Variant Perception

Where We Disagree With the Market

Consensus is pricing PTSB as deal #15 of an unbroken 14-deal flywheel; the evidence says this transaction is structurally different in four specific, observable ways — and the 14-deal pattern is not the right base rate. The 13-analyst sell-side panel sits at €174.80 with 12 Buy ratings, P/TBV has re-rated from 1.36x (end-2023) to 2.99x (FY2025), and Fitch has already labeled the deal "transformational, synergy potential significant" without quantifying anything. We agree the platform is real (36% CIR, 27% RoTCE, 14-of-14 self-funded acquisitions, CEO open-market buy three weeks before this report). We disagree on the assumption inside the multiple that PTSB integration economics rhyme with Knab and Barclays Consumer. The H1 FY2026 print on 21 July 2026 publishes the first three resolving variables — underlying risk costs, standalone CIR, and the CET1 walk to ~17% — on a known date; this is a cleanly testable variant view.

Variant Perception Scorecard

The score reflects an unusual setup: consensus is loud and observable (13 analyst PTs, 12 Buys, an explicit re-rate path on the chart), and the variant disagreement is bounded by published dates (21 July 2026 H1 print, Jun-Aug 2026 PTSB scheme circular, Q4 2026 / Q1 2027 deal closing). Variant strength is 72, not higher, because the deal-execution risk is widely discussed in the public record — what is mispriced is not the existence of the risk but the assumption that 14 prior integrations supply the right base rate. Evidence strength of 76 reflects a concrete trail across forensics (overlay release, Day-1 gains), competition (Irish spreads, MREL drag), and web research (no quantified synergies, branch preservation). The resolution window is short enough that this is an underwriting question, not a debate.

Consensus Map

The map shows where the loudest signals sit. PTSB integration and the multiple are High-confidence consensus reads — observable in price, ratings, and analyst targets. The €100B threshold is the quietest assumption: nobody is modelling MREL subordination explicitly, and the scale milestone is treated as upside rather than a structural cost. Earnings quality is Medium-confidence consensus because forensic findings have been published in research notes (the Petrus precedent) but have not moved the multiple — the market has chosen to read past them, which is itself the assumption.

The Disagreement Ledger

Disagreement #1 — PTSB structural difference. Consensus says: BAWAG has done 14 deals, every one of them at >20% RoTCE on deal capital, integrated within ~12 months, self-funded. Apply the same template to PTSB. Our evidence disagrees on four specific structural variables: Irish mortgage NIMs were already compressing 15-20bps in FY25 in a 3-bank oligopoly that will defend share against a foreign acquirer; the state-backed scheme commits BAWAG to preserving the 98-branch network (which inverts the cost-synergy playbook that drove deals 1-14); the 26% premium versus pre-process price contrasts with the 0.6-0.8x book purchase discipline of the prior 14; and the 40% one-shot balance-sheet expansion is an order of magnitude larger than every prior bolt-on. If we are right, the market has to concede that the next deal is the wrong base rate for projecting deal #16 — meaning the platform multiple is paying for an option that is not transferable. The cleanest disconfirming signal is mgmt finally publishing a quantified cost-vs-revenue synergy split at the BofA Sep-23 conference; if cost-synergy guidance lands inside the 15-20% peer benchmark, our objection collapses.

Disagreement #2 — RoTCE run-rate vs FY24 peak. Consensus reads 27% as the platform output and projects 25%+ through the cycle. Forensics shows the headline number was lifted by two distinct, non-recurring sources: a fully released ECL management overlay (worth ~€100-150M of provision relief in FY24) and ~€140M of acquisition Day-1 gains and investment-property revaluation across FY24-25. The FY25 risk-cost step from 16 to 41bps is the first half of that adjustment; Q1 2026 at 46bps is the second half running. If we are right, the market has to concede that the FY26 NP consensus of €990M is built on the wrong baseline — the right run-rate net profit is closer to €750M and the right multiple is closer to its own pre-2023 regime of 1.5-2.0x P/TBV. The cleanest disconfirming signal is a 21 July H1 print where standalone underlying risk costs (ex-Knab/Barclays mix change) come in below 40bps and CIR converges to ≤33%; both metrics inside their platform bands would confirm the consensus run-rate read.

