Back to Portfolio
PBR

Petroleo Brasileiro S.A. — Petrobras

$17.75 114.4B market cap 2026-06-06
Petroleo Brasileiro S.A. - Petrobras PBR BUFFETT / MUNGER / KLARMAN SUMMARY
1 SNAPSHOT
Price$17.75
Market Cap114.4B
2 BUSINESS

Petrobras at $17.75 looks irresistibly cheap - 5.7x earnings, 4x EV/EBITDA, 26% ROE, $36B operating cash flow from genuinely world-class low-cost pre-salt barrels. The catch is that you do not own an oil company; you own a minority stake in a Brazilian state instrument. The federal government controls 50.3% of the vote (~69% with its development banks) and has, within living memory (2011-2014), forced the company to subsidize fuel below import parity, destroying tens of billions and driving net debt to ~$130B. Management's "we don't pass volatility to the domestic market" policy is shareholder-friendly when oil falls but a value transfer to consumers when oil rises. The dividend that makes the stock look like a high-yield bargain is cyclical: it was $5.93/ADR in 2022 ($100 oil) and just $0.84/ADR in 2025, with management guiding extraordinary dividends as "pretty low" probability. My through-cycle dividend-discount, reserves-based, and distributable-FCF methods converge on $12-16/ADR fair value. After a +57% twelve-month rally the stock trades above that range with no margin of safety against the political tail, heading into a 2026 election. Howard Marks's new 2.93% Oaktree position is a credible carry/contrarian bet sized for a diversified book - but it is not a reason for a concentrated value investor to pay up. Wait for the price.

3 MOAT NARROW

Structurally low-cost pre-salt deepwater (sub-$6/bbl lifting on best fields), dominant Brazilian offshore infrastructure, privileged acreage via Transfer-of-Rights; but rents are extracted and pricing can be overridden by the controlling state.

4 MANAGEMENT
CEO: Magda Chambriard

Disciplined operationally (capital discipline, projects tested at $50 Brent, $65B debt ceiling) but ultimately serves the controlling government - dividends, pricing, and strategy are state-directed, not minority-shareholder-first.

5 ECONOMICS
28.6% Op Margin
15% ROIC
26.5% ROE
5.7x P/E
16.7B FCF
79.8% Debt/EBITDA
6 VALUATION
FCF Yield14.6%
DCF Range12 - 16

Overvalued by ~10-25% vs $12-16 through-cycle fair value; price embeds peak-cycle economics after +57% 1yr rally.

7 MUNGER INVERSION
Kill Event Severity P() E[Loss]
State control: the Brazilian federal government owns 50.3% of voting stock (~69% with BNDES/BNDESPar) and can force below-parity domestic fuel pricing, redirect dividends, and set strategy - a 2014-style value destruction is the dominant tail risk. HIGH - -
Commodity cyclicality - net-debt-neutral Brent breakeven $59/bbl; distributions collapse below ~$55 oil. MED - -
8 KLARMAN LENS
Downside Case

State control: the Brazilian federal government owns 50.3% of voting stock (~69% with BNDES/BNDESPar) and can force below-parity domestic fuel pricing, redirect dividends, and set strategy - a 2014-style value destruction is the dominant tail risk.

Why Market Right

Government forces below-import-parity domestic fuel pricing (2014 replay); Ordinary dividend redirected toward CapEx, jobs, Braskem, or energy-transition spend; Sustained sub-$55 Brent pushes net debt above ceiling and cuts distributions toward zero

Catalysts

Production ramp: +11% in 2025, new FPSOs (P-78, P-80, P-82, P-83) and Buzios/Mero growth into 2027; Market-friendly outcome in the 2026 Brazilian general election could re-rate governance discount; Sustained $75+ Brent would trigger extraordinary dividends under the 45%-of-FCF policy; Foz do Amazonas (equatorial margin) exploration optionality - Guyana-style upside if successful

9 VERDICT WAIT
B Quality Moderate - net debt $60.6B (~1.4x EBITDA); 62% of $69.8B gross debt is FPSO/vessel leases tied to producing assets; pure finance debt only $26.4B. Improved vs 2015 crisis but cyclically exposed.
Strong Buy$10
Buy$13
Fair Value$16

Wait. Accumulate only below $13; Strong Buy below $10. No position at $17.75 - the political tail is un-discounted and economics are peak-cycle.

