Marathon Petroleum Corp MPC
Quantitative scorecard
Thesis
Thesis
Marathon Petroleum is a high-quality cyclical, not a compounder. It is the largest US independent refiner by capacity (~3M bpd across 13 plants) and owns ~64% of MPLX, a logistics MLP that provides a steady toll-road cash stream inside an otherwise commodity P&L. Capital allocation under Mike Hennigan and now Maryann Mannen has been textbook for a cyclical: divest Speedway at the top of the retail multiple cycle (2021, $21B to 7-Eleven), keep MPLX as a yield engine, and aggressively retire shares.
The numbers that matter, taken at face value from the scorecard:
- 10y average ROIC: -0.78%. Composite 70 is propped up by capital allocation (20/25) and current valuation (17/25), not by underlying business quality.
- 5y FCF conversion: 1.29x - earnings understate cash generation, typical of a mid-cycle refiner with depreciation tailwinds.
- Net debt/EBITDA: -0.13x (net cash). Interest coverage 5.0x.
- P/E TTM 24.4 vs 10y avg 18.3. Shares are not cheap on earnings, but EV/FCF of 13.5 and P/IV 0.90 suggest the market is part-way to pricing a normalization.
- Reverse-DCF implied growth: 4.2%. Plausible for MPLX but a stretch for the refining segment.
Why I am not buying: the ten-year-look-forward test fails. In 2036 the US light-duty fleet will be materially more electrified, California/Washington/Northeast carbon regimes will be tighter, and Mexican/Indian refining capacity will likely have grown. Mid-cycle gasoline crack spreads in 2036 are genuinely uncertain. A business whose 10y average ROIC is already negative does not deserve a compounder allocation when the next decade may be tougher than the last.
Where I could be wrong: US refining capacity has fallen ~1M bpd since 2019 with no greenfield builds since 1977; if attrition continues faster than demand decline, normalized margins are structurally higher than the 2010s. MPLX is a real $40B+ midstream business that arguably deserves a tech-multiple-style sum-of-the-parts treatment. At $180 (P/IV ~0.65) I would reconsider as a cyclical position, not a core hold.
Moat
Moat Assessment - NARROW (cost advantage only)
Pricing power: NONE. Refining sells gasoline, diesel, and jet at globally-determined crack spreads. MPC cannot raise prices; it can only pull throughput up or down at the margin.
Switching costs: NONE. Wholesale buyers (gas stations, airlines, military) buy on price and reliability. After Speedway was sold, MPC has no captive retail.
Network effects: NONE in refining. MPLX has modest network density advantages in the Permian and Marcellus.
Intangibles: NONE meaningful. Refining permits are valuable but not a brand.
Cost advantages: NARROW and real.
- Scale: 3M bpd is ~16% of US refining capacity. Largest single refinery (Galveston Bay) is ~593k bpd, complexity index ~12+, can run heavy/sour discounted crudes. Garyville (~597k bpd) is similarly advantaged.
- Gulf Coast logistics: integrated MPLX pipelines feed crude in and product out at low marginal cost.
- Permit moat: the last greenfield US refinery was built in 1977. New entry is effectively impossible. Existing Nelson Complexity assets cannot be replicated.
$10B / 5-year stress test: A $10B competitor cannot build a new refinery (permitting alone takes a decade and would face guaranteed litigation). They could buy a refinery from PBF or HF Sinclair, but that does not erode MPC. The cost moat is durable against new entry but does NOT protect against margin compression from demand decline or imports.
Erosion risks: EV adoption, California/Washington LCFS regimes, biofuels/RD displacing diesel, Saudi/Indian/Chinese refining export capacity. These hit revenue, not the cost position.
Verdict: NARROW. The moat protects against new supply but not against demand erosion or cyclical price collapse. A NARROW moat in a commodity business is not a compounder moat - it is a license to earn cyclical returns with a floor.
Management & Capital Allocation
Management & Capital Allocation - Grade B+
The five capital choices:
-
Reinvest in the business: Disciplined. Maintenance capex ~$1.5B/yr; growth capex concentrated in MPLX midstream (Northwind, Whiptail, BANGL acquisitions in 2024-2025). Returns on incremental refining capacity are deliberately low - they are not building new refineries, they are doing high-return debottlenecks.
