closest structural analogue is a percentage-rent retail landlord whose tenants are all in the same business, not a franchisor.

SOURCE QUALITY: Tier A dominant. The FY2025 10-K, the Q2 2026 10-Q and the 2026-08-04 8-K supply the revenue split, the property backing, the comparable-sales table, the guest-count disclosure, the equity decomposition, the capital-return figures, the restaurant counts and the loyalty metrics. Tier B on three flagged items: the franchisee relationship survey (source articles returned HTTP 403; search-indexed snippets only), the California FAST Act study (18 locations, one state) and the June-2025 GLP-1 quantification (over twelve months stale). Tier C on aggregator multiples and consensus price targets, both of which are carried as unresolved conflicts and used for nothing. Evidence mode: PRIMARY_FILINGS. Research grade: A.

0. Company Fundamentals — what McDonald's is and how it earns

What it is. 46,028 restaurants across 114 countries at 2026-06-30, of which 44,016 (95.6%) are franchised and 2,012 are company-operated. Customers spent more than $139B across that system in FY2025. McDonald's booked $26,885M of that.

The gap between those two numbers is the company. McDonald's does not primarily sell hamburgers; it collects money from the roughly 2,400 operators who do.

How it earns. Franchised revenue of $16,548M in FY2025 breaks down as:

More rent than royalty. Behind the rent line sits $22.8B of net PP&E under franchise arrangements, including $7.1B of land, and $31,451M of contracted future minimum rents receivable. Franchised operations contribute roughly 90% of total restaurant margin dollars.

Cash-flow anatomy.

A 46-47% operating margin is what a rent-and-royalty model produces. It is not a restaurant margin, and it should not be read as operating excellence — it is a mix artefact of running only 4.4% of the system yourself.

Balance sheet. Total assets $59,920M against long-term debt of $39,863M and cash of $822M at 2026-06-30. Shareholders' equity is $(1,023)M — negative, and the reason is arithmetic rather than distress: common stock $17M + APIC $9,841M + retained earnings $71,987M + AOCI $(2,340)M less treasury stock $(80,527)M. Decades of buybacks have retired more stock than the sum of every other equity account. Retained earnings are strongly positive; there is no accumulated-loss problem. Rated S&P BBB+ / Moody's Baa1 (Fitch NOT VERIFIED), with FY2025 interest expense of $1,582M covered 7.8x by operating income.

Capital returns. Dividend $1.86 per quarter, $7.44 annualised, 50 consecutive years of increases — 60% of FY2025 EPS, and a 2.91% yield at $255.69. Buyback $15.0B authorised with no expiration, $11.73B remaining at 2026-06-30; Q2 2026 repurchases were 2,973,917 shares at an average of $288.62.

What drives it. Systemwide sales, through two channels with different characters: unit count (46,028 and growing ~2,100 a year, mostly on other people's capital) and same-store sales (comps, which are a product of guest counts and average check). The hidden variable — the one H-0 tests — is how much of the rent line is genuinely insulated from the second channel.

I. One-sentence verdict

McDonald's at $255.69 is a levered, land-backed annuity with a fading growth kicker, carrying a real and currently unexplained traffic decline that the market has already marked down 11.4% below the company's own Q2 repurchase average — and at an expected value of $256.53 against a price of $255.69, with an asymmetry of 1.09x and a market-implied bear probability (29.3%) that matches this tree's own (28%), there is no disagreement here to trade: the framework output is Hold-with-sizing, and the entry level is $238.00.

II. Company snapshot

Founded 1955, headquartered in Chicago. The world's largest restaurant company by systemwide sales, and — far less commonly stated — one of the world's largest owners of commercial real estate by site count.

The identity question this tree resolves: McDonald's is habitually described as "a real-estate company with a hamburger brand attached." The FY2025 10-K supports the first half emphatically. Rents of $10,442M exceed royalties of $6,018M; $22.8B of net PP&E stands behind them; $7.1B of that is land.

What the description implies but the filings do not support is insulation. Contracted future minimum rents receivable total $31,451M — the whole remaining schedule, not a year. Divided by any conventional ten-to-twenty-year weighted-average term, that implies a contractual floor of roughly $1.6-3.1B a year against $10,442M collected: 15-30%. The remaining 70-85% is percentage-of-sales rent, which is a claim on the same franchisee sales the royalty is a claim on.