🧠 ULTRATHINK Deep Philosophical Analysis

PBR - Ultrathink Analysis

The Real Question

The seductive question is "how is one of the world's great oil companies trading at 4x EBITDA?" That question has an easy, wrong answer: "it's a bargain." The real question is harder and uncomfortable: when you buy PBR, whose company are you actually a part-owner of, and does that owner share your goals?

Petrobras is not a business that happens to have a large government shareholder. It is a government that happens to operate a business. The Brazilian state owns 50.3% of the votes and ~69% counting its banks; it appoints the CEO; it sets fuel prices; it decides what fraction of $36B of cash flow becomes your dividend versus a jobs program, a refinery, or a fuel subsidy for Brazilian truckers before an election. The geology is spectacular. The cash generation is real. But the question that determines your return is not "how good are the barrels?" — it is "what is the probability that the controlling owner chooses, over my five-year holding period, to convert those barrels into my wealth rather than its political objectives?" Everything else is a distraction from that single conditional probability.

Hidden Assumptions

The market's hidden assumption: that the current shareholder-friendly regime — capital discipline, the 45%-of-FCF dividend formula, no fuel subsidies — is a property of the company rather than a property of the current political moment. It is the latter. The exact same legal entity, the exact same pre-salt fields, paid almost nothing to shareholders and nearly bankrupted itself subsidizing fuel just eleven years ago. Nothing structural prevents a return to 2014. The dividend policy is a board resolution, not a covenant; the board serves at the pleasure of the federal government.

The second hidden assumption is that the trailing dividend yield is the forward dividend yield. It is not. The $5.93/ADR of 2022 was $100 oil plus a market-friendly government — a once-in-a-decade alignment. Management itself, on the record, calls extraordinary dividends "pretty low" probability at today's prices. The yield the screen shows you already happened.

My own assumption I must check: that I have correctly priced the political risk at a 14% cost of equity. If the 2026 election produces a durably market-friendly, fiscally orthodox government — and Brazil's pendulum does swing — then my Ke is too high by 300+ bps and my fair value is too low. I am consciously betting that the distribution of Brazilian political outcomes is wide and left-skewed, not that the bad outcome is certain.

The Contrarian View

Steelman Howard Marks. For the bulls to be completely right, three things must hold simultaneously, and they are not crazy:

  1. The carry is the thesis, not the re-rating. You don't need the multiple to expand. At 4x EBITDA with $36B OCF and only $26B of true financial debt, even a mediocre dividend is a high-single-digit cash coupon, and you are paid to wait. Bonds don't need to appreciate to compound.
  2. Oil stays structurally bid. Years of under-investment by Western majors, OPEC discipline, and pre-salt's low cost mean Petrobras prints cash across a wide band of Brent. The downside case ($50 oil) still produces positive FCF; the company survives scenarios that kill marginal US shale.
  3. Brazil's politics mean-revert toward orthodoxy. Fiscal reality eventually disciplines every government; a heavily indebted state needs Petrobras dividends and tax flows, which actually aligns the controlling shareholder with cash distribution more than the bears admit. The state is a dividend-hungry owner, not a dividend-hostile one.

If all three hold, $17.75 is cheap and Marks looks prescient. The bear case requires only one of them to break — and the political one has broken before, hard. That asymmetry of "bulls need three, bears need one" is precisely why this is a WAIT and not a REJECT: the bull case is genuinely live, just not yet priced with a margin of safety.

Simplest Thesis

A world-class oil asset wrapped in a state-control structure is worth $12-16 through-cycle, and after a +57% rally to $17.75 you are paying for the asset while getting the wrapper's risk for free — so wait for the price, don't pay for the story.

Why This Opportunity Exists

The mispricing — in both directions — exists because two very different investor populations look at PBR and see incompatible things, and neither is wrong.

The quant/screen population sees 4x EBITDA, 26% ROE, 14% FCF yield and flags it as deep value; they systematically under-weight governance because it doesn't fit in a spreadsheet cell. They have bid the stock up 57% on a real production story.

The governance-aware population sees a state instrument and applies a permanent, large discount — sometimes too large, which is the genuine opportunity Marks is exploiting. The discount is correct in direction but the magnitude oscillates with the political cycle. It widens to absurdity in crisis (2015-2016, sub-$4) and compresses to complacency near elections-of-hope.

