Methodology

How AssetNext compares companies, which perspectives it keeps separate — and what the methodology deliberately does not attempt.

Part 1: The Foundation

The core idea: peer-relative, not absolute

A return on equity of 18% is no information without context. Only in comparison with comparable companies does it become a classification — exceptional, solid, or below average. The same applies to nearly any relevant metric: valuation levels, growth rates, margins, volatility. Numbers gain their meaning through the reference system in which they stand.

AssetNext therefore does not work with absolute thresholds but with peer-relative percentiles. A company is not measured against a universal norm, but against the distribution of comparable companies. The result is a classification on a scale from 0 to 100 — not as a normative judgment, but as a position: where does this company stand within its comparison group?

This decision has consequences. It dispenses with the apparent clarity of fixed categories ("cheap," "expensive," "growth-oriented") that are common in financial communication. Instead, it describes positions within a field. That is linguistically less spectacular but closer to what data can actually tell us.

How peer groups are formed

The quality of any peer-relative statement depends on the quality of the peer group. AssetNext uses exclusively functional peers for the Peer Score — companies with structurally similar business trajectories. The construction of these peer groups follows a multi-layered logic, explained here in its principles.

Functional similarity. Two companies are considered functionally similar when their long-term business trajectories move in similar ways — in the development of their revenues, their margin structure, and their capital efficiency over multiple years. What matters is not what a company nominally does in its industry code, but how it actually behaves economically. This similarity is calculated as a numerical measure ranging from 0 (no structural similarity) to 1 (fully aligned trajectories).

The minimum threshold. Not every candidate is included in a peer group. AssetNext requires a minimum similarity; companies below this threshold do not enter the Peer Score, even if the peer group becomes smaller as a result. A smaller group of truly comparable companies is more meaningful than a larger one diluted by weakly similar candidates.

Geographic opening. A company is first considered in the context of its home universe — a DAX company among German peers, a Nasdaq company among US peers. When the number of sufficiently similar peers at this local level is limited, the search space opens step by step along a defined path into broader universes: from narrow national indices through regional to global comparison groups. Each step is documented and made transparent in the report.

Locality as a soft preference, not a hard rule. This is where a methodological decision lies that most clearly distinguishes AssetNext from classical peer systems. AssetNext treats geographic proximity not as a hard condition, but as a soft preference. Specifically: a candidate from a more distant universe receives a small discount on its similarity. A local candidate is preferred at otherwise equal similarity — but an internationally located candidate that is significantly more similar can still displace a weakly similar local candidate. Classical peer systems working purely geographically or classificatorily often resort to less comparable but closer candidates when coverage in the narrow universe is thin. AssetNext avoids this compromise: comparability takes precedence over proximity.

Similarity-weighted aggregation. Within a peer group, not all peers are equally similar. A peer with very high similarity contributes more to a comparison than one with medium similarity, even if both are qualified. AssetNext weights accordingly: the more similar a peer, the more strongly it enters the score calculation. The result is a peer-relative score that weights the most relevant comparisons most heavily.

Limits of this construction. Even carefully constructed functional peer groups have limits. Structural similarity in historical trajectories does not mean business model identity. For very young, highly idiosyncratic, or data-sparse companies, the functional peer base may remain too thin to form a reliable score. In such cases, AssetNext deliberately omits a score rather than outputting a substitute value constructed from ill-fitting comparison groups. A disclosed gap is more informative than filled-in pseudo-precision.

The four perspectives

Economic reality cannot be reduced to a single dimension. AssetNext therefore separates four perspectives that answer different questions and are not dissolved into one another.

Valuation. The question: what does the market currently demand for this business? The valuation perspective measures how the company is priced in the market — relative to its own long-term valuation level and relative to the valuation level of comparable companies. The core anchor is the price-to-earnings ratio, supplemented by peer-relative classification. Where no meaningful valuation data exists — for example, with negative earnings — a descriptive regime classification is used instead of a number. A distorted P/E is not a statement; transparency about its non-informativeness is.

Quality. The question: how substantial is the underlying business? Quality here refers to structural features of capital usage and profitability: capital returns, operating margin, efficiency of the earnings base. The classification is peer-relative — not "is ROIC X good," but "where does this company stand on the capital-return distribution of its peers." Quality in this sense is a business characterization, not an evaluation.

Stability. The question: how does the company behave under pressure? Stability bundles features of fluctuation characteristics and loss resistance across the available history: volatility, beta to the market, maximum drawdown, and other features of behavior under stress. This is not about prediction, but about describing past reaction patterns in peer comparison.