So the accurate description is narrower and less comfortable: a percentage-rent landlord whose tenants are all in one business, plus a royalty on that same business, plus a thin contractual floor underneath. Three claims on one variable, and that variable is now going the wrong way.

III. The five facts that drive everything

1. Rents $10,442M (63.1%) versus royalties $6,018M (36.4%) of $16,548M of franchised revenue, backed by $22.8B of net PP&E including $7.1B of land. The composition is not in dispute and licenses the entire "not a restaurant company" framing. ✅A

1. Contracted future minimum rents receivable of $31,451M against $10,442M of annual rent. On a ten-year weighted-average term that is $3,145M a year floored (30%); on twenty years, $1,573M (15%). For any term above roughly six years, the contractual floor is a minority of collected rent. The term itself is NOT VERIFIED and is the tree's largest single reversal risk. ✅B

1. Q2 2026 comparable sales ran at roughly one-third of the prior year in every segment simultaneously — global +1.3% (from +3.8%), US +0.8% (from +2.5%), International Operated +1.5% (from +4.0%), Developmental Licensed +1.9% (from +5.6%). Four segments differing in ownership model, currency, regulation and consumer income moved proportionally at once, which rules out any cause confined to one of those dimensions. ✅A

1. The residual comp is carried by price: the 10-Q states the US result as "positive check growth… partly offset by negative comparable guest counts." Fewer transactions, higher ticket. Revenue growth borrowed from the transaction base, because each price increase raises the probability of the next transaction lost. ✅A

1. MCD's derived 20.8x TTM P/E and 14.6x EV/EBITDA sit below YUM's 23.41x and 20.54x, and near DPZ's 18.2x and 16.06x — despite the strongest asset backing in the set. This is the fact that most damages the thesis and it is scored against it. The market is not paying a compounder premium for McDonald's; it has already repriced. ⛔B

IV. The H-0 thesis

H-0: McDonald's at $255.69 is two claims stapled into one security — a land-backed rent stream (63.1% of franchised revenue) and a traffic-linked royalty (36.4%) — and both are more exposed to falling guest counts than the "real estate company" framing implies, because the majority of collected rent is percentage-of-sales rather than contractually floored; the correct reading is a levered, inflation-passing annuity whose growth kicker is fading, and at 20.8x trailing earnings the price is closer to fair than to mispriced.

The contradiction that generates it. Fact 1 and Fact 2 cannot both be read the consensus way. If rent is 63.1% of franchised revenue and the contractual floor under it is only 15-30% of that rent, then "the majority of revenue is contractual" is false as stated. The majority of revenue is a claim on franchisee sales, dressed in a lease.

Facts 3 and 4 then say what is happening to franchisee sales: still rising, but on price against falling transactions. The best available offset — a first-party loyalty asset that grew 13% to roughly 220M ninety-day actives with about $40B of trailing-twelve-month member sales — did not prevent it.

Fact 5 is why this thesis resolves at 61% rather than 85%. The market has already taken the stock 11.4% below the price at which the company itself was buying in Q2, and MCD now trades below a pure-franchise peer with no comparable real estate. The thesis is right about what the security is and only partly right about whether the market knows.

Mechanism: rent-annuity-and-royalty-claim-blended-into-one-multiple — two economically distinct claims priced with one franchisor multiple, with the mix drifting toward the more traffic-sensitive leg because developmental-licensee unit growth adds royalty without proportional rent. It is a classification problem, not a level problem, and it points in both directions: a sum-of-the-parts read could as easily justify a lower blended multiple as a higher one.

V. Investment tree (condensed)

Full leaf detail with evidence, reasoning and dated falsification conditions is in leaves.md. Fifteen terminal leaves across five branches.

