LMT

We rate Lockheed Martin a Hold: the price is $584 and our fair value is $595. Its third-quarter results beat every forecast we made in our 2026-10-15 report, and that lifts our value $41, from $554 to $595. The price rose more, from $532.64 at the last report to $584 this morning. A buyer today earns 8.8% a year over five years, short of the 12% hurdle; a buyer below $507 would earn it. The Hold stands; the $554 does not.

What moved the value: sales of $20.7B were 5.7% above our $19.6B, segment profit and free cash flow landed inside our ranges, and the munitions backlog tripled in nine months to $142.5B. Two changes carry most of the move. Missiles and Fire Control (MFC), the missile and munitions segment, grows faster in our central case (+$20). And the outside-view exit multiple is 17.5x instead of the method's 16.5x, because the history it comes from counts a pension credit that our profit leaves out (+$16). Facts and guidance add $10, and Aeronautics (aircraft, including the F-35) and tax $4. Outside reviews then cut the value twice: later years' corporate costs had been carried below the October guide (−$11), and the share of the pension credit that was cash had been measured over the wrong years (−$8).

Terms used below.

• Bear, central and bull cases. The central case is our best estimate. The bear is a severe but plausible bad outcome, and the bull is the matching good outcome. Only the central case decides the call.
• Call rule. Buy when the price is at or below the purchase price ($507, which earns the 12% hurdle). Avoid when fair value is under 85% of the price ($497). Otherwise Hold.
• Fade. One way to value 2031: let the company's returns on new investment drift toward its cost of equity over 5, 10 or 20 years.
• Outside view. The other way: apply a multiple that comparable companies trade at, here enterprise value divided by operating profit (EV/EBIT), to our 2031 profit. The method starts from a benchmark multiple of 16.5x. The exit price averages the two ways.
• Program charges. Losses booked when a fixed-price contract is expected to cost more than it pays (reach-forward losses).
• The release and the call. The Q3 earnings press release and the earnings call, both on 2026-10-22.

1. Two of five forecasts landed in range, and every miss was too low

Graded on main in ../2026-10-15-lmt/grades.json from the Q3 release (8-K exhibit 99.1, filed 2026-10-22).

Two of five were in range, and every miss was on the low side. The 2026-10-15 model's full year was right: $80.75B of 2026 sales from the July segment midpoints, against $81.0B now. The miss was in the split. That model put too little of the second half into Q3, and did not carry MFC's acceleration into the quarter. One quarter is not a record, but the guidance record in output.txt points the same way. Since July 2025, every LMT full-year guide has ended at or above its midpoint; in 2024 and early 2025, segment profit ended 14-19% below it.

2. Nine steps take the value from $554 to $595

model.py first runs the 2026-10-15 inputs through the shared method (method.py, judged-exit-2026-10-18), then applies each change on top of the last (output.txt, "Bridge"). The 2026-10-15 report's own fade gave $554; the shared method on the same inputs gives $555.8, so moving onto the shared method changes little.

$555.8 + 0.9 + 8.9 + 20.1 + 2.2 + 2.1 + 23.9 − 11.3 − 7.6 = $595.0. The steps interact; the order is fixed in STEPS in model.py and nothing is left unallocated. The old bear ($331) and bull ($707) move to $360 and $681 from the move onto the shared method alone, because the old fade treated the cases' mature returns differently from the shared method.

By kind:

• Release and guidance (steps 1, 2 and 7): −$1.5.
• Tax rate, a law change read from the release: +$2.1.
• Our judgment on MFC and Aeronautics (steps 3-4): +$22.3.
• Corrected basis of the exit multiple (steps 6 and 8): +$16.3.

Why the MFC path rises (step 3).

