WATCHconviction 3published 2026-09-11

CBNK — Capital Bancorp, Inc. · 2026-09-11 · Verdict: WATCH · Conviction 3

Price $36.36 (screen run 2026-09-11 close) · Mkt cap $592M · P/E 10.7x TTM GAAP (11.3x core) · P/TBV 1.55x · ROTCE 15.5% (Q2 2026) · TBVPS growth 13.6% y/y · Dividend yield 1.5% · CET1 13.14% · ADV $1.7M Sources read: 10-K filed 2026-03-16 (Items 1, 1A, 7), 10-Q filed 2026-08-07, DEF 14A 2026-04-07, 8-Ks 2026-03-18, 2026-04-27, 2026-06-01, 2026-06-17, 2026-07-27, 2026-08-14, Form 4s (12m). No analyst call transcript was in the bundle; the transcript file is the Q2 earnings release and investor deck.

Desk stats

1. What the business actually does

Capital Bancorp is the $3.9 billion holding company for Capital Bank, N.A., a branch-light commercial bank in the Washington-Baltimore corridor with seven branches (10-K 2025, Item 1). Four segments. Commercial Banking is the bulk: C&I, owner-occupied and non-owner-occupied CRE, construction, plus national verticals in lender finance and government-guaranteed SBA/USDA lending, funded by national deposit verticals serving homeowners associations, title companies and PACs. OpenSky is a nationwide digital secured credit card for credit rebuilders, $145.3M of net loans and $166M of matched deposits. Windsor Advantage, acquired with IFH, is an outsourced SBA/USDA servicing platform for other banks with a $3.4 billion servicing portfolio. Capital Bank Home Loans is a small mortgage arm that lost money last year. The unusual feature is funding: OpenSky's secured cards create noninterest-bearing deposits that collateralise their own credit lines, which is why headline NIM is 5.64% while core NIM excluding cards is 4.04%.

2. Why it is mispriced — the edge case

Honestly, there is not a strong one, and that caps the verdict at WATCH. At 1.55x tangible book for a bank earning 15.5% on tangible equity, this is roughly where the market prices banks of that quality. The one structural quirk, screeners reading a 5.64% NIM and a 10.4x P/E and concluding the bank is exceptionally profitable and cheap, cuts against the bull case rather than for it, because the card book that inflates NIM also charges off at 8.53% annualised (8-K 2026-07-27). The real debate is narrower: whether the 2026 earnings dip is a deliberate investment year that reverses, or the start of a credit and control problem. Insiders own 33.53% (DEF 14A 2026-04-07), so the register is not full of indifferent holders.

3. Credit quality, margin and deposits

Credit is the reason to hesitate. Nonperforming assets were $60.8M, 1.56% of total assets at 30 June 2026, up $24.7M y/y; nonperforming loans reached 1.85% of loans from 1.32%; substandard loans rose to 2.3% from 1.7% (8-K 2026-07-27). The deterioration is concentrated, with three relationships at 37% of nonperforming loans including a $9.7M legacy credit that went nonaccrual in Q2 and $15.9M from two acquired IFH loans, which is better than broad-based rot. But the reserve has not kept pace: allowance to nonperforming loans is 95.5%, down from 131.2% a year ago, and allowance covers only 35% of construction nonaccruals, which are 3.79% of that book (10-Q 2026-08-07). Total allowance is 1.76% of loans, down 9bp since December. Offsetting that, non-owner-occupied CRE is 18.2% of loans at a 48.5% weighted average LTV with office just $27.2M, 0.9% of loans. There is no office problem here.

On cards, net charge-offs ran 8.53% annualised in Q2 versus 6.82% a year earlier while gross unsecured balances grew from $32.7M to $51.2M (8-K 2026-07-27). Management has slowed: new-customer direct-to-unsecured offers are test-only, with balances "expected to remain de minimis through year end as management monitors performance." Prudent, and it also removes a growth engine. Funding is a genuine strength: loans to deposits 91.5%, noninterest-bearing 26.6% of deposits, cost of deposits 2.29% and falling, insured and protected deposits 66.6%.

The control environment is the other problem. The FY2025 10-K disclosed a material weakness and concluded that internal control over financial reporting and disclosure controls were not effective at 31 December 2025, with remediation still in progress at the Q2 filing (10-K 2025, Item 1A; 10-Q 2026-08-07). On 14 August 2026 the company dismissed Elliott Davis and engaged Crowe, citing the need for "a larger independent registered public accounting firm with additional resources and industry specialization" and naming the material weakness as the only reportable event (8-K 2026-08-14). No disagreements, and hiring a bigger bank auditor is the right response, but a first-year audit over a control environment already known to be weak is when surprises surface.