H-0 · A levered land-backed annuity with a fading royalty kicker · 支持 (h0 61%)
│
├── A · The asset half — what is the rent a claim on?                    支持
│   ├── L1A 1.1  Rent > royalty, backed by owned land            [crux] ✅A 强力支持
│   ├── L1A 1.2  The contractual floor is a MINORITY of rent     [crux] ✅B 支持
│   └── L1A 1.3  Land worth more than $7.1B historic cost  [supporting] ⊗B 证据不足
│
├── B · The operating half — what is the royalty a claim on?           支持
│   ├── L1B 2.1  Comps at 1/3 of prior year, all segments        [crux] ✅A 强力支持
│   ├── L1B 2.2  Price carries the comp; guest counts negative   [crux] ✅A 强力支持
│   ├── L1B 2.3  GLP-1 too small; decline unexplained    [load-bearing] ⛔B 证据偏反
│   └── L1B 2.4  Loyalty is not arresting the decline    [load-bearing] ✅A 支持
│
├── C · The counterparty — who pays both?                              支持
│   ├── L1C 3.1  Franchisee relations deteriorating      [load-bearing] ✅B 支持
│   ├── L1C 3.2  Wage inflation passed to the menu       [load-bearing] ⚠️B 部分支持
│   └── L1C 3.3  Unit growth keeps adding units, third-party funded
│                                                          [supporting] ✅A 支持
│
├── D · The capital structure — who gets the cash?                     支持
│   ├── L1D 4.1  Negative equity is a buyback artefact   [load-bearing] ✅A 强力支持
│   ├── L1D 4.2  Dividend takes ~73% of FCF; buyback flexes
│   │                                                    [load-bearing] ✅A 支持
│   └── L1D 4.3  Buyback execution not counter-cyclical    [supporting] ⚠️A 部分支持
│
└── E · The price — what is already in the number?                  证据偏反
    ├── L1E 5.1  Market paying a compounder multiple      [crux] ⛔B 证据偏反
    └── L1E 5.2  Consensus has not repriced          [supporting] ⊗C 证据不足

Polarity note. Every hypothesis is stated so that supported means H-0 more likely. L1B 2.4 therefore carries ✅ because loyalty is failing to arrest the decline. Reading the glyphs as sentiment inverts the tree.

h0 derivation — computed, not felt

h0 = Σ(tier_weight × verdict_value) / Σ(tier_weight) over terminal leaves. Weights: crux 5 · load-bearing 3 · supporting 1. Values: 强力支持 0.80 · 支持 0.625 · 部分支持 0.45 · 证据偏反 0.225 · 已证伪 0. 证据不足 excluded from numerator and denominator.

h0 = 27.375 ÷ 45 = 60.8% → 61% · band supported (55-74%). (Post-2026-10-06 Stage-6 update — L1B 2.3 部分支持 → 证据偏反, L1C 3.1 部分支持 → 支持; see update_2026-10-06.md. Build anchor, 2026-09-06: 27.525 ÷ 45 = 61.2% → 61%.)

Coverage: excluded weight is 2 of 47, or 4.3% — far below the one-third threshold that would require a low-evidence-coverage flag. Correlated drivers: L1A 1.2, L1B 2.1 and L1B 2.2 share franchisee sales as an input; no mechanical discount is applied because they test different propositions, and the correlation is instead absorbed inside the verdicts, where L1B 2.3 and L1C 3.2 are both held at 部分支持 partly for leaning on the same single traffic disclosure.

Branch B is the strongest and Branch E the weakest, which is the whole finding in one line: the operating deterioration is well evidenced, and the claim that the market has not noticed is not. (Post-2026-10-06: Branch A now leads at 71% and Branch B sits at 66% — management named causes for the traffic decline, so L1B 2.3 no longer supports "unexplained"; Branch C rose to 支持 on the NEXT rent-relief commitment. The rollup tier is unchanged.)

VI. Business model taxonomy

The tree is cut by claim type, not by geography. Geography is MECE but explains nothing here — all four segments decelerated proportionally at once.

The one declared overlap. Percentage rent is reported in Branch A and economically driven by Branch B. L1A 1.2 owns it explicitly rather than hiding it, which is why that leaf is a crux.

The three business models inside the system, and why the mix matters:

• Conventional franchise — MCD owns or leases the site and subleases it. Generates both rent and royalty. The core of US and International Operated.
• Developmental licence — the licensee owns the property and provides the capital. MCD collects royalty only, little or no rent. Predominant in Developmental Licensed markets.
• Affiliate — minority equity, royalty income.

More than 1,800 of the ~2,600 FY2026 gross openings are funded by developmental licensees and affiliates. That is capital-efficient and it is also a mix shift: unit growth adds royalty faster than it adds rent, tilting the revenue base toward the more traffic-sensitive of the two legs. Capital efficiency and revenue quality are trading against each other here, and only the first half of that trade is usually reported.