The opportunity, then, is not "PBR is cheap." It is "PBR's governance discount is mean-reverting and election-driven, and the way to win is to buy when the discount is at its widest — in a political panic or oil trough — not after a rally has compressed it." Today the discount has compressed. The edge is gone. The opportunity will return — it always does with this name — at a lower price.

What Would Change My Mind

Concrete, falsifiable triggers that would move me from WAIT to ACCUMULATE/BUY at or near current prices:

  1. Price re-rates the risk for me: PBR trades at or below $13 (accumulate) or $10 (strong buy) on an oil or political pullback. This is the cleanest trigger — same company, different price.
  2. Structural governance improvement: a binding change that protects minorities — e.g., the dividend policy is hard-wired into bylaws with supermajority-to-change protection, or a credible privatization/control-dilution path emerges. (Low probability, but it would re-rate the whole story.)
  3. A market-friendly 2026 election result with an early, concrete signal — a finance minister and Petrobras board that reaffirm import-parity pricing and the dividend formula under fiscal pressure — would justify cutting my cost of equity below 14% and lifting fair value toward $18-20.
  4. Proof the dividend is durable at low oil: two-plus consecutive years of meaningful ordinary dividends at sub-$65 Brent without fuel subsidies would falsify my "yield is cyclical" claim.

Conversely, what confirms the bear and would push me toward REJECT: any reintroduction of below-import-parity domestic fuel pricing. That single act tells you the controlling shareholder has reverted to type, and no multiple is cheap enough to compensate.

The Soul of This Business

The soul of Petrobras is a contradiction it can never resolve: it is simultaneously a commercial champion and a public utility of the Brazilian state, and these two souls fight for the same cash flow. In the pre-salt — in Búzios and Tupi and Mero — Petrobras is one of the finest deepwater operators humanity has produced; oil columns "as tall as Pão de Açúcar," barrels lifted for less than $6, a reserve base growing 175% of what it pumps. That is the soul of an engineer, and it is magnificent.

But the other soul lives in Brasília. It is the soul that paid BRL 277 billion in taxes and royalties, that counts "300,000 jobs" as a deliverable, that froze diesel prices for 300 days "for society," that turns over its entire C-suite every time the presidency changes hands. To the Brazilian state, Petrobras is not an investment to be optimized — it is an organ of the nation, an instrument of industrial policy, energy security, and electoral peace.

A minority ADR holder owns a thin, junior, foreign-currency, withholding-taxed claim on the first soul, exercised entirely at the discretion of the second. When the two souls are aligned — friendly government, high oil — you get 2022 and a $5.93 dividend. When they conflict — populist government, inflation panic — you get 2014 and ruin. The business is not fragile in its rocks; it is fragile in its governance, and that fragility is invisible on the day you buy and decisive on the day you wish you hadn't. The disciplined investor's response is not contempt and not infatuation, but patience: this great, compromised asset goes on sale every few years when its political soul frightens the market. Wait for that sale. It is coming. It is just not here at $17.75.

1. Executive Summary

Three-sentence thesis. Petrobras is a world-class, low-cost pre-salt oil producer generating $36B of operating cash flow and a 26% ROE, trading at ~5.7x earnings and ~4x EV/EBITDA — optically one of the cheapest large-cap energy companies on earth. But the buyer of PBR is not buying an oil company; they are buying a minority stake in a Brazilian state instrument whose dividends, fuel pricing, capital budget, and strategy are controlled by a federal government that owns 50.3% of the voting stock and ~69% with its development banks. The stock has already re-rated +57% in twelve months and now trades above my through-cycle fair value of $12–16/ADR, with the headline "dividend yield" being a cyclical peak that the company itself signals is unlikely to repeat at $50–65 Brent — so despite Howard Marks's new position, the disciplined verdict at $17.75 is WAIT, not buy.