• MFC grew 24% in Q3, and its 2026 guide is $17.25-17.45B, +20% (call).
• Its backlog went from $46.65B at the 2025 year end to $142.5B at 2026-09-27 (release), 8.2 years of 2026 sales.
• The jump follows the seven-year PAC-3 contract, "up to $58.6 billion" (call), of which $52.5B is in backlog (release), and a seven-year THAAD contract (call).
• Management says "mid-teens sales CAGRs over the next five years" (CFO, call; management claim).
• Our central averages 12.8% a year to $31.6B in 2031, below that claim. Its cumulative 2027-2031 MFC sales ($130B) are less than today's MFC backlog.
• The bull runs 20% falling to 12%; the bear, 11% falling to 5%.
• Margins stay at 13.7-14.0%, because the profit terms of the munitions framework agreements are not set ("left to work", call).

Why steps 6 and 8 correct the basis of the exit multiple, and why 17.5x.

• The outside view applies the method's benchmark multiple to our year-five EBIT. The benchmark is the mean of LMT's own and its peers' ten-year median EV/EBIT: 16.5x.
• Our EBIT leaves out the FAS/CAS pension operating adjustment (FAS is the company's financial accounting; CAS is the government's cost accounting standards). It is the CAS pension cost recovered in prices, net of FAS service cost: a credit of $1.25-2.05B a year inside LMT's GAAP operating income in 2016-2025, and $1.685B in the 2026 outlook (release).
• NOC, RTX, LHX and HII carry the same credit; GD does not.
• The benchmark's history is market data enterprise value over GAAP operating income, credit included. It is therefore measured on a larger profit than the one we multiply.
• With the credit taken out of the history, the benchmark is 19.5x (LMT 18.9x, peers 20.2x; anchor_basis.py, anchor_basis.txt, from 10-K tables).

Neither end is right, because the credit is not all paper. Over the benchmark's own window, 2016-2025, LMT recovered $19.98B of CAS pension cost and paid $8.85B into the trust (10-Ks; none in 2016-2017, $5.0B in 2018). So 65% of the $17.06B operating adjustment came back as cash the market could count. Keeping 65% of every company's credit in operating income gives a benchmark of 17.52x (LMT 16.16x, peers 18.87x; anchor_basis.txt). We use 17.5x. At 16.5x the central is $579; at 19.5x it is $626.

Our first draft took 18.0x, "halfway", supported by a 44% cash share over 2018-2025. An outside audit of the model showed that window leaves out two years of the benchmark's ten (2016-2017, credit recovered and nothing paid in) while keeping the 2018 prefunding, and that 44% used CAS cost rather than the adjustment as its denominator. Step 8 replaces the judgment with the matched figure. It is still the step most open to challenge: read forward, our own model has contributions equal to CAS cost, which would argue for 19.5x.

Step 7, found by an outside report check. The model's 2026 uses the October guide's −$540M for other unallocated items, but it carried 0.5% of sales into 2027-2031. That figure came from the 2026-10-15 report, which drew it from the 2023 to July-2026 range of 0.28-0.61%. The October guide is 0.67%, above that range, so 2027 fell by about $0.1B for no reason. We now carry the guide's share forward: −$11.3.

3. What the release and call changed, and what is still unverified

• 2026 guide raised for the second quarter running (release):

  • sales $80-82B (was $79.75-81.75B);
  • segment profit $8.575-8.675B (was $8.50-8.70B);
  • FCF (free cash flow) $7.8-8.0B (was $7.0-7.2B);
  • EPS raised.

  Our 2026 uses the segment midpoints from the call's prepared remarks.

• Aeronautics 2026 profit cut to $2.92-2.95B from $3.00-3.08B, because of the lot award timing: "we shift that assumption to 2027" (call).

  • The F-35 Lot 20 advance contract's costs now exceed its authorization and funding. Without more, "results of operations, cash flows, and financial condition will be negatively impacted" (release).
  • F-35 deliveries were 94 in nine months, against 143 a year earlier.
  • Aeronautics backlog fell to $57.3B from $59.4B.

• F-35 multi-year: management claim, not in a filing. The CEO cited a Department of War ten-year agreement announced that morning at 156 aircraft a year, "850 U.S. aircraft over a 10-year span" (call). He gave no terms. We already held 156 a year in every case, so this changes no input; it narrows the bear's room for a production cut.