VII. Valuation scenarios

Anchor $255.69. Horizon twelve months to ~2027-09. Two Tier-A-derived inputs: TTM diluted EPS $12.31 and 707.6M shares outstanding (1,660.6M issued less 953.0M treasury). Cross-check: $255.69 ÷ $12.31 = 20.77x, matching the derived ~20.8x. Net debt ~$39.0B; implied TTM EBITDA ~$15.1B.

Each target is forward EPS × exit multiple. Market cap and EV are computed at the scenario's own forward share count as a cross-check, not as the primary derivation.

Base — 50% — target $262.00

Traffic stays modestly negative; price carries a +1.0-1.5% global comp; ~2,100 net unit additions contribute the guided ~2.5pp; systemwide sales +3.5-4% constant currency; operating margin holds near 46.5%; interest flat; buyback retires ~2% of the count.

TTM EPS $12.31 × 1.065  (op income +~4.5%, share count −2%)  =  $13.11 → $13.10
Target = $13.10 × 20.0x = $262.00

The 20.0x exit is a half-turn below today's 20.8x — the "annuity drift" as the growth leg fades — and stays inside the peer band (DPZ 18.2x, YUM 23.41x). Cross-check: 707.6M × 0.98 = 693.4M shares; market cap $181.7B; EV $220.7B; EBITDA +4% = $15.70B → 14.1x, essentially today's 14.6x.

Return: +2.5% price, +5.4% with the dividend.

Bull — 22% — target $319.60

Traffic inflects; the value platform and the new US leadership convert loyalty engagement into visits; US guest counts turn positive within four quarters; global comps recover toward +3%; systemwide sales +5.5% cc; ~50bps of operating leverage.

TTM EPS $12.31 × 1.105  (op income +~8%, +50bps margin, share count −2%)  =  $13.60
Target = $13.60 × 23.5x = $319.60

The 23.5x exit is set at YUM's current 23.41x — a McDonald's with positive traffic should get at least what a pure-franchise peer with no real estate gets today. Cross-check: market cap $221.6B; EV $260.6B; EBITDA +7% = $16.16B → 16.1x, still well below YUM's 20.54x. The bull case does not require MCD to out-multiple its peer, only to close part of the gap.

Return: +25.0% price, +27.9% with the dividend. Note that $319.60 sits inside the published consensus cluster ($317.55 / $319.88 / $337.13 / $347.00) — this tree's bull case is roughly consensus's base case.

Bear — 28% — target $197.20

Price fatigue arrives; the next increase does not stick; US comps go slightly negative and global goes flat; the franchisee fee dispute resolves with a concession of roughly $100-150M annualised; systemwide sales +1.5% cc on unit additions alone; the buyback compresses to preserve the balance sheet, retiring only ~0.5%.

TTM EPS $12.31 × 0.942  (op income −~6%, share count −0.5% only)  =  $11.60
Target = $11.60 × 17.0x = $197.20

17.0x is the annuity multiple — what a levered, low-growth, 60%-payout, BBB+ income stream trades at once nobody is paying for growth. Yield cross-check: at $197.20 the $7.44 dividend yields 3.77%, a plausible clearing yield for a 50-year dividend grower whose growth has stopped. EV cross-check: market cap $138.8B; EV $177.8B; EBITDA −4% = $14.50B → 12.3x.

What the bear case does NOT assume: no dividend cut, no downgrade below investment grade, no unit-count decline, no rent impairment beyond the fee concession. This is a growth-stops bear, not a business-breaks bear — the right shape given 20/25 durability with zero fatal flags.

Return: −22.9% price, −20.0% with the dividend.

Summary and expected value

EV = 0.22 × 319.60 + 0.50 × 262.00 + 0.28 × 197.20
   =      70.312   +     131.000   +      55.216   =  $256.53

Expected value $256.53 against a price of $255.69 — an edge of +0.33%. With the 2.91% dividend, a total expected twelve-month return of roughly +3.2%, for equity risk.

Asymmetry: upside $63.91, downside $58.49, ratio 1.09x — below every action threshold, and the downside case is the more probable of the two tails.

What the price implies about the bear case. Holding the bull probability at 22% and both upper targets fixed:

255.69 = 0.22 × 319.60 + p_base × 262.00 + p_bear × 197.20,  p_base + p_bear = 0.78
255.69 = 274.672 − 64.80 × p_bear   →   p_bear = 29.3%

The market embeds a 29.3% bear probability; this tree assigns 28%. A 1.3-point gap, inside the error bar of every input feeding it. There is no disagreement to trade.

