Correction

The prior version of this memo claimed AMR equity implied PLV of $180-195/t vs commodity consensus of $220-235 — a $30-50 gap. That claim doesn't reproduce. At current PLV-HVA capture ratios, AMR is roughly fairly valued at commodity consensus. The thesis needed rebuilding.

What follows is the honest version.

The Mispricing (Corrected)

AMR doesn't sell PLV coal. It sells predominantly High Vol-A, which currently trades at a $90/t discount to PLV — versus a historical average of $10-20.

This spread is the single most important variable in the stock. Everything else is noise around it.

AMR capture ratio history (Non-GAAP realization / PLV, from investor presentation and 8-Ks):

Each $10 of spread width costs ≈1.5% capture. At historical spread ($15), capture is ≈64%. At current spread ($90), capture is ≈55%. That 9-point difference is worth $300M+ in EBITDA on the same PLV.

Equity-implied PLV derivation (showing the work this time):

AMR equity:          $174.80
Shares:              12.8M
Market cap:          $2.237B
Net cash:            $403M (366 cash + 50 STI - 13 debt)
Enterprise value:    $1.834B

EBITDA = Tons × (Realization - Cost)
       = 15M × (Capture% × PLV - $98)

At different multiples and current 55% capture:

Multiple  Implied EBITDA  Implied Realization  Implied PLV
  3.0x       $611M            $138.77            $252
  3.5x       $524M            $132.93            $245
  4.0x       $459M            $128.57            $237
  5.0x       $367M            $122.45            $225

At current capture (55%), equity-implied PLV = $225-252.
Commodity consensus = $220-235.
Gap = ≈$0.

There is no PLV gap at current capture ratios. The prior memo was wrong.

But watch what happens when spread normalizes:

At 64% capture (historical spread ≈$15) and PLV $220:

Realization:  0.64 × $220 = $140.80
EBITDA:       15M × ($140.80 - $98) = $642M
At 4x:        EV $2.57B → equity $2.97B → $232/share (+33%)

At 58% capture (partial normalization, spread ≈$50):

Realization:  0.58 × $220 = $127.60
EBITDA:       15M × ($127.60 - $98) = $444M
At 4x:        EV $1.78B → equity $2.18B → $170/share (-3%)

The stock's value is a step function of the spread:

• Spread stays at $90 → stock is fairly valued to slightly overvalued at $175
• Spread narrows to $50 → stock is roughly flat
• Spread normalizes to $15-20 → stock is worth $230+

This is a doorway state. Two completely different companies depending on one variable.

Why the Spread Blew Out, and Whether It Normalizes

The PLV-HVA spread was $10-20 for years (2017-mid 2024, per CNR Q4 transcript). It blew to $90 because:

1. Premium supply cratered. Grosvenor (5M mt/yr premium low-vol) offline since June 2024 methane explosion. Peabody acquisition deal collapsed Aug 2025. Won't restart until late 2027.
2. Blue Creek premium supply entering slowly. HCC's new mine produces premium low-vol but only reached full run rate late 2025.
3. Steel mills shifted to premium grades for efficiency, bidding up low-vol relative to HVA.

For spread to narrow, you need:

• More premium low-vol supply → Blue Creek at 4.5M tons in 2026, Grosvenor eventually in 2027-28
• Or less HVA demand → unlikely near-term

CNR's Deck Slone (Q4 2025 transcript): spread expected to "normalize in not-too-distant future." Blue Creek is the mechanism — as premium supply increases through 2026-27, the relative scarcity premium should compress.

My estimate: Spread narrows from $90 to ≈$40-50 over 12 months as Blue Creek ramps. Full normalization to $15-20 requires Grosvenor return (2028). Partial normalization gets AMR to roughly flat; full normalization is the +30% trade.

This is the bet. Not PLV level. The spread.

What AMR Actually Is

19 mines, 8 prep plants, 15M met tons/year. Second-largest US met coal producer behind CONSOL. 65% ownership of DTA (Dominion Terminal Associates — export terminal in Newport News, VA). Product mix: 37% High Vol-A, 31% High Vol-B, 19% Mid Vol, 13% Low Vol.

