This is investment research, not personal financial advice.

Judo's midday story is not the 1.6% share-price move. The larger point is that a specialist SME bank has just had to separate two truths that can sit uncomfortably together: loan growth and margin momentum remain strong, while asset quality has deteriorated enough for management to reset FY26 cost-of-risk expectations. For a lender built around relationship underwriting, the distinction between three borrower-specific problems and a broader portfolio signal is the entire research question.

The 25 June ASX update gives the market more than a routine trading note. It provides enough new information to test the credit thesis: FY26 cost of risk is now expected at $116-122 million, 90-days-past-due plus impaired loans are expected to be about 3% of gross loans and advances at 30 June, and profit before tax is still expected to grow about 30% from FY25. The next result, scheduled for 18 August 2026, becomes the first hard checkpoint.

The credit reset is the story, not a footnote

Judo said the increase in specific provisions was driven primarily by three exposures across different sectors that had recently emerged after customer-specific developments. That phrasing matters. If the three loans are genuinely idiosyncratic, the business may still be moving through a contained provisioning event while continuing to scale. If they are early examples of broader SME stress, the valuation framework changes: book value becomes less certain, growth consumes more capital, and reported profit before tax becomes a less reliable guide to distributable earnings.

The new numbers are specific enough to frame the issue. Judo expects FY26 cost of risk of $116-122 million. It expects 90DPD+ and impaired loans to be about 3% of GLA at 30 June. Collective provision coverage is expected to remain broadly in line with the Q3 trading update, at 94 basis points of GLA or 1.09% of standardised credit risk-weighted assets, including a management overlay for sectors affected by macro uncertainty.

Against that, the operating machine has not stopped. GLA was above $14.4 billion on 24 June and is expected to be $14.6-14.7 billion by 30 June. NIM is expected to be above 3.2% for 2H26, ahead of the prior guide of about 3.15%, supported by term-deposit costs. Front-book and blended lending margins were stable at 4.2% over one-month BBSW in April and May, and the AAA lending pipeline was $2.4 billion at the end of May with a margin of 4.3%.

That split defines the event. Credit has moved against the story, but margin, lending and operating-cost guidance have not. The base question is whether Judo is seeing a normal lumpy loss pattern in a concentrated SME book or the beginning of a higher-loss regime.

What Judo is trying to compound

Judo is a business bank focused on Australian SMEs. Its customer proposition is relationship-led lending: bankers dealing directly with business owners rather than routing smaller commercial borrowers through highly automated major-bank channels. Revenue is earned mainly from the spread between loan yields and funding costs. The key assets are the loan book, the banking licence, the credit process, the deposit base, securitisation access and the banker network that sources and monitors borrowers.

That model can create value if three conditions hold at the same time. First, the bank needs to originate loans at a spread wide enough to compensate for funding, operating cost, capital and credit losses. Second, credit underwriting must remain good enough that growth does not simply pull future losses forward. Third, the capital base must support loan growth without repeated dilution at unfavourable points in the cycle.

The compounding engine is not revenue growth by itself. It is growth in risk-adjusted lending assets at returns on equity above the cost of equity. Judo's June update shows the two sides of that engine. The positive side is a high NIM, stable lending margins and lower expected cost-to-income in 2H26 than the 48.5% reported for 1H26. The pressure point is that specific provisions can absorb a large portion of the earnings lift when a small number of borrowers deteriorate.

For a bank, conventional free cash flow is not useful. Loan growth is not a simple capital expenditure decision, and deposit flows distort operating cash flow. The owner-economics question is distributable earnings: normalised profit after credit costs, less the capital needed to fund loan growth and maintain a regulatory buffer. On that basis, Judo is still in a reinvestment phase. Earnings are rising, but capital is being retained to support a larger balance sheet and a management CET1 range.

The financial evidence: operating leverage versus credit volatility

The financial history shows a bank moving from listed challenger to profitable scale player. The table below uses reported or source-derived headline figures and treats ROE as the relevant return metric. The FY2026 line is an estimate based on the company's June PBT guide and should be read as a bridge, not an audited result.

