For information only. Not investment advice, a recommendation, an offer or a solicitation, and without regard to any recipient’s objectives or circumstances. Figures come from third-party sources believed reliable but not guaranteed, and market data may be delayed. Past performance is not a reliable indicator of future results. | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Swissquote Group Holding — Deep DiveOverviewRevision log
All figures CHF unless stated. Price CHF 38.38, 24-Sep-2026 [5]. Shares outstanding 148.6m (153.3m issued less 4.64m treasury at 30-Jun-2026) [1][2]. 1. VerdictBuy, 12-month horizon. Conviction: Medium.
We are not making an earnings call. Our difference with the market is a multiple call, and a narrow one. Our 2026–27 EPS sits within 1% of Bloomberg consensus; we are 6% below on 2028. At CHF 38.38 the stock discounts a cost of equity of about 9.0% with 3% terminal growth on our base earnings (9.2% on consensus earnings). Our target uses 8.5%. That 50bp is the call. The case for 8.5%. Crypto is now 4% of revenue against 12% a year ago [2]. 48% of revenue is non-transaction [2]. The company carries roughly CHF 530m of capital above its own 18% threshold, and management has said it will consider distributions when the surplus reaches CHF 500m, "very likely" in 2027 [4]. Two identifiable events inside the 12-month window can compress the risk premium: the FY2026 results on 18-Mar-2027, where the 2028 crypto assumption is to be restated, and a capital return decision. Action. Initiate a half position at spot. Complete the position on whichever comes first:
We would stop adding if the FY2026 results show margin on assets below 77bp with no capital return announced. The staging reflects the fact that the call rests on a 50bp cost-of-equity judgment (§6). Both triggers are ways of being paid for that judgment before taking full size. Where this cuts against the obvious reading. The Street's average target, CHF 43.50 [5], is the same as our probability-weighted value. We are not differentiated on the number. We are differentiated on why. Both extreme notes on the name are unreliable, as §12 explains. Kepler's Buy rests on an understated 2025 base and on the pre-March 2028 plan. CFRA's Sell rests on a 15.3x multiple applied to an earnings window that will be trailing by its own target date. 2. Investment thesis
3. Business and segment economicsReported structure, H1-2026 operating income of CHF 364.9m [1]:
By nature and client [2]:
Concentration and cyclicality.
Yuh.
4. Sector sub-section — digital brokerage bankStructural drivers.
Regulatory overhang.
Industry-specific cycle risk.
5. Accounting comparability and data integrityEarnings and cash quality pass — subject (exceptions only)⚠ Four definitions of 2025 EPS are in play.
⚠ Kepler's 2025 base is understated by ~CHF 12m.
⚠ "The cost curve is bending" is partly a bonus accrual.
⚠ Treasury share purchases are not an accretive buyback.
⚠ The tax rate is creeping up.
Not applicable. Cash-flow and working-capital checks do not apply to a deposit-funded bank. H1 operating cash flow of CHF 512m includes CHF 808m of customer deposit inflows [1], so it says nothing about earnings quality. Comparability audit — peer set
6. ValuationFramework and whyExcess return (RoTE–CoE), with required and excess capital valued separately.
Inputs
Note on the cost of equity (v1.1). Measured inputs do not anchor it.
A 5.2% CAPM rate would value the shares above CHF 100. It is not a usable anchor. The SMI is a defensive, pharma- and staples-heavy index that explains 9% of SQN's weekly variance. Swissquote's risk is idiosyncratic: volumes, crypto, rate mix.
House estimates vs consensus
Consensus: Bloomberg, 24-Sep-2026 [5]. Margin on assets computed on consensus client assets. House 2029–30 are extrapolations: PBT +9% and +8%. Result — base case, target date Sep-2027
Sensitivity — cost of equity × terminal growth (base earnings, CHF/share at Sep-2027)
Other sensitivities (base):
Cross-checks and convergence
Four independent routes land between CHF 40 and 45. The full-peer median is the outlier, and it is explained by mix. The convergence is real, but it is convergence on "moderately undervalued", not on "deeply mispriced". Normalised free-cash-flow yield: not applicable for a bank. In its place:
7. Variant perception
Net: −5.7% on 2028 EPS, ≈ −2% on our NTM anchor, ≈ −CHF 1 on target. We are not making an earnings call. Our difference with Kepler is partly earnings (−3.5% on its blended EPS) and partly base. Our difference with CFRA is almost entirely methodological (§12). Bridge to the Street
8. ScenariosProbabilities: 25 / 50 / 25 (default, no deviation). The evidence does not justify tilting. H1 franchise data argue up; the unresolved 2028 bridge and unmeasured risk-free rate/beta argue down.
Probability-weighted:
Asymmetry. Upside and downside around the base are roughly equal in the 12-month bear/bull (−17% / +37% vs spot). The 3Y bear roughly breaks even on total return. The bear is not a loss scenario because of the multiple. It is a loss scenario because earnings and the multiple fall together. Its implied 14.0x is simply the stock's 5-year average. 9. Catalysts
10. Risks
11. Quantitative scoresTBI H-Score, as of 10-Aug-2026, three days before the H1 results and the −14% share move [13]:
Reading the components as evidence:
12. Sell-side consensus reality checkPanel (Bloomberg, 24-Sep-2026) [5]:
Actions since 1-Aug-2026, classified by type [5][11][12]:
Last 30 days strictly: two reiterations (Oddo, UBS) and one quantitative downgrade. There is no fresh fundamental signal since the post-results wave. Targets moved down, ratings did not. The post-results reaction was almost entirely target cuts inside unchanged ratings. The single upgrade and the single fundamental downgrade offset each other. The Street disagrees on the multiple, not on the earnings.
Kepler.
13. Independent review synthesisReview round 1 (v1.2): what actually ran, via the Hermes orchestrator, 24-Sep-2026
Pass A reconciliation (one reviewer):
Pass B outcomes (facts; they override the draft):
Review round 2 (v1.3): blind Pass B re-run via Hermes, 24-Sep-2026 Blindness: the prompt received by the sub-agent was copied from its log. It matches the question list verbatim, with no file path and no draft [19]. Model: reported by the gateway as Question design (our error, see below): questions 1 and 3 embedded two figures from the draft (CHF 49.8m; CHF 18.4m / 26.3m).
Errors the process found in our own work: (v1.2: none new from Pass A. The accepted points are omissions, not errors.)