-
Acquire: Mostly bolt-ons through MPLX (Northwind Midstream Aug 2025, BANGL pipeline 2025, Whiptail). Disposed of Rockies operations Nov 2025. Speedway divestiture to 7-Eleven (Aug 2021, $21B cash) at ~14x EBITDA was an A+ trade - sold the cyclical-low-multiple retail business at a peak retail multiple.
-
Debt: Net cash position (net debt/EBITDA -0.13x). Interest coverage 5.0x. MPLX carries its own investment-grade leverage. Conservative for a cyclical.
-
Buybacks: This is the headline. Share count down ~4.8% over 10y on the scorecard, but the move post-Speedway is more dramatic - shares outstanding fell from ~650M in 2020 to ~310M today, roughly 50% retirement. Average price paid has been below current $246, but well above 2020 lows. They have NOT been opportunistic at the trough; they have been mechanical. Average P/IV when buying: roughly 0.85-1.00. Acceptable, not exceptional.
-
Dividends: ~$3.40/yr, ~1.4% yield. Modest, has grown steadily. Right call for a cyclical - keep dividend small, return cash via buybacks.
Communication: Investor presentations are clear about mid-cycle vs trough vs peak EBITDA bands. Mannen (CEO since Aug 2024) and the team set explicit capital return targets (50% of operating cash flow). Refreshingly free of energy-transition theater.
Grade: B+. The Speedway sale is one of the better corporate trades of the last decade. Buybacks have been steady but not opportunistic. Fails to qualify for an A because (a) average buyback P/IV is ~0.9, not 0.6, and (b) the underlying business is not the kind where great capital allocation creates compounding - it just returns cash.
Industry Structure
Industry Structure - Porter's Five Forces - Average
Rivalry: HIGH within US. VLO, PSX, MPC, PBF, HF Sinclair, DK, CVI all run similar kits. Differentiation is plant location, complexity, and crude slate flexibility. Margins are commodity.
Buyer power: MODERATE. Wholesale gasoline/diesel buyers are price-takers in aggregate but big airlines and trading houses extract concessions on jet fuel.
Supplier power: HIGH AND VOLATILE. OPEC+ and US shale producers control crude price; refiners take the spread. Gulf Coast access to discounted heavy crude (Mexico, Venezuela, Canada) is the key competitive variable.
Threat of new entrants: VERY LOW. No new US greenfield refinery since 1977. Permitting + capex + ESG headwinds make new builds effectively impossible. This is the strongest force in MPC's favor.
Threat of substitutes: HIGH AND RISING. EVs for gasoline (long-tail), renewable diesel and biofuels for diesel (already material in California), SAF for jet (early stage). Plus product imports from Saudi Aramco's Jazan, Reliance, Chinese teapots.
Value pool location: Mostly in the crude-to-products spread (crack), which is volatile. Secondary pool in the midstream toll-road business (MPLX). Tertiary pool in renewable diesel (Martinez and Dickinson conversions) - low-IRR optical investments to keep regulators happy.
Trajectory: US refining capacity has retired ~1M bpd since 2019 (LyondellBasell Houston, Phillips Rodeo, Lyondell Houston, others). If attrition continues and demand falls more slowly than supply, mid-cycle spreads structurally widen. If demand falls faster than supply rationalizes, spreads compress.
Verdict: AVERAGE. Strong barriers to new entry and rationalizing supply support cycle floors. Substitution and demand erosion put a ceiling on long-run growth. This is a 'rent your seat at the table' industry, not a 'own a piece of the future' industry.
Inversion (Bear Case)
MANDATORY INVERSION - The Bear Case
The single event that kills this: A major refinery incident at Galveston Bay or Garyville. BP Texas City (2005) killed 15 people, cost BP ~$3B in fines plus settlements, and crystallized the divestment that produced MPC's current Galveston Bay asset. A similar event at one of MPC's mega-complexes would: (a) take 200-600k bpd offline for 12-24 months, (b) drive $2-5B in remediation and litigation, (c) trigger management turnover and EPA consent decrees that constrain operations for a decade. One bad day erases 2-3 years of FCF and resets the buyback program.