Balance sheet: $524M liquidity ($366M cash, $50M short-term investments, $184M ABL), $13M debt. Net cash. $400M buyback authorization remaining on 12.8M shares outstanding.

2026 guidance: 14.4-15.4M met tons, cost $95-101/ton (includes Section 45X credit), capex $148-168M.

Pricing exposure for 2026:

• 27% committed/priced at $128.16/ton (domestic met at $136.75)
• 38% committed, unpriced
• 35% uncommitted, unpriced

73% reprices to market. But "market" for AMR means HVA, not PLV. At current $90 spread, PLV $240 gives AMR ≈$132 blended realization. Not $140+.

Why AMR Over HCC

The review asked the right question. HCC outperformed AMR by 30% annualized over the past 90 days. The market is choosing HCC as the preferred met coal vehicle. Why?

HCC's advantages:

• Blue Creek produces premium low-vol → sells at/near PLV, no HVA discount
• Capex cliff: $155-215M in 2026 vs $402M in 2025 → massive FCF inflection
• 4 Buy ratings vs AMR's 0
• Better 1Y momentum (+73.5% vs +11.7%)

The discount is justified at current HVA spread. HCC captures ≈75-80% of PLV because it sells premium product. AMR captures 55%. The market is correctly pricing this difference.

The AMR-specific thesis is spread normalization optionality.

If spread stays at $90: HCC is the better stock. No question. HCC is the cleaner pure met coal vehicle.

If spread normalizes: AMR has asymmetric upside that HCC doesn't. HCC already sells at PLV — spread normalization adds zero to HCC. AMR's capture jumps from 55% to 64%, which is +$300M EBITDA on the same PLV. HCC doesn't have this embedded option.

The question isn't "AMR vs HCC for met coal exposure." It's "do you have a view on spread normalization?" If yes, AMR is the only vehicle that expresses it. If no, HCC is the cleaner play.

The insider buying supports this framing. Courtis (Goldman commodity veteran, 21% of net worth in AMR) and Gorzynski ($7.3M at $188) are not making a generic met coal bet — they're making an AMR-specific bet. They see something in the spread or company-specific recovery that justifies AMR over the market's preferred HCC.

Supply: Net New Is +4-8M, Not 14-16M

The bear case cites "14-16M tons of new supply" hitting 2026. Roughly right gross, misleading net.

Mine-level verification (operator transcripts and 8-Ks, Jan-Feb 2026):

Legitimate subtractions:

• Leer South is restoration, not new supply. Returning to 2024 baseline. Net new = zero. (-3.6M)
• Moranbah North is partial year only. 2-3M of 5.3M nameplate. Remote mining phase. (-2-3M vs nameplate)
• Queensland flooding removed 2-5M tons temporarily. Cyclone Koji (Jan 2026) triggered force majeures at AMCI, Stanmore, Glencore, Coronado.
• BTU Wambo Underground closed Q3 2025. (-1-2M)

Note: Grosvenor (5M mt/yr) is NOT subtracted from new supply because it was already offline in the 2025 baseline. Bears aren't counting Grosvenor restart in their 14-16M; subtracting it double-counts.

Net genuinely new seaborne supply: +4-8M tons. Still well below the structural gap that multiple analysts estimate at 10-20M tons. Supply is tightening, not loosening. But the headline is less dramatic than the prior memo claimed.

The Pricing Mechanism (Why Earnings Lag PLV)

AMR's quarterly pricing creates a specific, predictable lag between commodity moves and earnings recognition. This is not edge — every coal analyst models it. But it explains why the market values AMR on trailing trough earnings.

Q4 2025 was the trough. PLV averaged ≈$210 in Q4 but AMR realized $115.31/ton because:

• Export-Other (50% of tons): Spot-indexed at $106.13 — these priced when PLV was weaker
• Domestic (22%): Fixed at $148.93
• Australian-indexed (28%): Lagged repricing at $114.96 — these reset quarterly with ≈1Q delay

CEO Eidson on the Jan 30 8-K: "benefit was pushed into Q1 of 2026."