Pattern. Every open question in this review concerns price and inputs (cost of equity, 2028 margin, capital return). None concerns the franchise: net new money, client assets and non-crypto revenue lines are all corroborated by primary data. That corroborates the business and warns on the entry point, which is why the recommendation carries an add-below level rather than full size at spot. 14. Verify before publishing
15. Note on methodThree reusable lessons:
Sources
|
Swissquote Group Holding — Deep Dive
Overview
Revision log
| Version | Trigger | Change | Target |
|---|---|---|---|
| v1.0 | Initial | Full deep dive | CHF 44.70 base / 43.50 PW |
| v1.1 | Beta regression and CHF sovereign curve received (verify item 5) | Measured beta does not support the CAPM derivation of the 8.5% CoE. CoE is now stated as a judgment against the market-implied rate. CHF curve steepening added as an NII driver. Spot moved from 38.38 to 38.34; returns in the body are computed at 38.38 (difference <0.2pp) | Unchanged |
| v1.2 | Hermes review round (one Pass A reviewer; Pass B voided) + SNB 24-Sep-2026 release | Second-tranche trigger made catalyst-based as well as price-based. Deposit-repricing risk added; curve-steepening NII claim tempered. SNB hold verified. Review log rewritten (§13) | Unchanged |
| v1.3 | Blind Pass B re-run (Hermes) | Regulatory capital lock-up risk withdrawn (FINMA cat. 3 minimum 12.0%, management threshold unchanged). FX-swap direction closed by mechanism. Kepler-base reconstruction corroborated. Review log extended | Unchanged |
All figures CHF unless stated. Price CHF 38.38, 24-Sep-2026 [5]. Shares outstanding 148.6m (153.3m issued less 4.64m treasury at 30-Jun-2026) [1][2].
1. Verdict
Buy, 12-month horizon. Conviction: Medium.
| 12M (Sep-2027) | 3Y (Sep-2029) | |
|---|---|---|
| Base target | CHF 44.70 (+16.3%; TSR +18.1%) | CHF 54.50 (+41.9%; TSR +13.9% p.a.) |
| Probability-weighted | CHF 43.50 (+13.3%; TSR +15.1%) | CHF 52.70 |
| Bear / Bull | CHF 31.90 / CHF 52.70 | CHF 36.70 / CHF 65.10 |
We are not making an earnings call. Our difference with the market is a multiple call, and a narrow one. Our 2026–27 EPS sits within 1% of Bloomberg consensus; we are 6% below on 2028. At CHF 38.38 the stock discounts a cost of equity of about 9.0% with 3% terminal growth on our base earnings (9.2% on consensus earnings). Our target uses 8.5%. That 50bp is the call.
The case for 8.5%. Crypto is now 4% of revenue against 12% a year ago [2]. 48% of revenue is non-transaction [2]. The company carries roughly CHF 530m of capital above its own 18% threshold, and management has said it will consider distributions when the surplus reaches CHF 500m, "very likely" in 2027 [4]. Two identifiable events inside the 12-month window can compress the risk premium: the FY2026 results on 18-Mar-2027, where the 2028 crypto assumption is to be restated, and a capital return decision.
Action. Initiate a half position at spot. Complete the position on whichever comes first:
- Price trigger: below CHF 35.50, where the market would be discounting a 9.5% cost of equity on our base earnings (−7.5% from spot).
- Catalyst trigger: the FY2026 results (18-Mar-2027) show margin on assets ≥79bp and a restated 2028 crypto assumption, or the board announces a capital return.
We would stop adding if the FY2026 results show margin on assets below 77bp with no capital return announced.
The staging reflects the fact that the call rests on a 50bp cost-of-equity judgment (§6). Both triggers are ways of being paid for that judgment before taking full size.
Where this cuts against the obvious reading. The Street's average target, CHF 43.50 [5], is the same as our probability-weighted value. We are not differentiated on the number. We are differentiated on why. Both extreme notes on the name are unreliable, as §12 explains. Kepler's Buy rests on an understated 2025 base and on the pre-March 2028 plan. CFRA's Sell rests on a 15.3x multiple applied to an earnings window that will be trailing by its own target date.
2. Investment thesis
-
Asset gathering does not depend on crypto, and H1-2026 proved it. Net new money was CHF 5.07bn in H1-2026, the second-best half on record, while crypto volume fell 63.5% YoY [2]. Client assets reached CHF 96.3bn (+19.8% YoY), ahead of the medium-term plan [2]. Swissquote holds 3.0% of a CHF 2.0tn addressable pool of Swiss household financial assets [2]. The growth engine is the multi-asset bank, not crypto turnover.
-
Crypto is rebased, not deferred. Net crypto income was CHF 14.6m in H1-2026 [2]. That figure includes a CHF −5.3m mark on the SQX inventory, and the remaining inventory is only CHF 8.9m [1]. Revised 2026 guidance of ~CHF 730m revenue and ~CHF 365m pre-tax profit (PBT) assumes no recovery; management confirmed it is H1 doubled [2][9]. The downside on this line is now small in absolute terms. The upside is an option the base case does not pay for.
-
The balance sheet funds the rerating. The capital ratio is 25.2% against a FINMA requirement of 11.2% [2]. Retained earnings push the surplus above management's CHF 500m distribution threshold during 2027 [4]. At a 30% payout the dividend yield is only 1.7–1.9%, so the capital decision is the main lever for shareholder return.
-
The earnings trajectory is credible up to 2027. 2028 is less so. Consensus 2027 PBT of CHF 421m needs about 81bp margin on assets and +6% cost growth. We think both are achievable. Consensus 2028 PBT of CHF 496m needs a 54.6% margin. That is above management's own revised target of 53% [3], on revenue CHF 42m below management's CHF 950m. This is the one place we sit below the Street, and it matters for only 40% of our 12-month weight.
3. Business and segment economics
Reported structure, H1-2026 operating income of CHF 364.9m [1]:
| Segment | H1-26 | H1-25 | Share | What it earns on |
|---|---|---|---|---|
| Securities trading | 300.7 | 317.3 | 82% | NII on client cash (cash & fixed income line 118.9), share brokerage 58.8, FX & precious metals 57.9, funds 29.4, structured products 20.2, crypto 12.0 |
| Leveraged forex (eFX) | 46.3 | 41.9 | 13% | Precious metals 24.3 (up from 15.6), FX 11.4 (down from 15.5), CFDs 10.7 |
| Neobanking (Yuh) | 17.8 | – | 5% | FX 6.9, cash 4.8, funds 2.1, crypto 1.8; incremental PBT of CHF −1.0m [2] |
By nature and client [2]:
- 52% transaction-based, 48% non-transaction (interest, custody, management fees).
- Switzerland 56% of revenue and 66% of client assets.
- B2C 72%, B2B/B2B2C 28%.
- No customer exceeds 10% of operating income [1].
Concentration and cyclicality.
- The cash & fixed income line (CHF 118.9m) is effectively the NII engine. At 33% of revenue it is the largest single exposure.
- NII drivers, H1-26 vs H1-25 [1]:
- Interest on investment securities rose to 51.6m from 39.0m. The CHF 5.3bn portfolio yields about 200bp with 2.8 years' duration [2].
- Lombard loan income rose to 25.8m from 19.9m.
- Central bank balances fell to 2.1m from 7.0m.
- FX-swap income fell to 18.4m from 26.3m.
- Deposit costs fell to 7.4m from 10.4m.
Yuh.
- 423k accounts and CHF 4.0bn of client assets [2].