Why the moat is narrower than bulls think: The 'cost advantage' is relative, not absolute. Valero (VLO) operates an arguably better Gulf Coast portfolio with higher Nelson complexity per dollar of EV. Phillips 66 has a better midstream/chemicals mix. PBF runs the best East Coast complex. MPC's edge is scale, not unit economics - and scale in a commodity earns commodity returns. The proof is in the scorecard: 10-year average ROIC of NEGATIVE 0.78%. No moat business produces that number. The cost advantage merely keeps MPC from being one of the casualties; it does not make it a compounder.
Why management is worse than it appears: The Speedway sale was excellent, but it was a one-time portfolio cleanup, not a repeatable skill. Buybacks at 0.9 P/IV are good capital allocation only relative to peers; a true value-allocator buys back near 0.6 P/IV and lets cash build at 1.2x. MPC's program is mechanical, formula-driven (50% of OCF), not opportunistic. The Northwind Midstream and BANGL acquisitions in 2025 were done at full cycle prices with limited synergy disclosure. CEO Mannen took over in August 2024 with no track record of allocating capital through a downturn.
What bulls are extrapolating that won't hold:
- Mid-cycle crack spreads of $20+/bbl. The 2022-2024 spreads were post-Russia, post-COVID-rebound anomalies. 2010-2019 average was closer to $12-14/bbl.
- US gasoline demand flat through 2030. EV penetration of new car sales is now 8%+ and accelerating; once EV stock crosses ~25% of fleet, gasoline demand falls non-linearly.
- MPLX standalone valuation of $40B+. MLPs trade at 8-10x EBITDA in good times and 6-7x in bad; the segment is not immune to oil & gas capex declines.
- Buyback flywheel continues at 5-7% per year. At trough cash flow it cannot.
Valuation trap (multiple compression / regime change): TTM P/E of 24.4x is not a 'cheap stock' multiple - it implies depressed earnings. If earnings normalize HIGHER, the multiple compresses (typical refining trap). If earnings collapse to trough ($1-1.5B FCF), the multiple expands but the stock falls 40-50%. The IV low of $132 reflects this scenario. Bulls are double-counting: paying a peak-multiple on peak-ish earnings while assuming a structural normalization that argues for trough multiple on peak earnings.
If I am right, the stock could be worth $130-150 within 3 years. Mechanism: 2026-2027 cycle trough drops crack spreads to $10-12, EBITDA halves, FCF falls to $1.5B, market re-rates to 6-7x EV/EBITDA cyclical-trough multiple. From $246 that is a 40-45% decline, with a fast recovery on the other side - exactly what you would expect from a high-quality cyclical, NOT a compounder.
Lollapalooza Bias Check
Lollapalooza Bias Check (active biases in me, the analyst)
Anchoring (HIGH): I am anchoring on the IV base of $272 versus current $246 - 'only' a 10% discount. The IV range from $132 to $346 is enormous (2.6x spread), and treating the midpoint as the answer ignores the asymmetry of cyclical valuation. A more honest read is 'this stock could rationally be worth anywhere from $132 to $346 depending on which point in the cycle prevails over the holding period'.
Recency bias (HIGH): 2022-2024 refining margins were spectacular due to the Russia-Ukraine refined-product disruption and post-COVID demand snapback. My instinct is to extrapolate. The 2010-2019 mid-cycle was much more pedestrian, and the scorecard's NEGATIVE 10y ROIC reflects that more sober reality.
Authority/social proof (MODERATE): Buffett owns Occidental and Chevron, so 'energy is investable' has authority cover. But Buffett does not own refiners - he owns the upstream optionality on US energy independence. That is a different thesis.
Confirmation (MODERATE): I started this writeup leaning Pass because of the negative ROIC, and I have to actively look for reasons the bull case might be right (capacity retirements, MPLX optionality, Speedway-like trades).