4. Balance sheet and capital allocation

CET1 13.14% and Tier 1 leverage 10.59% leave room to grow. Share count fell from 16.58M to 16.29M over the year, and the prior authorisation retired 419,643 shares (8-K 2026-03-18). But the current $15M authorisation, effective 16 March 2026 and expiring 31 December 2026, is barely used: $3.5M in Q1, then 1,213 shares for $36 thousand in Q2 at an average of $30.03, with $12.4M unspent (8-K 2026-07-27). A bank that believes its own 13.6% tangible book compounding does not stop buying at 1.3x tangible book. The dividend was raised 16.7% to $0.14 instead. The board has proposed eliminating its classified structure and all supermajority provisions (8-K 2026-06-17), a real governance improvement. CEO Ed Barry's 2025 total compensation was $1.38M with incentives tied to core EPS and ROTCE; the company missed its commercial deposit, DDA and OpenSky secured targets and the payout reflected it (DEF 14A 2026-04-07). Form 4s show no open-market buying or selling in twelve months, only grants and tax withholding.

5. Management: said versus did

In April the chairman said organic growth "permits us to accommodate an increase in noninterest expenses, while, at the same time, providing our stockholders with reasonable returns" and that they "have not yet seen any macroeconomic signs of credit deterioration in our markets" (8-K 2026-04-27). One quarter later nonperforming assets rose again and a $9.7M legacy relationship went nonaccrual. The expense increase was delivered: core noninterest expense went from $38.2M in 2Q25 to $43.2M in 2Q26 and the core efficiency ratio from 62.8% to 66.1%. The returns have not followed. They were straight about the spending and too early on credit.

6. Valuation

Base (50%): ROTCE settles at 13-14% while investment spend runs and card losses stay elevated; tangible book compounds about 11% to roughly $29 by end-2027; multiple holds at 1.5x, giving about $43.50 plus dividends, roughly +21% over five quarters. Bear (25%): the legacy construction credit and the two acquired IFH loans resolve badly, reserves are rebuilt from 95% coverage, card charge-offs pass 10%, EPS falls toward $2.75 and the multiple derates to 1.15x tangible book on control doubts, giving about $28.50, roughly -21%. Bull (25%): remediation is confirmed clean under Crowe, credit stabilises, Windsor scales with industry SBA volumes, efficiency returns to 62% and ROTCE to 17%, multiple 1.9x on about $27 of tangible book, giving about $51, roughly +40%. Probability-weighted value is about $41.60, roughly 14% above price over five quarters plus a 1.5% yield. Reverse DCF: at 1.55x tangible book with an 11% cost of equity, the price implies a sustainable return on tangible equity of about 12.7% if tangible book compounds at 8%, or today's 15.5% with only 2.8% growth. The market already pays for most of what the bank earns now and prices neither an acceleration nor a collapse.

7. Catalysts and timeline

Q3 2026 results in late October: whether nonperforming assets stabilise, whether allowance coverage returns above 100%, and the first quarter under Crowe. The FY2026 10-K around March 2027 carries the remediation statement. Resuming the buyback before the $12.4M authorisation expires on 31 December 2026 would be a strong signal.

8. Risks and pre-registered kill criteria

Specific risks are credit concentration in a few large relationships, SBA and USDA program dependence at both the bank and Windsor (10-K 2025, Item 1A), federal workforce and spending reductions in the DMV market (10-Q 2026-08-07), and the control environment. Kill criteria: 1. Nonperforming assets above 2.0% of total assets, or allowance to nonperforming loans below 85%, in two consecutive quarters. 2. Credit card net charge-offs above 10% annualised in two consecutive quarters, or gross unsecured card balances above $75M without that rate falling. 3. The material weakness not reported as remediated in the FY2026 10-K, or any restatement, late filing, or second reportable event under the new auditor. 4. Core efficiency ratio above 68% in two consecutive quarters, meaning the investment spend is not producing operating leverage.

9. Verdict and one-paragraph summary

WATCH, conviction 3. Capital Bancorp is a genuinely good small bank: a 15.5% return on tangible common equity, tangible book per share compounding 13.6% a year, 26.6% noninterest-bearing deposits, almost no office CRE, 13.1% CET1, and a board and management team owning a third of the company. The problem is that at $36.36, or 1.55x tangible book and about 11x normalised earnings, you pay a fair price for that quality just as three things turn the wrong way at once: first-half earnings fell 3% because core margin compressed 38bp while expenses grew 12%; nonperforming assets nearly doubled year over year to 1.56% of assets with reserve coverage of nonperforming loans slipping below 100%; and the company carries an unremediated material weakness in internal control while changing auditors. None is fatal and all three are observable each quarter, which is why this is a watch rather than a pass. The trigger to buy is a quarter showing nonperforming assets flat or down with coverage rebuilt, plus management restarting the buyback it stopped at $30 a share.

Research for discussion, not investment advice. Positions and sizing are the reader's decision.

Source markdown: 2026-09-11_CBNK.md · how these notes are built · every verdict tracked since publication.

Research and education only. Nothing here is investment advice or a recommendation to buy or sell any security. No price targets are recommendations; positions and sizing are the reader's decision. Past performance does not predict future results.