- Carried with CHF 143.7m of acquisition intangibles, of which goodwill is CHF 154.9m group-wide [1].
- No material profit contribution yet [1]. We ascribe no separate value. A sum-of-the-parts would add nothing at this scale.
4. Sector sub-section — digital brokerage bank
Structural drivers.
- Growth in retail self-directed investing in Switzerland and Europe. Europe contributed CHF 1.75bn of H1 net new money, 74% of it from Benelux, France and Germany [2].
- Cash held by clients: 15% of client assets. Its currency mix sets NII. Client deposits are 64% CHF (SARON −0.05%), 16% EUR (€STR up to 2.18%), 15% USD (SOFR 3.65%) [2].
- The CHF curve is steepening (24-Sep-2026 [16]):
- 2Y ~0.35%, 3Y ~0.44%, 5Y ~0.53%, 10Y ~0.61%, against a 0% policy rate.
- Since 31-Dec-2025 the 3–5Y sector is up ~45–50bp and the 10Y ~33bp. The 3M bill is still around −0.03%.
- With a 2.8-year bond book, reinvestment yields on the CHF portion rise. This supports the NII upside flagged in §7.
- The cost is FVOCI marks: H1-26 already carried CHF −10.7m of unrealised losses through OCI [1].
- v1.2 caveat. On 24-Sep-2026 the SNB held its policy rate at 0%. Its conditional inflation forecast assumes 0% over the whole horizon to 2028 [18]. The curve's pricing of tightening is therefore the market's view, not the SNB's.
- Rising CHF rates would also raise pressure to pay interest on client savings balances (see risk 8). The net NII effect of a steeper curve is positive only while deposit repricing lags asset repricing.
- The same SNB release notes that key rates were raised in both the euro area and the US over the quarter [18]. That supports non-CHF NII and the rate differential behind FX-swap income (income rises with the USD/EUR–CHF differential: this follows from the covered-interest-parity mechanism and is not a company disclosure, §14 item 6).
- Product broadening into structured products (+21% issuance), securities lending (H1-26 revenue equal to all of 2025), and online silver trading [2][4].
Regulatory overhang.
- FINMA category 3. The threshold is CHF 17bn of total assets; the balance sheet was CHF 16.9bn at 30-Jun-2026 [2]. Reclassification is expected in H2-2026, and management says most readiness costs are already in the base [8]. It raises the regulatory minimum capital ratio from 11.2% to 12.0%; management's 18% distribution threshold is unchanged [19][9]. That is immaterial against a 25.2% ratio.
- MiCA. Swissquote Bank Europe received a CASP licence from Luxembourg's CSSF [2]. This helps EU crypto distribution but does not change the H1 volume picture.
Industry-specific cycle risk.
- Transaction revenue is pro-cyclical with volatility. Crypto is the most violent sub-cycle: volumes fell 63.5% YoY in H1.
- NII is now driven more by balance-sheet growth and non-CHF rates than by SNB policy. SARON was flat at −0.05% through H1 [2].
- The combination that hurts is low volatility plus falling USD/EUR rates. That is the bear case in §8.
5. Accounting comparability and data integrity
Earnings and cash quality pass — subject (exceptions only)
⚠ Four definitions of 2025 EPS are in play.
| Definition | 2025 EPS | Trailing P/E at spot |
|---|---|---|
| Reported [1] | 2.45 | 15.7x |
| Company-clean (PBT 370, H1-26 presentation p.24 [2]) | 2.12 | 18.1x |
| Kepler "adjusted" [8] | 2.04 | 18.8x |
| S&P "normalised" in CFRA [9] | 1.48 | 25.9x |
- Locked: company-clean for the 2025 base (PBT 370.3m = 420.2 reported − 49.8 net one-off) and Bloomberg adjusted consensus for forecasts.
- The S&P normalisation is unusable: its net income of CHF 224m does not reconcile to any disclosed item.
⚠ Kepler's 2025 base is understated by ~CHF 12m.
- Kepler removes a CHF 62m "exceptional" and carries 2025 opex at 365m against the company's 353.0m [8][3].
- The company line "net gain from remeasurement of joint venture and others" nets to CHF 49.4m in H2-25, plus 0.46m of joint-venture result in H1-25 [2].
- Kepler therefore strips the gross gain but leaves the offsetting one-off costs inside its base. The blind fact check (v1.3) puts the gross gain at ~CHF 61.9m and the offsetting one-offs at ~CHF 12.5m [19]. That is consistent with Kepler's CHF 12.0m opex gap. The franc-level figures and the annual-report note reference are not verified (§14 item 14).
- Consequence: Kepler presents 2026E PBT of 365m as +1.7% YoY. On the company's clean base it is −1.4%.
⚠ "The cost curve is bending" is partly a bonus accrual.
- Payroll rose only +1.4% YoY despite average FTE up 16.3% [2]. The H1 cash bonus accrual fell to CHF 7.4m from 10.1m [1].
- Excluding that effect, total opex grew +6.1% YoY, not +4.6%.
- The own-cost capitalisation CFRA flags (CHF 19.6m vs 14.3m [1]) is not a growing P&L deferral: see the box in §10.
⚠ Treasury share purchases are not an accretive buyback.
- CHF 56.9m of treasury shares were bought in H1-26, against 28.1m in H1-25 [1]. Treasury holdings reached 4.64m shares (3.03%) [2].
- Management states the shares cover employee options and will be reissued on exercise [4].
- We model a flat 148.6m share count. Bloomberg's market cap (CHF 5.88bn) uses issued shares. On outstanding shares it is CHF 5.70bn, and P/TBV is 4.79x rather than Bloomberg's 4.94x [5].
⚠ The tax rate is creeping up.
- H1-26 effective rate: 16.0% [1]. Consensus implies 15.6% for 2026, 14.7% for 2027 and 14.1% for 2028 [5].
- We use 15.5% throughout. On 2028 that is worth −1.6% of EPS versus consensus.
Not applicable. Cash-flow and working-capital checks do not apply to a deposit-funded bank. H1 operating cash flow of CHF 512m includes CHF 808m of customer deposit inflows [1], so it says nothing about earnings quality.
Comparability audit — peer set
- Fiscal years. Seven of nine peers have December year-ends. IG Group (May) and AJ Bell (September) are calendarized: CY2026 maps to FY May-27 and FY Sep-27 respectively [6].
- Accounting regime. IFRS for all except Interactive Brokers (US GAAP). Leases are immaterial for P/E and P/B.
- Metric definitions.
- One source, Bloomberg adjusted consensus, for every name.
- The peer growth columns were pulled in USD (
currency=USDin the query [6]). That distorts every non-USD name. SQN's CY26 revenue shows −0.1% in USD against +1.9% in CHF on the company-clean base. We use the multiples and discard the growth columns. - Bloomberg's own SQN YoY fields (revenue +3.4%, PBT −5.5%, EPS −6.9% for FY26 [5]) reconcile neither to reported nor to clean 2025. The implied revenue base is 712.7m against 723.3m reported. We recompute growth ourselves.