The price at which there would be. Solving 319.60 − P = 2 × (P − 197.20) gives P = $238.00, where asymmetry is exactly 2.0x (upside $81.60, downside $40.80), the yield is 3.13%, the trailing P/E is 19.3x, and the market-implied bear probability rises to 56.6% against this tree's 28%. That is a genuine disagreement. A looser 1.5x threshold sits at $246.16.

Sensitivity — what would change the answer. Moving the bear probability from 28% to 20% raises expected value 2.4% and leaves asymmetry exactly at 1.09x: asymmetry is a function of the target spread and the price, not of the probabilities, and a shape problem cannot be fixed by becoming more optimistic. The one input that genuinely moves it is L1A 1.2 — if the FY2026 minimum-rent maturity schedule shows a short weighted-average term, the rent leg is substantially floored, the bear multiple should not compress below ~19x (target $220.40), and asymmetry improves to 1.81x. Obtaining that maturity schedule is the highest-value verification available on this name.

VIII. Competitive analysis

Peer figures Tier B, aggregator-sourced, as-of dates NOT VERIFIED.

Three corrections before this table is used. MCD's figure is trailing and the peers' are forward — growing MCD's TTM EPS by the ~6.5% in the base case gives a forward-equivalent near 19.5x, still below YUM and above DPZ. Nothing else in the set owns its real estate. And CMG and SBUX are not comparables; they bound the range. The genuine comparables are YUM and DPZ.

What it says. On EV/EBITDA — the measure least distorted by capital structure — MCD is the cheapest of the three, at 14.6x against YUM's 20.54x and DPZ's 16.06x, while being the most levered of the three ($39.9B of debt at ~5.5x FCF), which normally raises EV/EBITDA rather than lowering it. On P/E it sits between them. It has the strongest asset backing and the lowest EV/EBITDA in the set. That combination is the hardest fact for any "MCD is expensively priced" thesis to survive, and it is why L1E 5.1 is scored 证据偏反.

Two readings, and the evidence does not choose. Either the market discounts the real estate because it correctly sees that most of the rent floats with sales anyway — in which case the market is ahead of the consensus narrative and this tree is merely catching up. Or the market ignores the real estate because no quick-service framework has a place for it — in which case there is a mispricing. This tree takes the first reading as more likely.

The biggest gap in this section. YUM and DPZ comparable sales for the quarters overlapping MCD's Q2 2026 are NOT VERIFIED. If the entire category decelerated equally, L1B 2.1's finding is a macro observation rather than a McDonald's observation and the thesis weakens materially. This is trigger T6 and the first item on the data-gap list.

What is deliberately not done here. A sum-of-the-parts valuation. Splitting the rent line requires knowing the floored share, which is precisely the quantity marked NOT VERIFIED at L1A 1.2. Assigning a REIT multiple to a component whose size rests on an unverified lease term would manufacture precision from the one input already flagged as unknown. Qualitatively the decomposition still informs: the leg deserving the highest multiple (the contractual floor) is the smallest, and the leg growing fastest by unit count is developmental-licensed royalty, which carries no rent at all. A sum-of-the-parts would not obviously produce a higher number than the blended 20.8x.

IX. Forward triggers and red flags

Full playbooks in triggers_redflags.md. Every entry links to a node ID; unlinked triggers are not listed.

T2 is the one to act on. It settles the crux leaf carrying 3.125 of the 27.525 h0 numerator, moves h0 by up to six points in either direction, and moves asymmetry from 1.09x to as much as 1.81x.

T4 must never be read alone. If $238.00 is reached because RF1 fired, the bear case is being confirmed rather than discounted, the targets themselves move down, and the trigger is void. A price trigger without a thesis check is how a value trap gets bought.

RF2 has the widest blast radius — it is the only identified mechanism by which pressure reaches the rent leg, and it converts the bear case's fee-concession assumption into the base case.

RF4 is asymmetric and must be read as two different events. A buyback pause is L1D 4.2 working exactly as modelled and is already inside the bear case. A break in the dividend streak would be the first in fifty years and would invalidate the durability Q3 score. Do not read the first as a warning of the second.

X. Long-term durability assessment

Full test in durability_test.md. Result: 20/25 — Medium durability, upper half of the 17-21 selective-hold band. Zero fatal flags.

