PUBLIC RESEARCH / SDB ZÜRICHZÜRICH · EUROPE/ZURICHSOURCE DATE PER ANALYSIS

SDB / KI · SI · HI

Public information. Multiple perspectives.

We bring public information together and analyse it using AI, SI and HI approaches. Online tools and local analysis complement each other; people review the interpretation.

01

Gather

Collect company reports, investor-relations releases and public market data, with their sources and timestamps.

02

Interpret

Compare facts, assumptions and counterarguments. AI-supported online tools and local analysis help examine the connections.

03

Review

Check findings and calculations, explain their limits and publish the approved record with its sources.

AI, SI and HI are SDB terms for analysis approaches in use or under development. They are not standardised or measured performance levels.

Learn the process ↗
Historical research documentation
SDB / RESEARCH MODEL

A model.
Multiple source versions.

A and B analyse independently. E reviews the frozen findings. C brings them together. Changes, limitations and historical scenarios remain traceable.

HISTORICAL ENTRIES117
METHODOLOGY ITEMS515A 227 / B 192 / E 96
DIRECTION NOT SPECIFIED—
LATEST EDITORIAL VERSION10.10.2026
01 / ROLES & PROCESS

Separate perspectives. Shared review.

A + B → E → C

Role outputs are documented research versions. Completing an entire checklist proves neither model quality nor an advantage in predicting prices.

02 / UPDATES & REFINEMENT

Changes with provenance.

Documented method and source updates sit alongside the historical originals. Later findings do not change any original dated version.

Sources & bindings ↗

Updates are loaded from the local method register.

View method bindings and file boundaries
Complete model register as JSON ↓
03 / DOCUMENTED CALCULATIONS

The formula and its inputs.

Examples from the frozen PKE version of 08.10.2026. These displays are calculation identities or conditional sensitivities; they do not implement a price-reaction algorithm.

PKE case model ↗
PKE / A / EARNINGS BRIDGE

Reconcile EBITDA.

EBITDA = EBIT + D&A
Adj. EBITDA = EBIT + D&A + SBC

In the present PKE-A model: 4,158 + 0,480 + 0,100 = 4,738 million USD adjusted EBITDA. The SBC adjustment belongs to this issuer-aligned model; other companies may use different definitions.

PKE / A / SHARE DENOMINATOR

Calculate EPS.

EPS = Stammaktionärsgewinn
÷ gewichtete verwässerte Aktien

3,759488 million USD ÷ 22,05 million shares ≈ 0,170498 USD in the Q2 preliminary model at the time. Earnings, tax rate and share count were model assumptions; the calculation does not validate these inputs.

PKE / B / CONDITIONAL SENSITIVITY

Separate EPS and the multiple.

R = (EPS_s ÷ EPS_h)
× (PE_s ÷ PE_h)

Documented relative valuation approach from B. EPS and the valuation multiple are assessed against an explicitly assumed hurdle. R is not an opening gap, a probability or a trading profit.

FROZEN PKE MODEL VALUES

A and B side by side.

All method items ↗
Revenue and adjusted EBITDA in million USD / EPS in USD per share / preliminary assumptions, not actual results.
HORIZONMODELREVENUEADJ. EBITDAEPS
04 / HISTORICAL SCENARIOS & DATED VERSIONS

Every version remains visible.

All archived versions can be read individually. The documented direction is the historical statement of that entry. A missing direction remains unresolved and is not interpreted as neutral.

Role filters use the original role version; “A/B/C” is a shared historical entry. New and older revisions remain separate records. Times are displayed in Europe/Vaduz; a date without a time is not assigned an invented timestamp.

Historical research versions with date, role, status, direction and direct access to the original
AS-OF DATECOMPANYSCENARIO / STATUSROLEDIRECTION / STRENGTHORIGINAL

The register does not summarise a “current signal” for each company. No profitability, success rate or price range is inferred from historical labels.

HOW WE WORK / HYPERINTELLIGENCE

Research globally.
Evaluate with frozen inputs.

SDB stands for Swiss Dimensional Broker: a Swiss home, multidimensional analysis and the sharing of information. No trading accounts or execution of client orders.

Working backwards from the event.

We begin with the target metric and time window: for example, a company quarter or the historical question of an opening gap. We then check what information was actually available before the event and which assumptions support the causal chain.

Business structure, CEO/CFO responsibilities, relevant technical leadership, presentation structure and public call participation form part of the context. A title or appearance alone does not provide a reliable direction.

Communication as a source.

We examine publicly available statements, answers, definitions and changes in outlook. Observations and interpretations are documented separately. A person’s inner motives or future decisions cannot be reliably inferred from them.

Company results and calls can occur before, during or after trading. We take the date from the company’s announcement and check the release and call separately.

01 / RESEARCH

Public sources.

Global research, filings, investor relations documents and documented AI assistance. Sources, times and definitions are recorded.

02 / EVALUATION

A fixed data snapshot.

A and B work independently. E reviews frozen findings; C consolidates them. OFFLINE_FROZEN_ONLY limits the permissible inputs for each run.

03 / PUBLICATION

Output you can review.

Approved analyses, changes and retrospective reviews are published publicly. The website, subscription and external market feeds require an online connection.

A published method does not guarantee forecast accuracy or profitability. Dated findings and their limitations are the basis of transparency.

Sources on company disclosures

SEC Form 8-K, Item 2.02 ↗ · SEC Regulation FD ↗ ·

Editorial source review dated 10.10.2026. These sources do not constitute an audit or certification of SDB.