- Peer earnings-quality pass: COVERAGE GAP. Peer filings were not reviewed in this round. Peer multiples rest on Bloomberg consensus alone. Cost: the platform-peer anchor in §6 is a cross-check, not a primary input.
6. Valuation
Framework and why
Excess return (RoTE–CoE), with required and excess capital valued separately.
- Swissquote is a licensed bank. Its value is the return it earns on the equity regulation and management policy require, plus the surplus.
- Applying one multiple to total earnings would capitalise the thin yield on CHF 530m of surplus capital at 16x. That undervalues the surplus relative to its distribution value.
- Why no DCF. Operating cash flow is dominated by deposit flows (see §5), so projecting free cash flow for a bank is meaningless.
- Why no SOP. Yuh contributes no material profit [1], and eForex is 13% of revenue at a group-level cost structure that is not disclosed by segment.
Inputs
| Input | Base | Basis |
|---|---|---|
| Cost of equity | 8.5% | Judgment, not CAPM (see note below). The market-implied rate on our base earnings is 9.0% (9.2% on consensus); we take 50bp off for reduced crypto dependence and capital-return visibility |
| Terminal growth | 3.0% | Swiss household financial assets grow ~3% p.a. [9]; Swissquote gains share on top |
| Capital threshold | 18% of RWA | Management's own distribution threshold [15]; FINMA minimum is 11.2% [2] |
| RWA growth | 12% p.a. | Balance sheet +17.3% YoY [2]; RWA density stable |
| Return on excess capital | 1.3% after tax | HQLA / short-bond yield mix |
| Excess capital valuation | 1.0x | Distribution guided for 2027 [4]; sensitivity at 0.8x and 0x below |
| Roll-forward weight | 60% FY2027 / 40% FY2028 | Next-twelve-month (NTM) window at the Sep-2027 horizon. A strict calendar weight would be 25/75: see sensitivity |
Note on the cost of equity (v1.1). Measured inputs do not anchor it.
| Input | Value |
|---|---|
| Swiss 10Y spot [16] | ~0.61% |
| Beta vs SMI, weekly, 2Y (24-Sep-2024 to 24-Sep-2026, 103 obs.) [17] | Raw 0.70, adjusted 0.80 |
| Standard error of beta | 0.225 (95% range ~0.25–1.15) |
| R² | 0.087 |
| Residual volatility | ~4.6% weekly, ~33% annualised |
| CAPM on these inputs | 0.61% + 0.80 × 5.75% ≈ 5.2% |
A 5.2% CAPM rate would value the shares above CHF 100. It is not a usable anchor. The SMI is a defensive, pharma- and staples-heavy index that explains 9% of SQN's weekly variance. Swissquote's risk is idiosyncratic: volumes, crypto, rate mix.
- The market charges ~3.8pp over CAPM (9.0% implied). Our 8.5% charges ~3.3pp.
- v1.0 presented 8.5% as "normalised risk-free 1.0% + beta 1.3 × ERP". The 1.3 beta was assumed, not measured, and the measurement does not support it. The number stands, but its basis is judgment against the market-implied rate. It should be read as such.
- A regression against a more relevant benchmark (Stoxx Europe 600 Financial Services, or a European brokers basket; 5Y monthly) is the next check. It is not a precondition for the view.
House estimates vs consensus
| CHF m | 2026E House | 2026E Cons. | 2027E House | 2027E Cons. | 2028E House | 2028E Cons. |
|---|---|---|---|---|---|---|
| Net revenues | 732 | 736.9 | 810 | 809.8 | 888 | 908.2 |
| PBT | 365 | 369.8 | 421 | 421.1 | 469 | 495.8 |
| PBT margin | 49.9% | 50.2% | 52.0% | 52.0% | 52.8% | 54.6% |
| Net income | 308 | 312.1 | 356 | 359.3 | 396 | 426.1 |
| EPS | 2.07 | 2.09 | 2.39 | 2.40 | 2.67 | 2.83 |
| Margin on avg client assets | 79bp | 79bp | 81bp | 81bp | 82bp | 84bp |
Consensus: Bloomberg, 24-Sep-2026 [5]. Margin on assets computed on consensus client assets. House 2029–30 are extrapolations: PBT +9% and +8%.
Result — base case, target date Sep-2027
| Item | Value |
|---|---|
| NTM net income at horizon (60/40) | CHF 372m |
| Required tangible equity (18% × RWA ~5.4bn) | CHF 980m |
| Excess tangible equity | CHF 533m |
| RoTE on required capital | 37.2% |
| Justified P/E on required earnings | 16.7x |
| Value per share | CHF 44.65 |
| Implied P/E on NTM EPS | 17.8x (+0.7 SD vs 5Y average of 14.6x [7]) |
Sensitivity — cost of equity × terminal growth (base earnings, CHF/share at Sep-2027)
| CoE \ g | 2.0% | 2.5% | 3.0% | 3.5% |
|---|---|---|---|---|
| 7.5% | 45.9 | 49.4 | 53.8 | 59.2 |
| 8.0% | 42.3 | 45.3 | 48.8 (Kepler 48) | 53.0 |
| 8.5% | 39.3 | 41.8 | 44.7 | 48.1 (Kepler) |
| 9.0% | 36.8 | 38.8 | 41.2 = market today | 44.1 |
| 9.5% | 34.6 | 36.3 | 38.3 | 40.7 |
- Market today. Spot rolled forward at the cost of equity less the dividend is ~CHF 41.2. That is the 9.0% / 3.0% cell: the market prices Swissquote at its own historical average risk premium.
- Street average (CHF 43.50). Sits between 8.5%/2.5% and 8.5%/3.0%.
- Kepler (CHF 48). Needs 8.0%/3.0%.
- CFRA (CHF 33). Lies off the grid: it needs CoE ≥ 9.5% and g ≤ 2%.
Other sensitivities (base):
- Roll-forward weight: 100% FY27 → 42.7; 60/40 → 44.7; 25/75 → 46.4.
- Excess capital valued at 0.8x → 43.9; at 0x → 41.1.
Cross-checks and convergence
| Route | Value | Note |
|---|---|---|
| Excess return, capital split (primary) | 44.7 | |
| Simple justified P/E on total tangible equity (RoTE 24.1%, CoE 8.5%, g 3%) × NTM EPS 2.50 | 39.9 | Capitalises the surplus at a low yield, so structurally lower |
| Platform peers CY27 P/E (Avanza 18.1x, Nordnet 19.9x, AJ Bell 16.7x; average 18.2x) × consensus 2027 EPS | 43.8 | Closest business models: savings/custody platforms with NII |
| Full-peer CY27 median (14.55x) × consensus 2027 EPS | 34.9 | Includes CFD brokers at 9–12x (IG, Plus500, XTB), a structurally lower-multiple model |
| Street average target | 43.5 | [5] |
Four independent routes land between CHF 40 and 45. The full-peer median is the outlier, and it is explained by mix. The convergence is real, but it is convergence on "moderately undervalued", not on "deeply mispriced".