• Munitions framework terms (call): year-zero price on CAS, an inflation escalator, retained efficiency gains, cash neutrality during the ramp, and a recovery mechanism if volumes are cut. Profit terms are not set. We did not use them.

• Pension: management expects "a pension contribution in excess of $1 billion to be required next year", with a possible 2026 prefunding (CFO, call). Ours is $1.6B a year from 2027, equal to CAS cost, so the cash line nets to zero.

• Cash taxes: part of the FCF raise is lower cash tax from IRS guidance. Management calls it "more than a single-year benefit" but smaller in 2027 (call). We carry no cash-tax benefit after 2026.

• Ultra Maritime ($3.45B, closing in Q4 2026 for cash, call) stays out of both sides of the bridge.

• 10-Q pending. These balances stay at their 2026-06-28 values (Q2 10-Q):

  • accrued program losses: $427M in Aeronautics, $1.1B in MFC, $365M in CMHP, all classified;
  • venture investments;
  • environmental balances.

  A new loss accrual or a Lot 20 provision in the Q3 10-Q would reopen the bridge and the charge allowance.

4. The central forecast: sales of $107.8B in 2031 and a $595 value

The central case (output.txt):

• Sales: $86.8B in 2027 (+7.1%) and $107.8B in 2031, 5.9% a year from 2026.
• Segment profit: $9.37B in 2027 before a $0.52B charge allowance (0.6% of sales), and $12.3B in 2031.
• Other unallocated: 0.67% of sales.
• Exit multiple (outside view): 17.5x EV/EBIT.
• 2031 profit: EBIT excluding FAS/CAS $10.7B; NOPAT (operating profit after tax) $9.0B at a 16.0% tax rate.
• Capex: $2.4-2.7B a year.
• Shareholders: dividends rise 4.6% a year, and no case has buybacks after 2026. Net debt falls from $14.3B at the 2026 year end to net cash of $1.4B in 2031.

Management frames 2027 as "high single-digit" sales growth (call); our 7.1% is at its low end.

Sensitivities of the central fair value (output.txt):

5. The price is just below our value, and Alpha ranks LMT 73 of 709

Last trade at 14:01 UTC was $584.15, up 5.9% on the previous close of $551.51 (2026-10-21). The day carried both the Q3 release and the Department of War's F-35 announcement (call). We do not split the move between them. Our $595 is 1.9% above $584.15 and 7.9% above $551.51.

The rule gives Hold at either price. The purchase price ($507) is below both, and fair value is above 85% of both. At $584 the central five-year return is 8.8% a year (bear −0.8%, bull 14.2%); at $551.51 it would be 10.1%.

The Street's mean target is $620.23 (market data, updated 2026-10-03, before the release). Discounted a year at our cost of equity it is $572, and our $595 is 4% above that. The 2026-10-15 report noted target cuts running down into the print. We do not use targets or estimates as evidence for an input.

Odds. alpha research odds (odds.json, 4,000 draws over three weighted worlds) puts 36% on fair value reaching $584.15:

• 26% in the base world;
• under 1% in the bear-leaning world;
• 95% in the bull-leaning world.

The simulated median is $547, 8% under the central $595, because our ranges are skewed. Charges can run to 2.5% of sales but not below 0.2%, and MFC growth has more room down (−6 points) than up (+4). Read the central as the most likely path and $547 as the middle of our spread. The Hold holds at either.

Alpha's view. Alpha is Capfolio's stock-ranking model. alpha rank LMT --raw --json --allow-stale-model-cache was run read-only on 2026-10-22 (data/alpha-rank.json; as of 2026-10-21; cache status "partial", freshness "rerate_required"). LMT ranks 73 of 709, with a score of 89.8 and tier Strong (90th on 2026-10-15). Grades: quality B, growth B, valuation B, expectations C, momentum C (rank 466). Our Hold does not conflict with it: Alpha ranks stocks against each other over 126 trading days, while we value the company against its own five-year cash.