Fatal flags fired: 0. Note explicitly that negative shareholders' equity does not trigger the balance-sheet flag — the cause is $80.5B of cumulative treasury stock, not accumulated losses, and that distinction is the difference between a healthy company and a broken one.

The question the score leaves open, and this tree's answer. The durability test ended on the observation that the asset half is nearly bulletproof and the operating half is visibly weakening, and asked whether $255.69 reflects that. It does — almost exactly. The market's implied bear probability of 29.3% sits within 1.3 points of this tree's 28%. That is a perfectly ownable outcome, but it is a different security from "the world's best quick-service compounder," and it should be sized as such.

XI. Tracking dashboard (summary)

Full dashboard in dashboard.md. Review cadence quarterly; review_due 2026-12-05 (Q3 2026 results).

The four numbers that decide this thesis, in order:

1. US comparable guest counts — direction. Q2 2026: negative. Two crux leaves rest on this single disclosure, and it is the bull case's only entry condition.
2. Comparable sales by segment. Global +1.3% / US +0.8% / IOM +1.5% / IDL +1.9%. Watch for any negative print, or a return above +3% globally with three of four segments improving.
3. Rent versus royalty growth in the franchised-revenue detail. Baseline 63.1% / 36.4%. A spread of ≥2pp for four quarters confirms the mechanism; a spread inside ±1pp retires it.
4. The minimum-rent maturity profile. $31,451M total; weighted-average remaining term NOT VERIFIED. The number to get.

Data gaps ranked by how much they would change the answer: (1) YUM and DPZ Q2-Q3 2026 comps — currently the biggest unknown in the analysis; (2) the weighted-average lease term; (3) the magnitude of the guest-count decline, of which only direction is disclosed; (4) a current-year GLP-1 quantification; (5) full-text sourcing for the franchisee survey after the 403; (6) reconciliation of the aggregator multiples, one of which implies an EBITDA below EBIT; (7) the developmental-licensee per-unit rent/royalty split; (8) the Fitch rating.

Cross-ticker context. cycle_exposure: uncorrelated — first-order independent of the AI-capex cycle dominating the rest of this corpus. Not a hedge; it has its own idiosyncratic problem. US-domiciled, so no sovereign_exposure band applies. Nearest structural analogue in the watchlist family is COST, but the shared driver is consumer traffic generally rather than a common counterparty — treat MCD as an independent consumer-traffic exposure, not as a diversifier against the AI-capex block.

XII. Investment Scorecard

Format B (15 questions — long-term hold). Per MANUAL_en.md Part K.6. Verdict values: ✅ = 1.0 · ⚠️ = 0.5 · ✗ = 0 · ⊗ = 0 (non-scoring). Weight tiers: Critical 5x · Load-bearing 3x · Important 2x · Confirming 1x. Tier suffix on each marker is the evidence tier behind that answer.

Section A — Business quality

Q1 [Critical 5x] Does the company have a durable competitive advantage? ✅A Yes. Owned land under 46,028 restaurants in 114 countries, 95.6% franchised, $22.8B of net PP&E behind the rent line, and the strongest brand in the category. The advantage is real and it is asset-backed rather than reputational. Its trajectory is narrowing — durability Q2 scores 3/5 on exactly that — but the question asks whether a durable advantage exists, and it does. Score: 1.0 × 5 = 5.0

Q2 [Critical 5x] Is the business model persistently cash-generative at high returns on capital? ✅A Yes. 46-47% operating margin, FY2025 free cash flow $7,186M after $6,672M in FY2024, H1 2026 $3,706M, and FCF conversion of 83.9%. The margin is a mix artefact of running only 4.4% of the system — but the cash is not an artefact. Score: 1.0 × 5 = 5.0

Q3 [Load-bearing 3x] Is revenue growth durable for 3+ years? ⚠️A Partially. Unit growth is durable and third-party funded — ~2,100 net a year contributing ~2.5pp, corroborated by +1,915 realised TTM. Same-store growth is not: comps at a third of the prior year with negative guest counts. Low-single-digit systemwide growth is durable; mid-single-digit is not established. Score: 0.5 × 3 = 1.5