Normalised free-cash-flow yield: not applicable for a bank. In its place:
- NTM earnings yield at spot: 6.0% (NTM EPS 2.32 / 38.38).
- Distributed as a 1.9% dividend, with the balance accumulating as surplus capital.
7. Variant perception
| Item | Our view vs consensus | Size | Direction | Grounded in |
|---|---|---|---|---|
| 2026–27 EPS | In line | ±1% | — | Guidance is H1 doubled [2]. 2027 needs 81bp margin on assets and +6% costs, both plausible |
| 2028 PBT margin | 52.8% vs 54.6% | −2.9% PBT if at management's own 53% | Below | Management cut the 2028 margin target from 55% to 53% when consolidating Yuh [3]. Consensus (and Kepler at 54.9% [8]) still model the old plan |
| 2028 margin on assets | 82bp vs 84bp | −2% revenue | Below | H1-26 was 79bp; the 2026 guide is ~78bp [2]; management's plan assumes 90bp [9] |
| Tax rate | 15.5% vs 14.1% (2028) | −1.6% EPS | Below | H1-26 effective rate 16.0% [1] |
| FY26 NII | Upside to consensus 216.4m | +10–15m | Above | H1 alone was 115.9m and management guides FY26 above FY25's 217.6m [4]. Consensus is internally inconsistent: verify item |
Net: −5.7% on 2028 EPS, ≈ −2% on our NTM anchor, ≈ −CHF 1 on target. We are not making an earnings call. Our difference with Kepler is partly earnings (−3.5% on its blended EPS) and partly base. Our difference with CFRA is almost entirely methodological (§12).
Bridge to the Street
| Target | EPS basis | Implied P/E | vs 5Y avg (SD = 4.5x) | |
|---|---|---|---|---|
| House base | 44.7 | NTM 60/40 FY27/28 = 2.50 | 17.8x | +0.7 SD |
| Street average | 43.5 | Consensus NTM 60/40 = 2.57 | 16.9x | +0.5 SD |
| Kepler (Buy) | 48.0 | 40/60 FY27/28 = 2.65 | 18.0x | +0.8 SD |
| CFRA (Sell) | 33.0 | 60/40 FY26/27 = 2.16 | 15.3x stated; 12.9x on a horizon-consistent window | −0.4 SD |
| Deutsche Bank (Buy) | 59.0 | Consensus NTM 2.57 (DB EPS: COVERAGE GAP) | ~23x | +1.9 SD: a sentiment statement |
| Spot | 38.38 | NTM now = 2.32 | 16.5x | +0.42 SD [7] |
8. Scenarios
Probabilities: 25 / 50 / 25 (default, no deviation). The evidence does not justify tilting. H1 franchise data argue up; the unresolved 2028 bridge and unmeasured risk-free rate/beta argue down.
| Bear (25%) | Base (50%) | Bull (25%) | |
|---|---|---|---|
| Narrative | Crypto stays dead; margin on assets drifts to ~75bp; USD/EUR rates fall and compress FX-swap and non-CHF NII; costs keep rising ~9% | Guidance delivered; margin on assets 81–82bp; costs +6–8%; surplus returned from 2027 | Crypto recovers to ~half of 2025 (~CHF 45m); margin on assets 85bp+; AI productivity visible in 2027 |
| CoE / g | 9.75% / 2.5% | 8.5% / 3.0% | 8.25% / 3.25% |
| 12M (Sep-2027) | |||
| Price | CHF 31.90 | CHF 44.70 | CHF 52.70 |
| Price return / TSR | −16.8% / −15.1% | +16.3% / +18.1% | +37.4% / +39.1% |
| EPS FY27 / FY28 | 2.22 / 2.36 | 2.39 / 2.67 | 2.59 / 2.90 |
| PBT margin FY27 / FY28 | 50.0% / 50.6% | 52.0% / 52.8% | 52.9% / 53.1% |
| Implied NTM P/E | 14.0x (5Y average) | 17.8x | 19.4x |
| 3Y (Sep-2029) | |||
| Price | CHF 36.70 | CHF 54.50 | CHF 65.10 |
| Price return / TSR p.a. | −4.4% / +0.3% | +41.9% / +13.9% | +69.5% / +20.6% |
| EPS FY29 (house extrapolation) | 2.50 | 2.91 | 3.18 |
| PBT margin FY29 | ~51% | ~53.5% | ~54% |
| Implied NTM P/E | 14.3x | 18.2x | 19.8x |
Probability-weighted:
- 12M: CHF 43.50 (+13.3%; TSR +15.1%).
- 3Y: CHF 52.70.
Asymmetry. Upside and downside around the base are roughly equal in the 12-month bear/bull (−17% / +37% vs spot). The 3Y bear roughly breaks even on total return.
The bear is not a loss scenario because of the multiple. It is a loss scenario because earnings and the multiple fall together. Its implied 14.0x is simply the stock's 5-year average.
9. Catalysts
| Catalyst | Type | Horizon | Direction | Note |
|---|---|---|---|---|
| FINMA category 3 reclassification | Micro | H2-2026 | Neutral / – | Costs largely in the base [8]; confirm capital requirement |
| Preliminary FY2026 figures | Micro | ~mid-Jan 2027 | ± | Precedent: preliminaries issued 16-Jan-2026 [14] |
| FY2026 results + restated 2028 crypto assumption | Micro | 18-Mar-2027 [2] | ± key | Management promised to update the 2028 crypto share of revenue [4] |
| Capital surplus reaches CHF 500m → board decision on buyback | Micro | 2027 | + | "Very likely" a buyback, per CFO [4] |
| AGM (dividend) | Micro | 12-May-2027 [2] | + | |
| Chatbot phase 2; AI-assisted development at full adoption | Micro | Dec-2026 / 2027 [2] | + | 25–30% shorter development cycles in early telemetry [2] |
| Crypto market volumes / volatility | Macro | Continuous | ± | Each CHF 10m of crypto revenue ≈ +2.7% PBT at >50% incremental margin |
| USD/EUR policy rates vs CHF | Macro | Continuous | ± | Non-CHF deposits are 31% of client cash [2]; FX-swap income sensitivity to be verified |
| SNB policy / CHF curve | Macro | Next assessment 10-Dec-2026 | Low; + if non-CHF rates stay higher | Held at 0% on 24-Sep-2026; the SNB's forecast assumes 0% to 2028; ECB and Fed hiked during the quarter [18]. The 2Y at ~0.35% prices some CHF tightening the SNB does not signal [16] |
10. Risks
- Transaction revenue shock. 52% of revenue is transaction-based [2]. A volatility trough in equities, FX and crypto simultaneously is the core bear case.
- Margin on assets keeps sliding. It went from 98bp in 2024 to 88bp in 2025 to ~78bp guided for 2026 [2]. Asset growth offsets a stable margin, not a falling one.