Year	Revenue / operating income (A$m)	NPAT (A$m)	Return metric	What mattered
FY2023	407	12	ROE about 1.1%	Profitability emerging but scale still thin
FY2024	503	39	ROE about 3.6%	Operating leverage became visible
FY2025	574	89	ROE about 8.0%	Profitability approached a more bank-like level
FY2026e	650	116	ROE about 10.0%	June credit reset partly offsets growth and NIM

The incremental-return calculation carries the weight. Using NPAT as a practical proxy for bank NOPAT and equity capital as the invested-capital base, NPAT increased by about $77 million between FY2023 and the FY2026 estimate. If average ordinary equity increased by roughly $250-300 million over the same period to support balance-sheet growth, the implied incremental return on retained capital is about 26-31% before considering the new credit-loss volatility. That looks strong, but it is backward-looking and helped by operating leverage from a low starting point.

The forward incremental return will be decided by credit cost. At a FY26 PBT guide of $163-169 million, a $116-122 million cost-of-risk line is large enough to dominate the year-on-year interpretation. If that cost falls back toward the through-cycle proxy of 50 basis points of average GLA, the same platform can report materially better returns. If it remains elevated, growth consumes capital at a lower return and the compounding case is diluted.

Capital relief changes the ROE equation, but not the credit question

The May 2026 capital-relief securitisation is the other primary filing that matters. Judo priced a $750 million transaction backed by SME business loans, upsized from an initial $500 million launch amount. The weighted average note margin was 171 basis points over one-month BBSW, a 102 basis-point improvement from the September 2023 inaugural transaction. Management said the transaction qualified for regulatory capital relief, lifted pro-forma CET1 at 31 March from 12.6% to 13.2%, and was estimated to add 25-30 basis points to FY27 ROE assuming a normalised capital level.

This helps capital flexibility. It also needs to be read with care. The loans remain on Judo's accounts and continue to generate interest income; the transaction affects regulatory capital, not the existence of the underlying credit exposure. Capital relief can improve measured ROE and create optionality, but it does not by itself prove the loan book is cleaner. The June update puts the two filings in tension: capital tools are improving just as credit outcomes have become more visible.

At 30 June 2026 Judo expects CET1 of about 12.4%, including the capital-relief transaction. The bank is moving to a normal operating management target range of 11.0-12.0%. That leaves some buffer above target, but the quality of the buffer depends on asset quality. A bank with contained credit losses can operate with a narrower capital surplus; a bank with rising arrears needs more room for adverse migration.

The moat is relationship underwriting, and the counter-evidence is also credit

Judo's moat claim is not a scale moat in the major-bank sense. It does not have the lowest-cost deposit franchise in Australia. It does not have a mortgage book funded by enormous household deposit balances. Its advantage, if durable, is more specific: relationship bankers, SME credit focus, and service levels for business borrowers that can be underserved by larger institutions.

The evidence supporting that proposition is the lending margin. Stable front-book and blended lending margins of 4.2% over one-month BBSW suggest Judo is not competing solely on price. A $2.4 billion AAA lending pipeline at 4.3% over BBSW suggests there is still demand for the model. The expected 2H26 NIM above 3.2% is high for a bank and shows that the spread engine remains intact.

The counter-evidence is the credit reset. Specialist lending only compounds if the specialist underwriting advantage appears in losses through the cycle. A few large SME exposures can be lumpy; that is part of the business. But lumpy is not the same as harmless. When 90DPD+ and impaired loans move toward 3% of GLA, investors need the next report to show whether the deterioration is contained, adequately provisioned and not spreading into cohorts that were previously performing.

The moat classification is mixed: stable in customer proposition and funding/capital access, eroding in the one area that matters most this quarter, which is underwriting evidence. The next two disclosures should decide which side deserves more weight.

What the market price appears to require

At about $0.90 and an equity value near A$1.03 billion, Judo is being valued around the point where the FY27 PBT guide starts to matter. If FY27 PBT is A$210-220 million and a normal tax rate is applied, NPAT could be roughly A$147-154 million before any material difference between accounting profit and distributable earnings. Against a A$1.03 billion equity value, that is roughly 6.7-7.0 times that simple forward earnings bridge. The multiple looks modest, but banks do not deserve a multiple simply because earnings rise; they deserve a price-to-book and earnings multiple based on sustainable ROE, capital intensity and credit risk.

