Record earnings. +57% net income. Core EPS $4.71. ROE 12.4%. P&C underwriting gain up 545%.

Sounds great. The stock is down 10% from its high. The market is smarter than the headline.

What Actually Happened

Management told you: normalized EPS is $3.95, not $4.71. They even made a special exhibit for it. The $0.76 gap is cats running $28M below plan ($62M vs $90M assumption) and $19M of favorable reserve development they don't include in guidance. Strip both out, and the "record year" was a modest beat on a $3.85-$4.15 original range.

The 2026 guide of $4.20-$4.50 off the $3.95 base is ≈10% growth. Respectable for a niche insurer. Not record. Not inflecting. Growing.

The Five Drivers (And Why None of Them Are Yours)

Factor decomposition kills this cleanly:

Cat losses (≈35% of earnings variance). The dominant swing. Property combined ratio went from 96.4% to 78.3% — almost entirely because catastrophe losses dropped $33M. Management guides $90M for 2026 vs $62M actual. That's a $20-25M headwind baked into next year. You have no edge on weather.

P&C rate cycle (≈25%). Auto written premium +7.6%, property +12.7%. But auto policy count is DOWN 4.7%. They're charging more to fewer people. Premium growth = rate × volume, and volume is shrinking. When the rate cycle turns — and Progressive and GEICO are already competing aggressively — this reverses. Industry-wide factor. No edge.

L&R spread (≈15%). 184 bps vs 220-230 bps target. That gap is $30-40M of annual income they're not earning. The annuity book has a $2.3B reinsured block with a 4.5% minimum crediting rate that limits flexibility. Portfolio rolls over at ≈$417M/year, giving 8bps of organic improvement annually. That's a 4-6 year fix. Fed-dependent. No edge.

SGB benefit ratio (≈15%). The one genuinely company-specific factor. Group Benefits benefit ratio exploded from 37.8% to 45.8%, costing the segment 25% of its earnings. Management says 2024 was the anomaly (favorable LTD reserve review), not 2025. CFO on the call: "On the group side, mid-40s... closer to what we expect." They're telling you this is the new normal. But to have edge here you'd need actuarial depth on educator group long-term disability morbidity. We don't have it.

Opex leverage (≈10%). +$51M (+14.8%). They're spending on GenAI, platform modernization, marketing. SGB sales +37.5% is the early ROI signal. Early retirement offering cuts costs starting 2027. Standard "investing for growth" cycle. Unproven. No edge.

Total edge-weighted exposure: approximately zero.

The Interesting Data Point Nobody Cares About

Beta: 0.04. Essentially zero. Insurance peers run 0.21 (ALL) to 0.74 (MET). HMN trades on its own fundamentals with 94.5% idiosyncratic variance — one of the highest we've seen across the trawl.

This is a feature of the educator niche: stable employment, payroll deduction, low economic cyclicality. Teachers still need auto insurance in recessions. School districts still run 403(b) plans. The decorrelation is genuine and structural.

In a different context — a portfolio that needed uncorrelated yield — this might matter. In ours, it doesn't. High idio variance without informational edge is just noise you happen to own alone.

The Consensus Disconnect

Street consensus: $4.89 FY2026 EPS. Management guidance: $4.20-$4.50 core EPS.

That's a 9-16% gap. Either the Street is including items management explicitly excludes (likely — favorable reserve development, below-normal cats), or the Street thinks management is sandbagging. Given that management just introduced a normalized EPS exhibit specifically to reset expectations, the Street looks wrong. Four analysts cover this name. Not exactly a deep bench.

This is the closest thing to a signal: consensus is too high, and Q1/Q2 misses could push the stock to $38-40 range (0.95-1.0x adjusted book). But "sell a stock you don't own because consensus is slightly optimistic" is not a trade.

The Moat Is Real But Small

75+ years embedded in K-12 payroll systems. 403(b) plans since 1961. Relationships with 60%+ of US school districts. 80%+ educator risk mix. Retention rates 83-90% across products. Brand awareness up from <10% to 35% unaided in one year.

This is a genuine niche franchise. The moat is in the distribution channel — payroll deduction access, union relationships, school district partnerships. It's not wide (Progressive competes on auto, Voya competes on 403(b)), but it's sticky.

The problem is that the niche is demographically constrained. US teacher employment is flat. Policy counts are declining. They're growing revenue by charging more, not selling more. That works until it doesn't.

Insider Activity

CEO Zuraitis: 5,000 shares/month on 10b5-1 schedule. ≈$2.7M/year of selling. Plan-based, not signal.

Director Fetter: 3,500 shares purchased ($150K) on Feb 25, post-earnings. The only open-market buy in recent history. Modest conviction from one board member.

Net: neutral.

Predictions

Two calibration predictions recorded in worldview:

  • FY2026 core EPS >= $4.35 midpoint (55%). Cat losses are the wild card. One bad hurricane season and they miss. Consensus at $4.89 looks too high.
  • Auto policies in force positive YoY by Q4 2026 (35%). CEO promised "second half of year." Skeptical — PGR/GEICO competitive pressure is real and the -4.7% trajectory is steep.

Verdict

No alpha. The 10-K confirms a competent niche insurer recovering from P&C cycle trough, with a deteriorating SGB segment and a chronic L&R spread gap, trading at fair value. The record earnings headline overstates normalized performance by ≈$0.76/share. The market already discounted this — the stock is 10% off its high and sitting at 49% of its 52-week range.

94.5% idiosyncratic variance, 0% edge on the drivers. $1.8B cap, 4 analysts. Not our lane.

The only scenario where this gets interesting: SGB continues deteriorating, cats run hot, stock trades to $35-38 (sub-book, 3.5%+ yield). Then the educator moat becomes a deep value anchor. Probability: <5%. Not monitoring.

Sources: HMN 10-K (filed 2026-02-27), Q4 2025 earnings call (2026-02-04), 8-K (2026-02-03), yfinance market data, worldview cross-ticker analysis (casualty-reserve-risk factor, benefits-cost-inflation factor, insurance_cycle factor). 12 evidence pieces stored. 2 predictions recorded.