- Non-CHF rate cuts. USD and EUR deposits plus FX-swap income support NII. FX-swap income already fell 30% YoY [1].
- Cost discipline fails. Average FTE +16.3% YoY [2]; AI programme ~CHF 15m in 2026 and ~CHF 10m p.a. thereafter [9]; new HQ depreciation starts on delivery (CHF 59.8m under construction [1]).
- Capital not returned. If the board retains the surplus, our capital-split value falls towards CHF 41 (excess valued at 0x). A regulatory lock-up is unlikely: FINMA category 3 lifts the minimum only to 12.0% and leaves management's 18% threshold unchanged [19]. The residual risk is board discretion. Management calls a buyback "very likely" once the surplus reaches CHF 500m, but it is not committed [4]. If the decision slips past the 12-month window, the rerating catalyst slips with it even if fair value is intact.
- Yuh goodwill. CHF 154.9m group goodwill [1]; Yuh must keep growing to avoid impairment.
- Governance. Co-founders hold ~22% (CEO 11.1%, co-founder 10.4%) [9]; a chairman transition took place in 2026 [3]. Low risk, noted.
- Deposit repricing. Client deposits cost only CHF 7.4m in H1-26, down from 10.4m [1], because trading accounts pay no interest. If CHF or EUR rates rise, savings-account remuneration and competitive pressure from neobanks and Swiss banks would raise that cost. This is the other side of the curve-steepening tailwind in §4, and it is not modelled separately in our NII.
Boxed: rising capitalised development costs are not rising earnings flattery
CFRA reads the rise in capitalised own costs (CHF 14.3m → 19.6m) as "shifting staff cost off the payroll line" [9]. A reader will assume profits are increasingly deferred. They are not. What matters to the P&L is capitalisation net of amortisation of proprietary software [1]:
Proprietary software H1-2025 H1-2026 Additions (capitalised) 21.0 25.9 Amortisation (16.3) (21.1) Net P&L deferral 4.7 4.8 The deferral is flat YoY. The cost has moved from the payroll line to the depreciation line, which is presentation, not flattery. The real flattery in H1-26 is the CHF 2.7m cut in bonus accrual (§5).
A related point: CFRA says management's attribution of the depreciation increase to Yuh "does not reconcile". Primary data show Yuh's incremental depreciation was CHF 3.2m of the CHF 6.7m increase (acquisition amortisation 1.2m plus Yuh's own systems) [1][2]. Management's explanation is incomplete, not wrong.
11. Quantitative scores
TBI H-Score, as of 10-Aug-2026, three days before the H1 results and the −14% share move [13]:
| Multi | Quality | Growth | Value | Momentum | Risk |
|---|---|---|---|---|---|
| 76 | 86 | 75 | 60 | 48 | 32 |
Reading the components as evidence:
- Quality 86 and Risk 32 are discordant, and the discordance is the thesis. A high-return, well-capitalised franchise (RoTE ~26% clean, 25% capital ratio) with a volatile, sentiment-driven share price (5Y forward P/E range 7.2x–25.9x [7]). The score says what this report says: the business is better than the stock's behaviour.
- Value 60 predates the de-rating. It is likely higher now; the Value score should be refreshed.
- Momentum 48 is stale and will be lower after the results move. Do not use it as a timing signal until refreshed.
12. Sell-side consensus reality check
Panel (Bloomberg, 24-Sep-2026) [5]:
- 14 analysts: Buy 7 / Hold 6 / Sell 1; rating 3.79/5.
- Average target CHF 43.50 (+13.3%), median CHF 44.26, range 33–59.
- Other panels differ. The S&P panel in CFRA shows 11 analysts, 0 Sell [9]. The company-collated panel of 13 brokers was last updated 5-Aug-2026, before the results, so it is stale [10].
Actions since 1-Aug-2026, classified by type [5][11][12]:
| Type | Count | Names |
|---|---|---|
| Upgrade | 1 | Kepler Hold → Buy, target 51 → 48 (14-Aug) |
| Downgrade | 3 | CFRA Hold → Sell, 42 → 33 (17-Aug, shown in Bloomberg as a 19-Aug reiteration: misclassified); Research Partners → Hold 38.60 (14-Aug); Sadif → Hold 34.21 (3-Sep, quantitative house, low signal) |
| Reiteration with target cut | 5 | UBS Neutral 42.50 → 39; Goldman Buy 50 → 48; Berenberg Hold 43.50 → 43; Deutsche Bank Buy 62 → 59 (3-Aug, pre-results); Vontobel Hold 40 |
| Reiteration | 4 | Oddo Outperform 55.50 (24-Sep); BNP Outperform 44; Jefferies Hold 36; AlphaValue Add 48.20 |
Last 30 days strictly: two reiterations (Oddo, UBS) and one quantitative downgrade. There is no fresh fundamental signal since the post-results wave.
Targets moved down, ratings did not. The post-results reaction was almost entirely target cuts inside unchanged ratings. The single upgrade and the single fundamental downgrade offset each other.
The Street disagrees on the multiple, not on the earnings.
- The 2027 EPS range across contributors is 2.27–2.89 (+27%) [9].
- The target range is 33–59 (+79%).
- Kepler and CFRA have identical 2026 EPS (2.07) and 2027 EPS within 6%. Their targets are CHF 15 apart.
Boxed: the CFRA Sell is largely a horizon artefact
CFRA applies 15.3x (its 5Y-average multiple) to a 60/40 blend of 2026/2027 EPS, i.e. CHF 2.16, for a target dated August 2027 [9]. By that date, 2026 is history and the NTM window is ~40% FY27 / 60% FY28.
On a horizon-consistent window, using CFRA's own 2027 EPS grown at consensus +18%, the same 15.3x gives CHF 39. That is the current share price, not CHF 33.
Half of CFRA's 15% downside comes from applying a forward multiple to trailing-at-target earnings: "same forward horizon on both legs" violated. CFRA's substantive points still stand: the undisclosed 2028 crypto assumption, margin on assets at 79bp versus the 90bp plan, and the bonus accrual. We adopt them in §7 and §5.
Kepler.
- Its note is issuer-sponsored research ("as a service to the issuer" [8]).
- Its 2025 base is understated (§5), and its 2028 model reproduces the pre-March plan (§7).
- Its CHF 48 target equals 18x on a correctly weighted 40/60 FY27/28 blend. The method is internally consistent; the inputs are optimistic.
- Its rating-history table shows a "Buy" on 4-Aug-2026 that contemporaneous press does not corroborate. The Hold → Buy upgrade was on 14-Aug [11]. Likely a table error.
13. Independent review synthesis
Review round 1 (v1.2): what actually ran, via the Hermes orchestrator, 24-Sep-2026
- Pass A: one reviewer, not two.
- The reviewer was reported as Claude Sonnet via the 1minai gateway. The delegation log records "Model: ?", so the model is unverified.
- A second reviewer (reported as GPT-5) was announced but no output was returned.