Metrics dashboard (FY2025, USD)

Metric Value Note
Revenue $90.8B -0.7% YoY; Brent avg $69 (-14%)
Net income $20.1B +168% YoY (2024 hit by FX/impairment)
Adjusted EBITDA $42.5B ex-exceptionals; reported $43.8B
Operating cash flow $36.6B flat YoY despite -14% Brent
CapEx $19.8B rising as production ramps
Free cash flow $16.7B 5yr avg $28.6B
ROE (FY2025 / 5yr avg) 26.5% / 30.5% exceptional, but cyclical
Gross debt $69.8B of which $43.4B leases (62%), finance debt $26.4B
Net debt $60.6B 1.4x EBITDA
Proved reserves ~10.9 Bn boe replacement ratio 175%
EV/EBITDA 4.1x P/E 5.7x, P/B 1.5x
FCF yield 14.6% (FY25) only ~45% reaches shareholders by policy

Verdict

WAIT. Quality grade B (an A-grade asset base trapped inside a C-grade governance and capital-allocation structure). Strong Buy below $10; Accumulate below $13. Current price $17.75 offers no margin of safety against the central risk — political control of cash flows — and embeds peak-cycle economics.

2. Business Model (primary source: 20-F FY2025)

Petrobras is an integrated national oil company:

  • Exploration & Production (E&P): ~84% of 2025 CapEx. The crown jewel is the pre-salt offshore Santos Basin (Búzios, Tupi, Mero, Sépia, Atapu) — some of the lowest-cost, highest-productivity deepwater barrels in the world. 2025 oil production grew +11%; Búzios alone passed 1 MMbbl/d. Total production reached ~4.2 MMboe/d. Lifting costs in the pre-salt are among the lowest globally (sub-$6/bbl on the best fields).
  • Refining, Transport & Marketing (RTM): Refinery utilization 91%, 68% high-value derivatives, 70% of throughput from pre-salt crude.
  • Gas & Power: Boaventura complex, Rota 3 pipeline; gas supply to market +15% in 2025.

The economic engine is genuinely excellent. The problem is who controls the cash it produces.

3. Phase 1 — Risk Analysis (Inversion)

Munger's question: how could this investment destroy capital? For PBR, the dominant risks are political/governance, not geological.

3.1 Political / state-control risk (THE central risk) — HIGH

  • The Brazilian federal government owns 50.26% of voting (common) stock directly and, with BNDES + BNDESPar, controls ~69% of the vote (20-F FY2025, Shareholder Information). It elects the board and appoints management.
  • History is unambiguous: under the 2011–2014 Rousseff administration, Petrobras was forced to subsidize domestic fuel below import parity to fight inflation, destroying ~$40B+ of value and helping drive net debt to a peak of ~$130B. The Lava Jato (Car Wash) corruption scandal followed.
  • Current management (CEO Magda Chambriard, appointed 2024 by the Lula government) insists the pricing policy "does not pass volatility to the domestic market" — a euphemism that, in a falling oil market (2025), helped margins, but in a rising market (the Middle East spike management discussed) means the state can again force below-parity domestic fuel prices and absorb the loss on shareholders' behalf.
  • P(material value-destructive political intervention over 5 yrs) ≈ 35–45%; impact -30% to -50%. Expected loss ≈ -12% to -18%. This single factor dominates the risk register.

3.2 Dividend-policy / capital-allocation risk — HIGH

  • The dividend is formulaic (45% of FCF) only when gross debt sits below the ceiling and when the controlling shareholder wants cash. The 2022 mega-dividend ($5.93/ADR) was a function of $100 oil and a market-friendly Bolsonaro-era management; 2025 ordinary dividends collapsed to ~$0.84/ADR as oil fell and CapEx rose.
  • Management explicitly guided that extraordinary dividends are "pretty low" probability at current prices and that surplus cash goes to investment first, then debt, then maybe dividends (Q2/Q4 2025 calls). The high trailing yield is a rear-view mirror.
  • The government's appetite to redirect FCF toward CapEx, jobs ("300,000 jobs"), Braskem/petrochemicals, energy-transition projects, and the Foz do Amazonas frontier competes directly with the minority shareholder's claim. P ≈ 60%; impact -15%. Expected ≈ -9%.

3.3 Commodity / cyclical risk — MODERATE-HIGH

  • Earnings swing violently with Brent: net income was $36.6B (2022, $100 oil) → $7.5B (2024) → $20.1B (2025). The 5.7x P/E is on near-peak normalized earnings.
  • Net-debt-neutral breakeven is $59/bbl (2026) per the 20-F business plan — meaning below ~$59 Brent, net debt rises and dividends shrink toward zero. A sustained $45–50 oil world (recession + OPEC supply) would gut distributions. P(2+ yrs sub-$55 Brent over 5 yrs) ≈ 30%; impact -35%. Expected ≈ -10%.