With PLV at $240+ through January-February, the lag works in reverse. Q1 realization should exceed $125/ton (our prediction: 85% probability). At $130 realization and $98 cost, Q1 EBITDA could be $120-150M vs Q4's $28.5M.

This isn't edge — it's arithmetic. But the Q1 print in May is the catalyst that forces the market to update its EBITDA estimate from trailing trough to forward run-rate.

Insider Buying: Courtis

$40.7M in Code P (open market) purchases in 2025, against $2.2M in sales. Buy/sell ratio 18.5:1.

The money is Kenneth Courtis, a director. Former Goldman Sachs MD and VP Goldman Asia, former Deutsche Bank MD. 30+ year commodity sector veteran. Net worth approximately $158M. He owns 816,537 AMR shares — roughly $142M at current price.

His 2025 purchases: 205,000 shares for $33.5M across six transactions, at prices ranging from $148 to $188. He put 21% of his net worth into a single stock. This is not a governance checkbox. This is a commodity veteran who has seen multiple coal cycles making a concentrated bet at what he believes is the bottom.

Michael Gorzynski (10% owner): 38,576 shares at $188.50 ($7.3M) in December.

Stock is now $174 — between Courtis's weighted average cost (≈$160) and Gorzynski's entry ($188). Insiders are underwater on the recent tranche and still haven't sold a share.

What Kills This

Engaged, not noted.

Spread persists at $90. If premium low-vol supply stays scarce and steel mills continue preferring premium grades, the HVA discount doesn't normalize. AMR capture stays at 55%. At PLV $220 and 55% capture, EBITDA is ≈$345M and stock is worth $135-165 on 3-4x. Current price is overvalued by 5-25%. This is the most likely bear outcome — not a PLV crash, but spread persistence.

Why it might not happen: Blue Creek adds 4.5M tons of premium low-vol in 2026. Grosvenor returns 2027-28 with 5M mt. More premium supply should narrow the spread. CNR expects normalization "in not-too-distant future."

PLV crashes to $180. Global recession plus demand destruction. At $180 PLV and 55% capture, realization is $99 — barely above cost. EBITDA near zero. Stock worth $30-50 on net cash basis. Probability: 15%.

Why it might not happen: Byproduct supply economics. 70% of global met coal is mined as byproduct of thermal coal. Supply can't respond to price increases (no new greenfield met coal investment). Cost curve floor is rising. PLV $180 would cause mine closures that self-correct the price.

Demand destruction — India and China. Can't fill these gaps with primary data. India import pace (biggest demand driver) requires Kpler/BigMint subscription. China domestic production requires NBS monthly data (lagged). The entire demand side of this commodity thesis is unverifiable with available sources.

Liquidity trap. 300K ADV, $2.3B market cap, 15.6% short interest. Exit takes weeks, not days.

Factor Decomposition

Model: AMR vs SPY + XLB + MTUM (90-day daily)
  Market beta:    0.76
  Sector beta:    minimal
  Idio variance:  90% (vs SPY+XLB)
  Alpha:          +8.2% annualized

Model: AMR vs SPY + HCC (closest peer)
  HCC beta:       0.65
  Idio variance:  49%
  Alpha:          -30.4% annualized (massive underperformance vs HCC)

The first model is misleading — 90% idio just means met coal is its own asset class. The second is honest: AMR underperformed HCC by 30% annualized. That underperformance IS the HVA spread blowout. Same commodity, different product quality, different price realization.

Edge audit:

• Met coal factor (48.3%): Supply-side evidence is strong (mine-level verified). Demand is unverifiable. Edge: PARTIAL (supply only).
• Idio component (48.8%): This IS the spread normalization bet. AMR-specific product quality, capture ratio trajectory, insider buying signal. Edge: YES, but conditional on spread view.
• Market factor (2.9%): No edge, irrelevant.