Q4 [Load-bearing 3x] Is management capital allocation trustworthy and ROIC-maximizing? ⚠️A Partially. 50 consecutive years of dividend increases is among the strongest records in existence and was maintained through multiple recessions. Against it: the return model is leverage-funded at ~5.5x debt/FCF; Q2 repurchases averaged $288.62 against a $255.69 price; and a fee restructuring produced a franchisee relationship score of 1.37/5. CosMc's was wound down cheaply, which is a point in favour. Score: 0.5 × 3 = 1.5

Q5 [Important 2x] Does the company have pricing power? ⚠️A Demonstrably has it, and is visibly spending it. Check growth is positive across the system while guest counts are negative — that is pricing power being converted into revenue, not held in reserve. The FAST Act study shows 8-12% menu increases at studied locations. Pricing power against declining transactions is finite by construction. Score: 0.5 × 2 = 1.0

Section B — Balance sheet and capital allocation

Q6 [Important 2x] Is the balance sheet survivable through a bad scenario? ✅A Yes. BBB+ / Baa1; FY2025 interest expense of $1,582M covered 7.8x by operating income of $12,393M; $39.9B of debt against $7.2B of FCF at ~5.5x. Negative book equity of $(1,023)M is a treasury-stock artefact against retained earnings of $71,987M, not a solvency signal. Even the bear case leaves FCF near $6.5B. Score: 1.0 × 2 = 2.0

Q7 [Important 2x] Is the company buying back shares at a value-accretive price? ⚠️A Ambiguous, and the one observable print is unfavourable. Q2 2026 repurchases averaged $288.62 against a subsequent $255.69 — 11.4% above. One quarter establishes no pattern and hindsight always flatters the observer, but there is no evidence of price-gating either. $11.73B of remaining authorisation with no expiration preserves the option to do better. Score: 0.5 × 2 = 1.0

Q8 [Confirming 1x] Is FCF conversion high (FCF / net income > 80%)? ✅A Yes. FY2025: $7,186M ÷ $8,563M = 83.9%. FY2026 guided to the low-to-mid 80% range. Score: 1.0 × 1 = 1.0

Q9 [Confirming 1x] Is the dividend sustainable (payout ratio < 50%)? ⚠️A Sustainable, but it fails the stated threshold. Payout is 60% of FY2025 EPS, and measured against free cash flow the forward run-rate of $5,265M is ~73% — the binding number, since earnings are not cash. Coverage is 1.36x and the 50-year record is not in question. Scored against the question as written rather than against the sentiment. Score: 0.5 × 1 = 0.5

Section C — Valuation

Q10 [Critical 5x] Is the stock undervalued relative to intrinsic value? ⚠️A Marginally, at best. Expected value $256.53 against $255.69 is an edge of +0.33%, with an asymmetry of 1.09x. The market's implied bear probability (29.3%) matches this tree's (28%) within 1.3 points. The correct reading is fairly valued, not undervalued — and the honest scoring of "is it undervalued?" against that evidence is a half-point, not a full one. Score: 0.5 × 5 = 2.5

Q11 [Load-bearing 3x] Is there a clear catalyst or timeline for the discount to close? ⊗A No discount identified, therefore no catalyst. T2 (the lease-term disclosure) resolves an uncertainty, not a discount. T4 is a price level, which is a waiting instruction rather than a catalyst. T1 (a traffic inflection) would be a genuine catalyst but has zero supporting evidence today. Scored 0 rather than given a token half-point for optimism. Score: 0 × 3 = 0

Q12 [Important 2x] Is the valuation multiple appropriate relative to growth and quality? ✅B Yes. 20.8x trailing (≈19.5x forward-equivalent) and 14.6x EV/EBITDA for a 46%-margin, 95.6%-franchised, land-backed business is defensible on its own terms and is below YUM on both measures despite the strongest asset backing in the set. The multiple is appropriate; it is the asymmetry, not the multiple, that fails. Score: 1.0 × 2 = 2.0

Section D — Long-term hold assessment

Q13 [Important 2x] Does the business earn returns above its cost of capital chronically? ✅A Yes. A 46-47% operating margin, $7.2B of annual free cash flow and fifty years of rising distributions are not compatible with chronic value destruction. Durability Q3's fatal-flag check does not fire. Note that a conventional ROIC cannot be computed against negative book equity, so this is assessed on margin, cash generation and coverage rather than on a single ratio. Score: 1.0 × 2 = 2.0