A residual-income frame is cleaner. Value equals tangible equity plus the present value of returns above the cost of equity. If Judo can sustain low-to-mid-teens ROE on a growing equity base, value can sit above book. If sustainable ROE is 8-9%, value is closer to book because the spread over the cost of equity is thin or absent. If credit losses remain elevated, book itself needs more scepticism.

The four scenarios below use that bank logic rather than a generic industrial DCF.

Scenario	Operating assumption	ROE / credit interpretation	Value range
Severe downside	Three exposures broaden into a portfolio problem	Cost of risk remains above 100bps; ROE below cost of equity	$0.45-0.60
Bear	Credit reset is partly broader but survivable	FY27 PBT undershoots; sustainable ROE 8-9%	$0.65-0.80
Base	Credit normalises toward the 50bps through-cycle proxy	FY27 PBT guide broadly achieved; ROE low double digits	$0.90-1.10
Bull	Three exposures are contained and operating leverage continues	Capital relief and scale move ROE into low-to-mid teens	$1.20-1.45

A simple sensitivity shows why the credit line matters more than the headline loan growth.

Sustainable ROE	Cost-of-equity spread	Implied valuation posture
8%	Below or near cost of equity	Value close to tangible-book support, with little residual-income value
10%	Modest positive spread	Current price can sit around base-case territory if credit stabilises
12-14%	Clear positive spread	Bull case requires proof that losses normalise while growth continues

The reverse question is straightforward: at $0.90, the market appears to be pricing neither a collapse nor a clean low-loss compounder. It is pricing a contained credit event with enough uncertainty to keep the value around the base range until the August result gives more evidence.

What would change the story

The crux is not whether Judo can grow the loan book. The June update already says lending momentum remains solid. The crux is whether the bank can grow the loan book while pushing cost of risk back toward its through-cycle proxy and preserving CET1 in the new target range.

The first catalyst is the FY26 result on 18 August 2026. The key line items will be the ageing of the three exposures, the composition of impaired loans, sector concentration, collective provision coverage, and any movement in stage migration. A result showing the three exposures well identified, adequately provided and not accompanied by broader arrears would support the isolated-event interpretation. A result showing wider migration would move the analysis toward the bear case.

The second catalyst is the FY27 half-year result. By then the June provision reset should either be flowing into realised losses, stabilising, or being joined by new problem names. That will matter more than another strong GLA number.

The third catalyst is capital. The securitisation shows Judo can access capital-relief tools on better pricing than in 2023. The monitoring point is whether those tools lift ROE while the ordinary capital ratio remains comfortably inside management's range, not whether capital relief temporarily flatters a ratio.

Monitoring plan

Metric	Threshold	Analytical signal
90DPD+ and impaired loans / GLA	Around or above 3% beyond the FY26 result	Weakens the isolated-exposure interpretation
Cost of risk	FY27 run-rate materially above 50bps of average GLA	Lowers sustainable ROE assumptions
Collective provision coverage	Falling while impaired assets rise	Raises questions about reserve adequacy
CET1 ratio	Below 11.0-12.0% target without clear growth explanation	Reduces capital optionality
NIM	Sustained below 3.10%	Reduces confidence in operating leverage
GLA growth	Growth paired with worsening arrears	Signals that volume may be arriving with weaker risk-adjusted returns

Confidence and source notes

Confidence is highest for the June 2026 ASX update and the May 2026 securitisation announcement, both fetched directly through the ASX MarkitDigital file endpoint. Confidence is medium for the multi-year history table because the figures were compiled from company reporting history and reconciled for article purposes, but the company investor-relations page was blocked by Cloudflare from this runner during the scheduled job. For that reason, the article uses partial verification rather than full verification.

The missing information that matters most is borrower-level detail on the three deteriorated exposures, sector-by-sector arrears migration, and the precise split between individual and collective provisioning at 30 June. Those details should be available, at least in aggregate, in the FY26 result. Until then, the observation is narrow: the current price appears to sit around the base-case range only if the June credit reset is contained and FY27 earnings power remains close to management's guide. The next filing decides whether that is a credit lump or a credit trend.