- Pass A is therefore partially performed: one reviewer, possibly from the same model family as the author.
- Pass B: void.
- The answers were not produced by a blind sub-agent. The orchestrator, which had read the draft, wrote them itself in a script annotated "simulated via direct factual answers from primary docs & public knowledge".
- They reproduce the draft's own constructions (e.g. "49.4m H2 + 0.46m JV", and "H1-2026 presentation p.24" cited as a source for a gross annual-report figure).
- They introduce one unsourced figure: a gross Yuh gain of CHF 61.7m.
- Per protocol, a reviewer that reproduces the draft's numbers is not corroboration. None of the seven answers is logged as confirmation. The "CONFIRMS" classifications in the orchestrator's contradiction table are rejected wholesale.
- The earlier directed self-check stands, still labelled non-independent: H1-26 interim report [1], H1-26 presentation [2], call transcript excerpts [4], and the SNB release [18] added in v1.2.
Pass A reconciliation (one reviewer):
| # | Reviewer point | Decision | Reasoning / change |
|---|---|---|---|
| A1 | The 50bp CoE discount is a judgment wedge; ~CHF 4–5 per 50bp | Accepted in substance, restates draft | Already stated in §6 (grid: 8.5% → 9.0% moves the base from 44.7 to 41.2, i.e. CHF 3.5, not 4–5). No change beyond the staging in §1 |
| A2 | 25/50/25 is a symmetric placeholder, not conviction-weighted | Accepted, no change | The draft states it is the default with no deviation because the evidence does not justify tilting. That is the correct use of the default, not a flaw |
| A3 | The variant perception is grounded | Not logged as corroboration | It restates the draft, and misstates the old plan as "55% / CHF 950m"; the old plan was CHF 900m / 55% [3] |
| A4 | Half position plus add only on a −7.5% drop sits awkwardly with a Buy | Accepted (partial) | Valid: if the price never dips, a Buy stays half-sized indefinitely. §1 now adds a catalyst trigger. The price trigger is kept |
| A5a | Deposit repricing risk undersold | Accepted. Works against our v1.1 addition: it tempers the curve-steepening NII tailwind we added in v1.1 | New risk 8; §4 caveat |
| A5b | FINMA category 3 could lock up the surplus and delay the rerating | Accepted (partial) | Already partly covered (risk 5, verify item 9). Risk 5 now names the mechanism and the timing consequence. v1.3: partly withdrawn. Pass B shows the regulatory step-up is only 80bp and management's threshold is unchanged. The residual risk is board discretion, not regulation. This revision works in favour of the thesis |
Pass B outcomes (facts; they override the draft):
| Item | Result | Effect |
|---|---|---|
| Clean 2025 PBT | 370.3m; one-off net 49.8m (49.4 in H2 + 0.46 JV in H1) [2] | Confirms CFRA's correction; Kepler's base understated |
| Bonus accrual H1-26 | 7.4m vs 10.1m [1] | Confirms CFRA; weakens Kepler's operating-leverage pillar |
| Own-cost capitalisation | 19.6m vs 14.3m confirmed, but net deferral is flat (4.8 vs 4.7) [1] | Partly rejects CFRA's framing (boxed §10) |
| Depreciation split | IT systems +5.4m; acquisition amortisation +1.2m; Yuh incremental total 3.2m [1][2] | CFRA's figures correct; management's explanation incomplete rather than wrong |
| Treasury purchases | CHF 56.9m, to cover employee options [1][4] | Works against our thesis: withdraws a buyback-accretion argument we would otherwise have made. Share count held flat |
| 2028 targets | Revenue CHF 950m, PBT margin 53% (revised Mar-2026) [3] | Establishes the consensus-margin anomaly in §7 |
Review round 2 (v1.3): blind Pass B re-run via Hermes, 24-Sep-2026
Blindness: the prompt received by the sub-agent was copied from its log. It matches the question list verbatim, with no file path and no draft [19].
Model: reported by the gateway as openclaw-default. This is an alias; the underlying model is not identified, and the delegation header again shows "Model: ?". It therefore cannot be confirmed as different from the Pass A reviewer. The pass is blind but not demonstrably cross-model.
Question design (our error, see below): questions 1 and 3 embedded two figures from the draft (CHF 49.8m; CHF 18.4m / 26.3m).
| # | Answer (short) | Decision | Effect on draft |
|---|---|---|---|
| B1 | Gross Yuh gain CHF 61,866,800; offsets CHF 12,484,300; net CHF 49,838,100 | Accepted in order of magnitude only | The answer does not add up internally: 61.87 − 12.48 = 49.38, not 49.84. The 0.46m difference is the H1-25 JV result, which is not an offsetting item. The target figure of 49.8m was in our question. Corroborates ~62m gross / ~12m offsets; franc-level figures not relied on (§5) |
| B2 | 2028: revenue CHF 950m (from 900), PBT CHF 500m, margin 53% (from 55%) | Accepted | Paraphrased, not verbatim as asked. Consistent with [3]. No change |
| B3 | FX-swap income captures the USD/EUR–CHF rate differential | Accepted as mechanism, not as disclosure | The agent inferred rather than answering "not disclosed". The direction follows from covered interest parity (swapping USD/EUR liquidity into CHF earns the forward points). Item 6 closed on that basis. The ECB and Fed hikes [18] are therefore supportive (§4) |
| B4 | Call: "most likely… higher interest income in 2026 compared to 2025" | Accepted | Corroborates [4]. The consensus-side inconsistency (item 7) remains a vendor question |
| B5 | Threshold 18%; surplus ~CHF 300m at FY25; ~CHF 500m trigger; category 3 lifts the minimum to 12.0%, management threshold unchanged | Accepted: overrides the draft | Risk 5 rewritten; A5b partly withdrawn; item 9 closed |
| B6 | CFRA action: "not disclosed" in wire archives | Accepted as an honest non-answer | CFRA's own report [9] is the primary source for its action (17-Aug, Hold → Sell, target 33). Item 10 closed; the Bloomberg misclassification remains a vendor flag |
| B7 | SNB held at 0%, 24-Sep-2026 | Accepted | Matches our direct verification [18]. No change |
Errors the process found in our own work:
(v1.2: none new from Pass A. The accepted points are omissions, not errors.)
- In the initial chat framing we described Kepler's target as "~19.6x 2027 EPS". Kepler applies 18x to a 40/60 FY27/28 blend. The arithmetic was correct; the attribution of method was wrong.
- We dismissed CFRA's point that the CEO cited CHF 900m for 2028 revenue on the call as "the old target". The official target is indeed CHF 950m [3]. But the CEO's use of the superseded figure on the call is corroborated by call highlights [4], and it is a fair signal of loose messaging on the bridge.
- An interim price of CHF 40.30 (21-Sep, web screen) was used in chat before the Bloomberg pull showed CHF 38.38 on 24-Sep. The screen price is not today's price. All figures here use 38.38.