Where we differ from management and the method. Our central is a little above the price, by less than the range the benchmark multiple's two bases span ($579-626). We are more cautious than management on MFC (12.8% a year against "mid-teens") and on Aeronautics, and harsher on cash taxes. We are more generous than the method's default on the exit multiple. The 2027 guide in January settles it: MFC growth and segment profit against ours.

6. Bear case: why the other side could be right

The bear ($380) runs two bad years for five more:

• charges at 1.8% of sales a year (2024 and 2025 were 2.75% and 2.34%);
• segment margins 0.4-0.8 point under the central;
• MFC slowing from 11% to 5% as framework volumes slip.

Three things keep it live:

• Fixed-price risk is still in the book. Classified programs in Aeronautics and MFC took most of the 2024-2025 losses, and Lot 20's costs now run ahead of its funding.
• The munitions terms are not written. A year-zero price on CAS is cost recovery, not margin. A framework that is "cash neutral" during the ramp can still earn less than 14% on it.
• The exit multiple. The method's own 16.5x gives $579, and our 17.5x rests on how much of the pension credit the market counted as cash.

A bull would answer that we are too low:

• management's mid-teens MFC growth is worth +$49 a share;
• the multi-year F-35 removes the volume question;
• the 2024-2025 charges came from a specific set of programs, and the nine months show no new ones.

The bull ($757) runs that world.

The funding question (raised by an outside reviewer arguing the opposite call). Press reports say most of the FY27 munitions money sits outside the base budget:

• Most FY27 munitions procurement sits in a reconciliation request (a separate budget bill), not the base budget. BNP Paribas' Matt Akers is cited putting 90% of PAC-3 and 84% of THAAD funding there (Washington Sun, 2026-07-13; secondhand).
• The House passed $60B for defense against a $350B request on 2026-07-22 (Defense Daily; press).
• The Pentagon comptroller said priorities would move into the base budget by "cutting other investments or deferring them to 2028" if reconciliation fails (Breaking Defense, 2026-09-10; press).
• FY27 "opened under a continuing resolution" (call).
• LMT's backlog includes unfunded orders (10-K filed 2026-01-29).

So the $142.5B MFC backlog does not prove the money is appropriated. Priced in our model:

• an MFC ramp funded a year late and smaller (9, 11, 12, 11, 9%): −$27;
• MFC growth 3 points a year lower: −$41;
• revenue brought down to the Street's 2030 figure through MFC alone: −$52, which leaves the central at $544.

Each of these is still a Hold. We keep the central path: the outcome is a vote not yet taken, and a slower appropriation delays volume more than it removes it (a one-year slip with the same 2031 volume costs almost nothing). It is the first thing the January guide and the reconciliation vote will test.

What flips the call. No single input reaches Avoid (fair value under $497) except program charges. At about 2.1% of sales every year from 2027 to 2031, the central falls to $497. That is near 2024's 2.75% and 2025's 2.34%. The other single inputs need levels with no support:

• MFC growing 4.6% a year;
• every segment margin 1.5 points lower.

Several adverse changes together can reach Avoid without that. MFC on the Street's path, MFC margins drifting to 12.6%, charges at 1.0% and the method's 16.5x together give $484.

7. Next results preview: Q4 2026 and the first 2027 guide

The release is due in late January or early February 2027. alpha research due lists 2027-02-04; the last two Q4 reports came on 2025-01-28 and 2026-01-29; the company has not confirmed a date. The forecasts are in record.json.

Each point a year of MFC growth in 2027-2031 is worth about $14-16 a share (3 points: −$41 / +$49). A 2027 MFC guide near 18% would point toward the bull's path.

8. Catalysts and falsifiers

Catalysts:

• the Q3 10-Q, expected within days;
• Ultra Maritime's close and any 2026 pension prefunding, by 2026-12-31;
• Q4 2026 results and the 2027 guide, late January or early February 2027.