Effective edge: ≈50-55% of variance. Lower than prior memo claimed because demand side is a gap, not an edge.

Alpha Calculation

Scenario EV:      $200 (revised down — see scenarios)
Current:          $174.57
Horizon:          6 months (0.5 years)
Risk-free:        5%

Raw annualized:   ($200 / $174.57)^(1/0.5) - 1 - 0.05
                = 1.1458^2 - 1 - 0.05
                = 1.3129 - 1 - 0.05
                = 26.3%

Conviction:       42% (demand gaps unfilled, spread thesis unproven)
Edge%:            55% (supply verified, demand not, spread conditional)
Alpha:            26.3% × 0.42 × 0.55 = 6.1% annualized

6.1% alpha. Small. Not a high-conviction swing — a conditional, thesis-dependent position.

Scenario

EV: $192 (+10%)

The asymmetry is thinner than the prior memo suggested. Base case is +12%, not +26%. The bull requires both PLV and spread to cooperate. Bear-1 (spread persists) is +20% probability and results in -17%. This is not a fat pitch.

Catalysts and Timeline

The problem: spread normalization is a 12-24 month process, not a 6-month catalyst. Partial normalization by Q4 2026 is the realistic upside. Full normalization is a 2027-28 story.

Sizing Constraints

At 6.1% alpha, this warrants small sizing — not a high-conviction swing.

Key constraints:

1. Spread thesis is conditional. If HVA spread doesn't narrow, stock is overvalued here. This is a bet, not a certainty.
2. Demand gaps unfilled. India and China data behind paywalls/lagged. Can't verify the commodity bull case.
3. 11-week vulnerability. No hard catalyst until May Q1 print.
4. Liquidity. 300K ADV. Position exit takes weeks, not days.
5. HCC alternative. For generic met coal exposure without spread dependency, HCC is the cleaner vehicle.

The Feb 27 earnings call is the first information event — management commentary on Q1 pricing trajectory and HVA spread outlook matters more than Q4 numbers (already in the Jan 30 8-K). The Q1 print in May is the real test.

Kill trigger: Q1 2026 blended realization below $120/ton with PLV above $220 = spread not normalizing = thesis invalidated.

What I Don't Know

1. PLV-HVA spread trajectory. This IS the thesis and I don't have a primary source timeline for normalization beyond CNR's vague "not-too-distant future." Blue Creek ramp is the mechanism but it's HCC's mine, not AMR's — AMR can't control this variable.
2. India import pace. Biggest demand driver. Paywalled data.
3. China domestic coal production. NBS data lagged.
4. Buyback acceleration threshold. What EBITDA level triggers $400M deployment? Feb 27 call may clarify.
5. Whether Courtis's thesis is spread normalization. We're inferring this from his buying pattern. He hasn't publicly stated his thesis. A Goldman commodity veteran might see something else entirely.

Bottom Line

The prior memo was wrong about the magnitude of the mispricing. There is no $30-50 PLV gap — at current HVA capture ratios, AMR is roughly fairly valued to slightly overvalued at $175.

The actual thesis is narrower and more conditional: AMR is a leveraged bet on PLV-HVA spread normalization. At current $90 spread, the stock is fair. At normalized $15-20 spread, it's worth $230+. The spread blew out because premium supply cratered (Grosvenor offline); it should normalize as Blue Creek ramps and eventually Grosvenor restarts. But the timeline is 12-24 months, not 6.

The strongest signal in the whole analysis isn't the commodity or the supply math — it's Kenneth Courtis putting 21% of his $158M net worth into the stock. A Goldman Sachs commodity veteran who has seen every coal cycle making a concentrated bet at these levels. He likely sees the spread normalization path before the market does.

At 6.1% alpha, this is a small, conditional position — not heroic. The Q1 capture ratio is the validation checkpoint. If capture doesn't improve with PLV above $220, spread isn't normalizing and the thesis is dead. For pure met coal without spread dependency, HCC is the cleaner vehicle.