Q14 [Important 2x] Is there a reinvestment runway at high incremental returns for 5+ years? ⚠️A Real but structurally small. ~2,600 gross openings on $3.7-3.9B of capex is genuinely high-return, but more than 1,800 of those are other people's capital — so McDonald's own reinvestment is limited and the majority of cash leaves the business as dividends and buybacks. A company returning more than it reinvests has a shorter internal compounding runway by construction. Score: 0.5 × 2 = 1.0

Q15 [Confirming 1x] Is there meaningful unpriced optionality? ⚠️A Some, unquantifiable. Loyalty at ~220M ninety-day actives with ~$40B of member systemwide sales is a real first-party asset that did not exist a decade ago and is barely monetised. The land carries embedded value above its $7.1B historic cost. Neither can be sized from disclosure — L1A 1.3 is 证据不足 for exactly this reason — so the optionality is scored as present rather than as valuable. Score: 0.5 × 1 = 0.5

Scorecard summary

Verdict tally (Section XII): 5 ✅ · 9 ⚠️ · 0 ✗ · 1 ⊗

K.3.5 weighted-score derivation

Tier assignment: 3 Critical (5x) = Q1, Q2, Q10 · 3 Load-bearing (3x) = Q3, Q4, Q11 · 6 Important (2x) = Q5, Q6, Q7, Q12, Q13, Q14 · 3 Confirming (1x) = Q8, Q9, Q15. Theoretical maximum: 3×5 + 3×3 + 6×2 + 3×1 = 15 + 9 + 12 + 3 = 39.

xii_score = 26.5 / 39 = 68% (67.9% exact, rounded to the nearest point). Band: ≥65% — moderate, buy with sizing discipline.

Reconciliation with h0. xii_score 68% measures structural quality on the fifteen-question scorecard; h0 61% measures thesis confidence in the H-0 statement. The 7-point gap is the normal direction — business fundamentals are stronger than the narrow thesis — and it names the headroom that a traffic inflection (T1) would close. Both numbers are computed in this document from stated inputs; neither is a judgment set.

Final verdict

HOLD WITH SIZING — framework output, not personal advice. Not actionable at $255.69; the entry level is $238.00.

The scorecard says this is a high-quality business (68%, moderate band) whose price is the problem rather than whose quality is. Ten of fifteen questions score at or above a half-point, and the only zero is Q11 — the absence of a catalyst, which is a statement about timing rather than about the company.

The binding constraint is Q10. Expected value of $256.53 against a price of $255.69 is not an investment case; it is a fair price. Asymmetry of 1.09x is below every action threshold, and the market's implied bear probability sits within 1.3 points of this tree's own. When your estimate and the market's estimate agree to within a rounding error, the correct action is to wait for one of them to change.

The 2-minute pitch

McDonald's is described as a real-estate company with a hamburger brand attached, and the first half is true: it collects more rent ($10,442M) than royalty ($6,018M), backed by $22.8B of property including $7.1B of land. What is not true is the implication that most of the revenue is therefore contractual. The contracted minimum rents total $31,451M — the whole remaining schedule — which on any conventional ten-to-twenty-year term means only 15-30% of collected rent is actually floored. The other 70-85% is percentage-of-sales rent: a claim on the same franchisee sales the royalty claims.

Which matters now, because those sales are growing on price against falling transactions. Comps ran at a third of last year in all four segments at once, and the 10-Q says the US result is "positive check growth… partly offset by negative comparable guest counts." No named cause fits: GLP-1's only quantification is 0.9% of sales and is over a year old; the California wage study covers 18 restaurants. Loyalty grew 13% to 220M actives over the same period and did not stop it.

So you are buying a levered, land-backed annuity that passes inflation to the menu and thereby spends the traffic its percentage rent is calculated on. That is a perfectly ownable security — 20/25 durability, zero fatal flags, fifty years of dividend increases, BBB+ — but it is not the compounder the description implies.

And the market appears to know. MCD trades below YUM on both P/E and EV/EBITDA despite far better asset backing, and 11.4% below the price at which the company itself was repurchasing in Q2. Expected value $256.53 against $255.69. You are not being paid to take this position today. At $238.00 you would be.

Relevant risk types (per MANUAL_en.md Part K.4)

When NOT to buy — anti-pattern check (per MANUAL_en.md Part K.5)

Two anti-patterns fire. Both point the same way: this is a good business at a fair price with an unexplained problem, and the discipline that matters here is not confusing the first clause for a reason to ignore the third.