Falsifiers, each with its level:

• A 2027 MFC sales guide under 12% growth, or framework profit terms set below CAS-level margins: toward the bear (3 points a year ≈ −$41).
• New reach-forward losses above $1.0B in 2027: toward the bear. Charges at 1.8% of sales a year are −$80.
• An initial 2027 segment profit guide midpoint under $8.8B: the central is too high.
• MFC growing 18% or more in 2027 at a 14% margin, with charges at or under 0.3% of sales: toward the bull ($757).
• Required pension contributions above $2.0B a year from 2027: the central is too high.
• The Q3 10-Q showing new accrued losses or a Lot 20 loss provision: we revisit the bridge and the charge allowance.
• Charges at 2% of sales or more for two years running from 2027: the call turns to Avoid (2.1% a year sustained puts the central at $497).
• FY27 defense reconciliation failing, with PAC-3 and THAAD volumes deferred to FY28 or later and not restored in the base budget: MFC toward a 9-12% path, about −$27 to −$41.

9. Review: four checklist flags and the outside findings, each answered

Four outside reviewers read the draft: a check of the report's math and sourcing (Claude Opus 5.5), an audit of the model (Claude Opus 5.5), a reviewer arguing the strongest case for the opposite call (Claude Opus 5.5), and OpenAI's Codex. Our review checklist also raised four flags, each answered here.

Checklist flags:

• Size of the move since the last report (high). The net move since the last report is +$41; the largest step is the exit-multiple basis (+$16.3 net, steps 6 and 8). The MFC step leans partly on a management claim, but our path sits below it and under the backlog.
• Fair value close to the price (medium). Fair value is 1.9% above the price. A Buy needs the price at or under $507; an Avoid needs fair value under 85% of it ($497).
• Central case against backlog, guidance or the Street (medium). Our 2027 sales are above the Street's $85.9B, on the company's frame.
• A lenient input chosen over a harsher sourced one (medium). We use 17.5x where the method's 16.5x is harsher; the matched figure replaces the draft's judged 18.0x (step 8).

The other checklist items did not fire (the one on shared sector drivers has no sector note to compare with). Five of them (checklist items 14, 16, 17, 18 and 19) were checked against these figures:

• central fair value is 8.1% from the simulated median;
• the chance that fair value reaches the price is 36%;
• fair value is 1.04x the Street target discounted one year;
• our 10th to 90th simulated range, $436-655, against a target spread of $486-756;
• the judged exit multiple is 1.06x the benchmark multiple.

10. Sources

• Lockheed Martin Q3 2026 release, 8-K exhibit 99.1, filed 2026-10-22, accession 0001628280-26-066780 (primary); data/ex991q32026.htm.
• Q3 2026 earnings call, 2026-10-22, Quartr transcript via Bigdata.com (secondhand); data/q3-2026-call-transcript.txt.
• LMT 10-Ks filed 2021-01-28, 2024-01-23 and 2026-01-29; the Q2 2026 10-Q (primary).
• 10-K FAS/CAS tables for LMT, NOC, RTX, LHX, HII and GD (primary); anchor_basis.py, data/anchor-basis/fas-cas.json.
• Exit-multiple benchmark history: market data fiscal-year-end EV and GAAP operating income, fetched 2026-10-19; data/exit-anchor.json.
• Market inputs: risk-free 5.29% and ERP 4.20% as of 2026-10-01 (Damodaran), and weekly betas; data/damodaran-erp.txt, data/beta-weekly-log-returns.json.
• Price and Street: market quote of 2026-10-22 14:01:36 UTC and analyst ratings (updated 2026-10-03); data/market-2026-10-22.json, market.py, market.txt.
• Alpha rank, 2026-10-22: data/alpha-rank.json.

The brief that started this revision quoted a 2026-10-22 close of $599.97 before the session had closed. Market data showed $551.51 for 2026-10-21 and $584.15 intraday, so that figure was not used.