The annual filing says Limbach grew organically 3.6%. That's technically true and completely misleading. The earnings call — one day later — says ODR organic growth was 17%, accelerating to 23.9% in Q4. The 3.6% blends a business they're intentionally killing (GCR, down 23%) with one they're building. Anyone who screened the 10-K and moved on got the wrong number.

Meanwhile, the CEO and CFO dropped $1.3 million of their own money into the stock at $82 on December 31 — the same week their peers at Comfort Systems were liquidating $25 million at all-time highs.

Someone is wrong. The insider buying says it's the market.

The Business

Limbach is a $1B mechanical/electrical/plumbing contractor for mission-critical buildings — hospitals, data centers, industrial facilities. The company has spent five years pivoting from low-margin general contracting (GCR) to high-margin owner-direct relationships (ODR): direct maintenance, retrofit, and service contracts with building owners.

The pivot is structurally complete. ODR hit 75% of revenue in FY2025, achieving management's 70-80% target. GCR is being wound down at improving margins (17% → 24.5%) — disciplined exit, not fire sale.

What the 10-K Shows vs What's Actually Happening

The 10-K consolidates numbers that hide the real story. The 450bps margin compression (31.2% → 26.7% ODR) is 100% acquisition drag from Pioneer Power, acquired July 2025. The legacy business — the thing management actually built — is running at 28.2% gross margin, stable year-over-year.

The Insider Signal

This is the strongest insider buying signal in the building services sector this cycle:

• LMB: CEO McCann $813K + CFO Brooks $530K, open market, December 31 at $82. Three directors added ≈$78K each. Total $1.6M personal capital at 47% off peak.
• FIX (Comfort Systems, $40B): 30 insider sells totaling $25.4M in Q1 2026. CFO dumped $12.9M. Stock at all-time highs, +296% 1Y.
• EME (EMCOR, $33B): Zero insider purchases. CAO sold open market.

FIX insiders are cashing out at the top. LMB insiders are buying at the bottom. Both can't be right about building services — but both CAN be right about their own companies.

The Pioneer Power Question

This is where the thesis lives or dies.

Pioneer Power's gross margin is below 13.4% — worse than where Jake Marshall was when LMB acquired it in December 2021. Jake Marshall went from 13.4% to 28% over four years using LMB's integration playbook. Management says Phase 1 (systems integration) is largely complete and Phase 2 (margin improvement) is now the focus, with "exit margins higher than current levels" by end of 2026 and full alignment in 2-3 years.

The playbook is proven. The question is whether Pioneer Power is another Jake Marshall or something worse.

Cross-ticker evidence is NOT encouraging here. FIX expanded gross margins 310bps while acquiring six companies in 2025. EME expanded 30bps with its largest-ever deal. LMB compressed 450bps. LMB is the outlier — and the explanation that "we're smaller, less experienced at M&A" isn't exactly reassuring.

Management's counter: Pioneer was bought at 5-6x EBITDA specifically BECAUSE margins were low. The value creation is the thesis, not the bug. Jake Marshall proves the model works. Fair enough — but it takes 2-3 years, not 2-3 quarters.

The Datacenter Red Herring

Less than 5% of revenue. Building a dedicated national team. Fourth project with a hyperscaler ($10M, Columbus). Two "very strong emerging relationships" with hyperscale owners. CEO is "bullish" but "going to see how it goes."

Compare to FIX: 45% of revenue from datacenters, $12B backlog (+93% YoY). EME: $4.5B in datacenter RPOs (+60% YoY).

If you want datacenter MEPC exposure, LMB is the wrong vehicle. The datacenter narrative is sector beta in a company costume. LMB's core is healthcare — and that's fine. Healthcare facilities need MEPC work forever. It's just not the story some people are telling.

Factor Decomposition

Statistical: 83% idiosyncratic variance. Passes the 75% threshold. Returns are stock-driven, not market/sector. Backward alpha is zero — the stock roundtripped from $73 to $154 back to $82. No historical alpha to extract. Forward alpha is the whole game.

Economic factors, ranked by edge:

Edge-weighted: ≈53%. More than half of LMB's variance comes from factors where we have (or can develop) informational advantage. Five analysts, one research shop (Stifel). This is the edge zone.

Valuation

At $82 with guided $90-94M EBITDA ($92M midpoint) for 2026:

• EV/EBITDA: ≈11x (cheap for a services compounder)
• Forward P/E: 16.5x on ≈$4.50+ adjusted EPS
• FCF yield: ≈4.2% on guided 75% EBITDA conversion
• Analyst mean target: $118.60 (+44%). All from one shop (Stifel). Take it with salt.

The $50M buyback authorized December 2025 hasn't been deployed yet. At $82, that's 5.2% of shares outstanding. Optionality, not certainty.

Scenario Distribution

EV: $92.90 (+13.3% from $82)

What I'm Watching

Confirmation gate: Q1 2026 earnings (May 4). If ODR margin ticks toward 27%+, the thesis validates — insiders were right, margin recovery is tracking. If flat or declining, Pioneer Power is a different animal than Jake Marshall.

Kill trigger: Consolidated gross margin below 25% for two consecutive quarters.

The bear case I respect most: FIX and EME expand margins while acquiring. LMB compresses. The explanation is "we're earlier in integration" — but maybe the explanation is "we bought a lower-quality business." Pioneer Power's gross margin is BELOW 13.4%. Jake Marshall was AT 13.4%. Starting worse and claiming the same trajectory requires faith in the playbook that cross-ticker evidence doesn't fully support.

Evidence