Setup

BMRC reported Q1 2026 EPS $0.53, missing consensus $0.55 by 3%. Stock fell 2.6% on the print to $25.20.

That's the wrong number. The number that matters is in the credit footnote: non-accruals collapsed from 1.27% to 0.41% in one quarter, and every dollar of the $7.3M of Q1 charge-offs was already reserved against with specific allowances built in Q4. Net P&L impact from the cleanup: zero.

This is not credit improving. This is demonstrated forecasting precision — management identified the problem credits six months ahead, took specific reserves against them in Q4, and charged them off cleanly in Q1. The Q4 2025 restructuring promised $0.40 EPS accretion and 25bps NIM lift over twelve months. Q1 delivered both, with the credit clean-up running ahead of guide.

What the filing says

Credit (the meta-signal): NPL 1.27% → 0.41%. Classified 1.51% → 0.85%. Zero provision. ACL coverage 1.12x → 2.64x. Remaining $8.6M non-accrual is concentrated in a single $8.2M NOO CRE relationship described as a technical extension dispute with no expected loss.

NIM: Tax-equivalent 3.24% (+54bps YoY, +6bps QoQ). March spot 3.26%. Securities repositioning delivered the projected +21bps. Deposit beta 25% cycle-to-date — well below the 33% management threshold. CFO guides further expansion through 2026.

Earnings & capital: Diluted EPS $0.53 (+76.7% YoY). ROA 0.87%, ROE 8.67%, efficiency 66.0% vs 75.7% YoY. Q1 fundings $60.8M = highest Q1 since 2015 (+28% YoY). Loan yield 5.04%. LTD 61.72%. Total RBC 15.26%; $2.185B liquidity = 221% uninsured deposit coverage. 84th consecutive $0.25 dividend.

What the market thinks

$25.20 (-2.6% on print). Forward P/E 9.75. P/B 1.03x (book $24.37). Mean analyst target $29.60 (+17.5%); range $28-31. Piper Sandler downgraded to Neutral $28 on April 2 — pre-print, modeling the optical miss. Insider Form 4 over 90 days: all Code A awards, zero Code P purchases. Short interest 4.3% float, 6.9 days to cover.

Probability-weighted 12-month return:

Probability-weighted +10-14% sits below the sell-side mean implied +21%. The reason is timing, not disagreement: the sell-side targets were last touched between October and February — before this print, before the credit footnote, before the cross-ticker confirmation. Models updating over the next 4-8 weeks should drift target ranges to $30-33. The gap likely closes from above. Our number is also more risk-adjusted: we explicitly weight the 25% of mass in stall + break states that consensus targets do not show.

Why the gap exists

The counterparty. Three sellers, none of them informed about the credit footnote. (1) Tape readers and earnings-quality algos parsed the 3% EPS miss and Piper's pre-print Neutral and trimmed. (2) Index/ETF rebalancers — BMRC sits in KRE and KBE; California regionals are bifurcating, with BANC, CVBF, and WAL surfacing fresh deterioration the same quarter, and generic sector flow doesn't distinguish. (3) Anchored holders selling on optical miss without re-reading the press release. None of these mechanisms read the specific-allowance mechanism. The seller has rules, not edge.

Cross-ticker confirms the move is idiosyncratic, not sectoral. BMRC's 86bp NPL collapse is 3-9x any California regional peer's Q1 movement. WABC, the closest Northern California comp, stayed flat. BANC saw NPLs rise and coverage fall — the inverse pattern. The bifurcation is real and isolates BMRC's workout execution as the alpha source.

Idio variance is structurally below target. Estimated 40-55% before hedging — meaningful sector and style factor weight is unavoidable for a $400M cap regional. A KRE short overlay at ≈0.5x of BMRC notional strips the bank-sector beta and lifts %idio toward 55-70%. Still below Paleologo's 75% but in monitor territory. Bay Area CRE exposure (intentional thesis component) and the 3x floating-asset rate sensitivity (structural NIM tailwind) are kept, not hedged.

The bull evidence cluster is correlated. NIM expansion, credit cleanup, EPS accretion, and loan-yield uplift are all downstream of the Q4 2025 restructuring working as designed. That is one signal at approximately LR 1.5, not four signals stacking to LR 2.94. After compressing for correlation and including bear evidence (optical miss, Code A insider grants only, $25M buyback authorization untouched two quarters) the net memo signal is approximately 1.4.

Risks

1. NIB deposit franchise compression (highest velocity). NIB share dropped 36.7% → 35.9% on a $22M outflow management called "one anticipated client event" — unverified Tier 2 management color. Update trigger: Q2 8-K NIB deposits down ≥$30M further, or NIB share <35.0%. This is the fastest-moving risk in the book.
2. Bay Area CRE single-name surprise. $8.2M extension-dispute non-accrual or a new single name from the existing book. Update trigger: any 8-K Item 8.01 material-credit-charge disclosure before Q2 print, or Q2 provision >$1M.
3. Aggressive Fed cuts. 3x floating-asset/liability mix is asymmetric. Down-200bps simulation = -4.4% Year 1 NII. Update trigger: FOMC cuts >50bps cumulative before September AND deposit beta failing to compress proportionally.
4. KRE sector contagion. Update trigger: KRE down >5% with BMRC underperforming KRE by >3pp in a 30-day window — would suggest sector beta dominates regardless of stock-specific story.
5. Wine industry deterioration. ≈$21.6M OO-CRE exposure (7% of OO-CRE). Update trigger: any new wine credit downgrade in Q2 8-K shifts this from "watch only" to "reduce."

Catalysts

What would change our mind

• Q2 NIM <3.25% with deposit beta climbing through 30% — structural expansion was a one-quarter peak, not run-rate.
• New single-name non-accrual >$15M — workout was lucky on identified credits, not skilled at finding all of them.
• NIB deposits down another $30M+ in Q2 — confirms franchise compression, "one-off" was misdirection.
• Two consecutive quarters of provisioning >$1M — specific reserves no longer ahead of charge-offs.
• KRE-relative price weakness despite better idio numbers — sector beta dominates the stock-specific story.
• $25M buyback authorization expires/replaced unused — confirms the hoarding pattern (LR 0.7-0.9 bearish per our taxonomy).

Evidence

* Items marked are correlated — all downstream of the Q4 2025 HTM restructuring working as designed. Treat as one Tier 1 cluster at effective LR ≈1.5, not as independent stacking signals.