3.4 Balance-sheet / leasing risk — MODERATE

  • Headline gross debt $69.8B looks scary, but 62% ($43.4B) is FPSO/vessel/rig lease liabilities tied directly to producing assets that generate the cash to service them. Pure financial debt is only $26.4B. Net debt $60.6B = ~1.4x EBITDA. This is manageable, not fragile — a genuine improvement vs the 2015 crisis. Still, leases are real obligations and rising as new platforms come online. P(distress) low ≈ 8%; impact -40%. Expected ≈ -3%.

3.5 Legal / environmental / ESG — MODERATE

  • The Foz do Amazonas (equatorial margin) frontier drilling is politically charged environmentally; a spill or licensing reversal carries headline and capital risk.
  • Residual Lava Jato / governance-quality perception keeps the equity-risk premium permanently elevated. P ≈ 20%; impact -15%. Expected ≈ -3%.

3.6 FX / repatriation — MODERATE

  • Cash flows are BRL-linked; ADR holders bear real/USD risk plus Brazilian withholding tax on dividends. A BRL devaluation cuts USD distributions even if BRL earnings hold.

Aggregate expected downside (additive, illustrative): ≈ -40% of expected value before considering correlation — these risks cluster (a left-wing fiscal squeeze raises political, dividend, AND FX risk simultaneously). The tail is a repeat of 2014.

4. Phase 2 — Financial Analysis

4.1 Returns (DuPont, USD)

  • ROE FY2025 26.5%, 5yr avg 30.5% — outstanding. Decomposition: net margin ~22%, asset turnover ~0.41, equity multiplier ~2.9 (the leverage flatters ROE).
  • ROIC (NOPAT/invested capital): with op income $26B, ~21% tax-effected NOPAT ~$20B over ~$135B invested capital ≈ **15%**, comfortably above a Brazil-state WACC I estimate at ~11–12% (US risk-free 4.3% + Brazil/country + state-control premium). The spread is positive but thinner than the ROE headline implies once you charge the full cost of capital for state risk.

4.2 Owner earnings (through-cycle)

5-year FCF: $31.5B (2021) → $40.1B (2022) → $31.1B (2023) → $23.3B (2024) → $16.7B (2025). The downtrend reflects falling oil and rising CapEx as the production-growth plan kicks in. Normalized through-cycle FCF ≈ $18B at ~$65 mid-cycle Brent with $18–20B CapEx.

4.3 My valuation (three methods, all per ADR)

A) Dividend-discount (the right lens — minority owns only the distributed cash). Cost of equity for a Brazilian state oil minority stake: Ke ≈ 14% (4.3% RF + ~7% Brazil ERP + ~3% governance/control premium). Through-cycle distributions (ordinary + occasional extraordinary) ≈ $11–13B/yr.

  • $11B, g=0% → $12.2/ADR
  • $12B, g=0% → $13.3/ADR
  • $13B, g=1% → $15.7/ADR

B) Reserves-based sanity (10.9 Bn boe proved). State oil majors trade at deep discounts to IOCs ($12–15/boe); applying $8–10/boe and subtracting $60.6B net debt:

  • $8/boe → $4.1/ADR; $10/boe → $7.5/ADR. (Conservative; penalizes heavily for net debt + control.)

C) FCF-yield cross-check. At $17.75, trailing FCF yield is 14.6%, but only ~45% is distributable by policy → distributable yield ~6–7%, fair for the risk but not cheap.

Fair-value range: $12–$16/ADR (central ~$14). Current $17.75 is ~10–25% above fair value.

4.4 Sensitivity

Brent Norm. FCF 45% Div Div/ADR Implied value @14% Ke
$50 (bear) ~$7B $3.1B $0.49 ~$7
$65 (base) ~$15B $6.8B $1.05 ~$13–14
$80 (bull) ~$25B $11.2B $1.75 ~$18–20

The current price only makes sense if you underwrite sustained $80 Brent AND a permanently shareholder-friendly government — two optimistic assumptions stacked together.

5. Phase 3 — Moat Analysis

Moat type: Cost advantage + regulatory/resource position. Width: NARROW (asset moat is real; corporate moat is compromised).