Commitment / consistency (LOW): No prior position.
Deprival super-reaction (LOW): Not relevant - I do not own MPC.
Incentive bias (LOW for me, HIGH for sell-side): Sell-side analysts have incentive to publish optimistic mid-cycle EBITDA assumptions to justify Buy ratings. Their consensus 'mid-cycle' EBITDA is roughly 30-40% above the 2010-2019 actual average. I should weight scorecard ROIC over sell-side normalization.
Net effect: Anchoring + recency are pulling me toward a Buy rating I should not give. The discipline is to weight the 10-year ROIC and the inversion case more heavily than the 'P/IV 0.90' headline.
10-Year Outlook
10-Year Outlook Test
Same fundamental business model in 2036? Mostly yes for refining, definitely yes for MPLX midstream. But the customer mix shifts: less gasoline (EV penetration), more diesel/jet/petrochemicals/exports. Refining is structurally lower-volume but possibly higher-margin per barrel.
Customer base larger? No. US gasoline demand likely peaks in the late 2020s and declines 1-2% per year through the 2030s. Diesel holds up longer (trucking is harder to electrify). Jet fuel grows with air travel. Net: refined product volume probably 10-15% lower in 2036.
Profit per customer higher? Possible if capacity retires faster than demand. Possible if MPC successfully tilts product slate to diesel/jet/petchem/SAF. Not the base case.
Moat wider? No. The cost-advantage moat is roughly stable. The new-entrant moat actually widens (no one is building refineries) but that does not help margins, only floors them.
Single biggest threat: Faster-than-expected EV adoption combined with a recession that crushes industrial diesel demand simultaneously. Spreads compress to 2009 levels for 2-3 years.
Other threats ranked:
- EPA / state-level carbon regulations (California LCFS expansion to other states)
- Imports from Reliance, Saudi Aramco Jazan, Chinese teapot exports
- Major refinery incident with EPA consent decree
- MPLX MLP structure tax disadvantages if Congress changes K-1 treatment
Confidence: MEDIUM-LOW. The combination of energy transition uncertainty, commodity price volatility, and regulatory regime risk makes a 10-year forecast genuinely speculative. Per the brief's rule, MEDIUM-LOW confidence pushes recommendation toward 'Pass' or 'Too Hard'. I am going with Pass rather than Too Hard because the math at trough ($132 IV) is at least bounded, and the business is understandable. But this is right on the edge.
Position guidance
## Position Guidance **Recommendation: PASS** (do not initiate as a compounder position). **If you must own it:** - **Sizing:** maximum 1-2% of portfolio. This is not a core compounder hold. - **Entry:** below $180 (P/IV ~0.65) gives a margin of safety against cycle trough scenarios. At $246 (P/IV 0.90) the margin of safety against the bear case is essentially zero. - **Exit:** trim above $310 (P/IV >1.10) regardless of narrative. Refining stocks are bought at trough multiples on trough earnings and sold at peak multiples on peak earnings; the inverse is a value trap. - **Hedge:** if held in a portfolio with material upstream energy exposure (XOM, CVX, OXY), MPC is partially redundant - all of it is correlated to oil prices. Pair against a midstream MLP only if you specifically want refining margin exposure. **Better alternatives for a Buffett-Munger compounder slot:** - For energy/inflation exposure: prefer upstream with low-cost reserves and royalty trusts. - For commodity-business cash return: prefer companies with structural cost advantages that compound (railroads, certain industrial gases). - For high-quality cyclicals: wait for the actual trough. **What would change my mind:** 1. Stock falls below $160 with no fundamental change to MPLX value -> initiate cyclical position. 2. Capacity retirements accelerate to 2M bpd over next 3 years and demand stays flat -> structural margin re-rating. 3. Management announces large opportunistic buyback at trough multiples (a la Henry Singleton at Teledyne) -> capital allocation grade goes from B+ to A and changes the equation. **Bottom line:** MPC is a B+ business with B+ management at a fair price. A Buffett-Munger portfolio buys A businesses at fair prices, not B+ businesses at fair prices. Pass.