The Setup

EIG is a California-heavy workers comp insurer that just got punched in the face. Net income collapsed 91% ($118.6M → $10.8M) after cumulative trauma claims in California "accelerated beyond historical levels" — their own words in the 10-K, not mine. Combined ratio blew out to 110.9%. The stock trades at 81% of adjusted book ($41 vs $50.95).

The bull case writes itself: CFO Pedraja bought stock three times with his own money — $200K in August, $204K in November, $79K in February. The February purchase came AFTER he saw Q4 results. Dual actuarial firms confirmed reserves are adequate, no further strengthening needed. The reserve range actually narrowed $106M year-over-year, with the adverse high end falling $58.6M. Management bought back $97M of stock in Q4 at a 20% discount to book. Net investment income provides a $120M+ floor ($5.75-5.90/share) from $1.4B of float — the hidden engine that keeps EIG alive even at 110% combined.

Everything converges: insider buying, reserve confirmation, buyback at discount, NII floor, AM Best "A" reaffirmed. Clean, coherent, attractive.

And I'm passing.

Why PASS

Three reasons, in order of importance.

1. It's a sector bet. Regressed EIG against SPY and KIE (insurance sector ETF) over two years: KIE beta = 0.95, explaining 42% of total variance. Idiosyncratic variance = 58.1% — well below the 75% target. Even in the post-crisis window (last 6 months), idio only reaches 72.5% with KIE beta still at 0.86. EIG moves with insurance. If you want insurance exposure, KIE at 0.35% expense ratio doesn't come with California cumulative trauma risk attached.

2. Forward alpha is noise. Honest calculation:

Target: $50.95 (adjusted BVPS)
Current: $41.68
Horizon: 18 months
Annualized total return: 14.4%
Subtract KIE expected: ≈6%
Subtract risk-free: 5%
Raw idio return: 3.4%
× Edge% (72.5% idio): 2.5%
× Conviction (50% — CT uncertainty): 1.2%

1.2% annualized idiosyncratic alpha. That's not an investment. That's a rounding error.

3. The key driver is outside our competence. The entire thesis hinges on California cumulative trauma claim frequency. Will it normalize? Accelerate further? Plateau? The 10-K says it "accelerated beyond historical levels" but management believes 5% rate increases will catch up. The gap between 5% rate increase and 8-point loss ratio deterioration is the central question — and answering it requires WCIRB actuarial data and California workers comp litigation expertise we don't have.

What's Real

The convergent signals aren't fake. They're just not enough.

CFO buying three times — including post-results — is the strongest signal. Insiders who see the reserve studies buying with their own money at 20% below book is genuine information. But note: the CEO isn't buying. One officer with conviction at a $900M company is evidence, not proof.

Reserve range narrowing ($350M spread → $244M) with the high end dropping $58.6M is legitimately good actuarial news. Management carrying reserves at the exact midpoint of the range ($1,419M vs $1,295-1,539M endpoints) is conservative positioning, not aggressive.

The $556M Q4 portfolio rebalancing (sold $112M equity, redeployed to fixed income) locks in $120M+ NII for 2026. At any combined ratio below ≈115%, EIG earns money. The question is whether CT pushes them above that line.

Cross-ticker check: SIGI's evidence confirms non-California workers comp trends are favorable. EIG's problem is California-specific, not industry-wide. This is genuinely idiosyncratic risk — it just happens to be idiosyncratic risk in a domain where we have zero edge.

The Consensus Problem

Two analysts cover this stock. Truist has a Buy at $58 (from November 2024 — stale). Forward P/E of 15.8x implies consensus expects $2.60 EPS in 2026, which requires combined ratio normalizing to approximately 100%. That's a 1,100 basis point improvement from 110.9%.

Is that possible? Sure — if CT frequency stabilizes and rate increases compound. Is it probable? I genuinely don't know, and neither does anyone without deep California workers comp actuarial expertise. The 10-K's own language ("accelerated beyond historical levels") suggests management isn't confident either.

Factor Decomposition

The negative SPY beta (-0.23) is structural and interesting — workers comp is counter-cyclical (recession → fewer workers → fewer claims). Natural portfolio hedge. But in a bull market, EIG is a drag. And we're not buying EIG for its hedging properties.

Verdict

EIG has the profile of a stock that rewards a specialist. Someone who reads WCIRB frequency reports, understands California litigation trends, tracks actuarial reserve development across accident years — they can evaluate whether 5% rate increases catch an 8-point loss ratio gap. That person might have 15-20% forward alpha here.

We're not that person. Our forward alpha is 1.2%. The CFO might be right. The discount to book might close. But we'd be buying insurance sector beta with a side of California litigation risk we can't assess, hoping a convergence trade works out. That's not edge. That's hope with a spreadsheet.

PASS.