- (v1.1) The v1.0 cost of equity was presented as CAPM-derived with an assumed beta of 1.3. The measured adjusted beta vs SMI is 0.80, with R² 0.09 [17]. The number is retained, but its derivation was not supported by the data; it is relabelled as a judgment.
- (v1.3) Our Pass B questions 1 and 3 carried draft figures (the CHF 49.8m net one-off; the CHF 18.4m / 26.3m swap income). That breaches the "no context beyond what the question needs" rule. On question 1 it visibly anchored the answer: the agent reverse-engineered the net figure with arithmetic that does not close. Future Pass B prompts should name the line item, not its value.
Pattern. Every open question in this review concerns price and inputs (cost of equity, 2028 margin, capital return). None concerns the franchise: net new money, client assets and non-crypto revenue lines are all corroborated by primary data. That corroborates the business and warns on the entry point, which is why the recommendation carries an add-below level rather than full size at spot.
14. Verify before publishing
| # | Item | Why it matters | Status |
|---|---|---|---|
| 1 | Clean 2025 base | Growth optics for 2026 | Closed [2] |
| 2 | Bonus accrual; capitalisation; depreciation split | Operating-leverage claim | Closed [1][2] |
| 3 | Purpose of treasury purchases | Share count, accretion | Closed: employee options [4] |
| 4 | 2028 revenue / margin targets | Variant perception | Closed (secondary report of the 19-Mar-2026 release [3]; call corroborates) |
| 5 | Spot Swiss 10Y yield and SQN measured beta | The CoE is the call | Closed (v1.1): 10Y ~0.61%; adjusted beta 0.80 vs SMI, R² 0.09. CAPM uninformative; CoE relabelled as judgment [16][17]. Follow-up: beta vs a financials/brokers benchmark |
| 6 | FX-swap income: direction of sensitivity to CHF vs USD/EUR rate differentials | Bear-case mechanism | Closed (v1.3): income rises with the differential. Established by mechanism (covered interest parity); not disclosed by the company [19] |
| 7 | Consensus FY26 NII of 216.4m below FY25's 217.6m despite guidance above | Stale contributors or field definition (cb_is_nii) | Open |
| 8 | Bloomberg YoY growth bases do not reconcile (revenue base 712.7m vs 723.3m) | Growth figures quoted to committee | Open: recomputed in-house meanwhile |
| 9 | FINMA category 3 capital requirement vs management's 18% threshold | Excess capital size | Closed (v1.3): minimum 11.2% → 12.0%, management threshold unchanged [19] |
| 10 | CFRA action date/type (17-Aug downgrade vs Bloomberg 19-Aug reiteration) | Rating-action count | Closed on CFRA's own report [9]; Bloomberg misclassification flagged to vendor |
| 11 | H-Score refresh post-results | Momentum and Value stale | Open |
| 12 | SNB September 2026 decision | Macro catalyst | Closed (v1.2): held at 0%, forecast assumes 0% to 2028 [18]. Verified directly; the Hermes Pass B answer is not relied on |
| 13 | Independent Pass A and blind Pass B on a different model | Process requirement | Partially closed (v1.3): Pass B blind with prompt verified from log; cross-model not demonstrable (gateway alias only). Pass A: one reviewer, model unverified |
| 14 | Yuh gross gain and offsetting items to the franc (Annual Report 2025, business-combination note) | Explains the Kepler base error; not a valuation input | Open, low priority |
15. Note on method
Three reusable lessons:
- Check which earnings window a target multiple is applied to, relative to the target's own date. A forward multiple on earnings that will be trailing at the target date is the forward-vs-trailing error in disguise. Here it accounted for half of a published Sell's downside.
- Read the currency parameter in any Bloomberg growth query. A
currency=USDflag silently turns CHF, SEK, GBP and PLN growth rates into FX-contaminated numbers. - Do not derive a cost of equity from a home-index beta when R² is below ~0.2. A defensive home index (SMI) measures almost none of a transaction-driven stock's risk, and CAPM on it produces an unusable rate. Anchor on the market-implied rate and state the premium you add or remove.
Sources
- Swissquote Group Holding, Half-Year Report 2026: condensed consolidated interim financial statements (IAS 34), reviewed by Ernst & Young, 12-Aug-2026. Period end 30-Jun-2026.
- Swissquote, Press conference — First half-year results 2026, 13-Aug-2026.
- Swissquote FY2025 results release, 19-Mar-2026, as reported by FX News Group (19-Mar-2026). Primary release not retrieved: secondary.
- Swissquote H1-2026 earnings call, 13-Aug-2026: transcript excerpts via stockanalysis.com and Investing.com.
- Bloomberg consensus and market snapshot, SQN SW Equity, 24-Sep-2026 (BQL; user-supplied).
- Bloomberg calendarized peer valuation table, 24-Sep-2026 (BQL; user-supplied). Growth columns in USD.
- Bloomberg SQN SW 12-month blended forward P/E history, 5Y/10Y, 24-Sep-2026 (BQL; user-supplied).
- Kepler Cheuvreux, B. de Leudeville, "Crypto reset, NNM intact – upgrade to Buy", 14-Aug-2026. Issuer-sponsored research.
- CFRA, F. Ibrahim, Swissquote Stock Report, 19-Sep-2026 (includes S&P Global Market Intelligence data).
- Swissquote investor relations, company-collated analyst consensus, last updated 5-Aug-2026.
- cash.ch, analyst roundup on Swissquote, 24-Aug-2026 (Goldman target 48, Kepler upgrade).
- StreetInsider: Berenberg target change 18-Aug-2026; Deutsche Bank target change 4-Aug-2026.
- TBI H-Score screen (UD_TBI_*), Bloomberg, 10-Aug-2026 (user-supplied).
- finews.com / FX News Group, Swissquote preliminary FY2025 figures, 16/17-Jan-2026.
- Investing.com, "Swissquote Group reports 2026 guidance below consensus on growth investments", 19-Mar-2026: capital ratio 25%, CHF 300m above the 18% threshold.
- Bloomberg, CHF Switzerland Sovereign Curve, mid YTM: last vs 01-Jun-2026 and 31-Dec-2025, 24-Sep-2026 15:26 (user-supplied screen).
- Bloomberg, BETA SQN SW Equity vs SMI Index: weekly, 24-Sep-2024 to 24-Sep-2026, 103 observations (user-supplied screen).
- Swiss National Bank, press release "Monetary policy assessment of 24 September 2026" and introductory remarks at the news conference, 24-Sep-2026.
- Blind Pass B sub-agent via the Hermes orchestrator (gateway alias
openclaw-default), 24-Sep-2026. Raw log: /home/dev/.hermes/cache/delegation/live/deleg_dc77f845/task-0.log. Its cited sources (FY2025 release, H1-26 call, Finance Magnates, LiquidityFinder) not independently re-opened except where stated.
Document facts
- H-Score
- 76
- Sector
- Digital Brokerage Bank
- Geography
- CH
- Accounting
- IFRS
- Created
- 2026-09-24