Disagreement #3 — Self-funding tail. This is the lowest-conviction of the three because mgmt has been transparent about the capital path and the SRT toolkit is real. The variant view is narrower: the disclosed 250bp build is one-shot (the H1 dividend skip cannot be repeated), market-dependent (SRT pricing is the swing variable), and unrated by Moody's at a moment when crossing €100B triggers MREL subordination requirements. If any one of those breaks, "alternative actions" becomes the live option. Our claim is not that an equity raise is the base case — it is that the option is currently priced at zero, and the realized vol at p80 with price at fresh highs already shows the market paying a stressed risk premium for something. The cleanest disconfirming signal is an H1 disclosure showing the SRT priced inside model and the CET1 walk delivered as guided; the cleanest confirming signal is "alternative actions" language anywhere in the H1 commentary.

Evidence That Changes the Odds

Binder Error: Set operations can only apply to expressions with the same number of result columns

The evidence table is intentionally ordered by what would change the odds for an institutional reader, not by importance. Items 1, 4, and 5 are the most consequential for the variant view: they convert qualitative observations into adjustments on the consensus model. Item 7 is included as a counter to the variant read, and the fragility column is the auditing surface — if cost-synergy guidance lands inside peer norms, if H1 risk costs converge to platform levels, or if PTSB PPA is clean, the variant view loses force on those specific dimensions.

How This Gets Resolved

The resolution path is short and well-instrumented. Two of the three disagreements (RoTCE run-rate, self-funding) settle on the 21 July H1 print; the third (PTSB structural difference) settles in two stages — synergy quantification at the BofA conference on 23 September 2026, and Day-1 PPA at closing in Q4 2026 / Q1 2027. There is no "wait years to find out" tail here. A PM holding the position can size the variant view to a defined event window; one not holding the position can use the same calendar to time entry without committing in front of the disclosure.

What Would Make Us Wrong

The strongest disconfirming evidence sits inside the operating record itself. We have argued PTSB is structurally different from deals 1-14 — but management has integrated each prior bolt-on at >20% RoTCE on deal capital, including Knab (now disclosed as "integration largely complete" at Q1 2026) and Barclays Consumer Bank Europe (closed Feb 2025 with €18.4M Day-1 gain). The base rate is 14-of-14, not 9-of-14 or 12-of-14. If H1 risk costs converge to the 35-40bps range on the underlying book, standalone CIR holds ≤33%, and the SRT toolkit prices inside model, the variant view loses force on disagreements #2 and #3 simultaneously. The remaining objection (#1, PTSB structural difference) has a clean falsifier at the BofA conference: if cost-synergy guidance lands inside the 15-20% peer benchmark and revenue synergies are quantified rather than waved at, the structural-difference critique converts into ordinary execution risk.

We could also be wrong on the multiple. The 2.99x P/TBV is one standard deviation above the 2019-2023 regime, but the regime change is real: 27% RoTCE simply does not exist in the European bank universe at 1.5x P/TBV. CBK at 17.5x P/E on 8% ROE shows that the market pays for ROE, and BAWAG at 12.8x P/E on 27% RoE is — by a sector-relative lens — actually cheap. If we are mistakenly comparing post-rate-cycle BAWAG to pre-rate-cycle BAWAG, the right comparison is post-rate-cycle BAWAG to post-rate-cycle peers, where the platform gap is structural and the multiple is paying for what the operating record continues to deliver.

The CEO's open-market buy of €1.19M at €147.50 three weeks before this report is the single hardest piece of disconfirming evidence to dismiss. We have argued that insider buying is a floor, not a ceiling — but the same CEO bought through the 2022 Linz write-off, the 2023 Petrus short report, the 2020 COVID overlay, and 13 prior integrations, and was right every time. A variant view that requires the CEO to be wrong on the deal he negotiated is a high bar. Our defense is narrow: the buy says "I work here", not "PTSB cost synergies will reach 15%"; we are not disputing alignment, we are disputing pattern-matching. But this is the asymmetry to take seriously.

Finally, we may be early. Even if the variant view is right that PTSB is structurally different, the resolution doesn't fully settle until 2027-2028 — by which time the multiple may have expanded further on momentum and the cost of being wrong-and-early is the foregone return on the platform's continued compounding. Our defense is the dated calendar: 21 July H1 and 23 September BofA both publish before any 2027-2028 resolution, and either-or both can move the price ahead of integration.

The first thing to watch is the 21 July 2026 H1 print — specifically, underlying risk costs ex-Knab/Barclays mix change, standalone CIR, and the explicit CET1 walk to ~17%.