  • Genuine cost moat: pre-salt is structurally low-cost (sub-$6/bbl lifting on best fields), and Petrobras has the dominant operating position and infrastructure in Brazilian deepwater. Reserve replacement of 175% and +1.7 Bn boe added in 2025 show the resource engine is widening, not depleting.
  • Regulatory moat (double-edged): the Transfer-of-Rights regime and incumbency give privileged access to Brazilian acreage — but the same state that grants the moat also extracts the rents (BRL 277B in taxes/royalties in 2025) and can override pricing.
  • No pricing-power moat at the corporate level: in a commodity business, the only durable edge is being lowest-cost, which Petrobras is — but that edge accrues to the state, not reliably to the minority shareholder. Durability of the asset moat: 15–20+ years. Durability of shareholder value capture: election-cycle-dependent (≤4 years of visibility).

6. Phase 4 — Synthesis

6.1 The Howard Marks signal — honored, then weighed

Oaktree's new 2.93% Q1 2026 position (his largest buy) is a serious contrarian endorsement from a distressed-value master who specializes in "you can't lose money in a cash-flow if you buy it cheap enough." Marks may be playing the carry: even a depressed dividend on a $26B-net-financial-debt, $36B-OCF machine at 4x EBITDA offers a fat coupon with optionality on oil and a future market-friendly government. That is a coherent thesis. But Marks sizes for a portfolio of asymmetric bets and can hold through a 50% drawdown; for a concentrated value account demanding margin of safety, buying after a +57% run, above my fair value, with the central political risk un-discounted, is not the same trade. I respect the signal; I do not chase the price.

6.2 Expected-return tree (5-yr, per ADR from $17.75)

  • 30% — Friendly govt + $75+ Brent → fat dividends, re-rate to $22–26: +30–50%
  • 40% — Muddle-through, $60–70 Brent, ordinary dividends → roughly flat to +15% incl. yield
  • 30% — Political intervention / sub-$55 Brent / dividend cut → -30% to -50%
  • Probability-weighted ≈ low-single-digit % annualized — inadequate compensation for the volatility and tail.

6.3 Position sizing & entry

  • Strong Buy: ≤ $10 (where reserves + base-case dividends give a genuine margin of safety and ~10%+ distributable yield).
  • Accumulate: ≤ $13 (lower edge of fair value).
  • Current $17.75: WAIT. Target allocation only 1–2% even on a pullback, given the binary political tail.

6.4 Monitoring triggers

  1. Pricing policy — any move to below-import-parity domestic fuel pricing = exit signal (2014 replay).
  2. Dividend policy formula — change to the 45%-of-FCF rule or a cut to ordinary dividend.
  3. Gross-debt ceiling — breach of the $65B target or net-debt-neutral breakeven creeping above $65 Brent.
  4. CapEx discipline — business plan CapEx inflating materially above ~$20B/yr or non-core (Braskem capital injection, social spend) acquisitions.
  5. Brazilian electoral cycle (2026 general election) — composition of the next government is the single biggest value swing.
  6. Brent — sustained sub-$55 for >2 quarters compresses distributions toward zero.

7. Risk Register (summary)

# Risk P(event) Impact Expected loss
1 Political pricing intervention 35–45% -30% to -50% ~-15%
2 Dividend redirected to CapEx/state goals 60% -15% ~-9%
3 Sustained sub-$55 Brent 30% -35% ~-10%
4 Leasing/debt strain 8% -40% ~-3%
5 Foz do Amazonas / ESG / legal 20% -15% ~-3%
6 BRL devaluation cuts USD dividends 35% -15% ~-5%

8. Conclusion

Petrobras is the rare case where the asset is A-grade and the ownership wrapper is C-grade, netting to a B-quality investment. At a genuinely distressed price it is a superb risk/reward — Howard Marks is not wrong to own some. But at $17.75, after a 57% rally, above a $12–16 fair-value range, with peak-cycle economics and an un-discounted political tail heading into a 2026 Brazilian election, the margin of safety is gone. The right action is patience: WAIT for ≤$13 to accumulate, ≤$10 to back up the truck. You are paid to wait for the price, not to pay up for the story.

Primary sources: SEC Form 20-F FY2025/2024/2023 (CIK 0001119639); Q4 2025 and Q2 2025 earnings calls (management transcripts); AlphaVantage normalized USD financial statements; daily price history 2021–2026. Full file list in data/SOURCE_CHECKLIST.md. No sell-side research used.