Growth. The question: how is the business foundation developing? The growth perspective considers the development of revenue and earnings at the business level, not at the price level. Peer-relative classification: where does the company stand in the growth spectrum of its comparison group, and does this development support the other perspectives or stand at odds with them?

Why the separation is not dissolved. A single overall grade would lose exactly what the actual analytical value is: the tensions between the perspectives. A company can rank very high in quality but equally high in valuation — that describes a different situation than high quality at low valuation. Anyone averaging both into a single number swallows the informational content. AssetNext does compute an aggregated Peer Score for sortability. However, the four dimensions remain individually visible, because the constellation of the dimensions is the actual statement, not their average.

How scores are formed

Each dimension is translated into a percentile rank. A value of 80 means: 80% of peers stand weaker on this dimension, 20% stronger. The scale is not linear in the raw values, but in the relative position. This is deliberately chosen — it makes companies comparable across different peer groups without needing absolute thresholds.

The aggregated Peer Score combines the four dimensions into a single number, also on the 0–100 scale. It serves primarily for sortability and quick overview. The weighting of the dimensions is fixed and applies equally to all companies; it is not a judgment about what is more important, but a uniform aggregation rule that remains consistent across all reports.

Handling outliers. In every real data distribution, extreme values occur that can distort percentiles. AssetNext dampens such outliers in scoring so that a single extreme value does not tip the entire percentile rank of a peer group. The displayed raw values remain unaffected. What is dampened is the statistical weighting in the score, not the represented reality.

Joint treatment of share classes. Companies with multiple share classes (such as voting and non-voting) are treated as one company in peer formation and scoring. Otherwise, they would count each other as peers, which would structurally undermine the meaningfulness of a peer group.

Score availability. A score is only output when enough peers and enough available dimensions are present. If the data basis is too thin, no score is reported instead of a number. In this case, foregoing false precision is more informative than a value extrapolated from few data points.

Complementary perspectives

Alongside the four core dimensions, AssetNext captures additional features that are relevant for a complete picture. They are not part of the central score aggregation but are reported in the analyses.

Resilience is an event-based consideration of recovery patterns after price declines, relative to the respective index. What is measured is whether the company recovers faster, slower, or in parallel with the benchmark after stress phases. This is a descriptive characterization of historical behavior, not a forecast.

Market behavior refers to relative price development versus the benchmark index, shown descriptively. Technical indicators such as RSI or MACD are shown where relevant as additional context — they do not change score values and are not interpreted as trading signals.

Development context refers to indexed trajectories of revenue and operating margin across available annual values. The purpose is not extrapolation, but the depiction of historical trajectories — how the business substance has changed over time.

These complementary perspectives are part of the methodology, but not part of the Peer Score. The separation is deliberate: score aggregation works only with robustly peer-comparable metrics. Resilience and market behavior are informative but not aggregable in the same way.

Part 2: The Application Layers

The Peer Score and its complementary perspectives are not ends in themselves. They form the methodological foundation for several analytical layers that AssetNext constructs from this basis. Each layer has its own focus and, where necessary, its own aggregation logic.

Single-company analysis

The most fundamental application layer: structured company reports that place all described perspectives alongside a single company — Peer Score classification, dimensional profiles, resilience, market behavior, development context. No additional methodological layer, but the direct expression of the Peer Score system for a single company.

Structured direct comparison

Two companies are not simply placed side by side but compared along an analytical anatomy. For each comparison, AssetNext identifies the dominant dimension (where is the difference strongest?), a supporting dimension (where does the finding intensify?), and, if applicable, a counter-direction (where does the picture reverse?). This creates a comparison that shows not only numbers but evaluates their constellation.

An important methodological point: both companies are each positioned in their own peer context and then measured against each other. This avoids the classical error of direct comparisons, in which two structurally different companies are measured against an abstract uniform standard.

Portfolio analyses

For a portfolio of existing positions, the methodology extends beyond single-company analysis. AssetNext aggregates the peer-relative profiles of the included companies, weighted by position size, and constructs from this an overall picture of the portfolio itself: its structural signature, its concentration, its inclination toward particular characteristics.

The analysis identifies tensions at the portfolio level — for example, when individual positions stand at odds with the rest of the portfolio's character, when breadth is low, when positions are unusually aligned in certain dimensions. The portfolio is analyzed as an independent object, not merely as the sum of its parts.

Methodologically significant: this layer has its own aggregation and classification logic built on Peer Score data, but with additional steps — portfolio characterization, tension detection, positioning of the portfolio against a methodologically appropriate reference. How individual positions should be weighted remains a question outside the methodology: that is an allocation decision and the user's responsibility.

Market patterns

The Situations layer systematically searches for recurring structural patterns in current market activity. The mechanism operates in several steps: daily, pattern conditions are checked across covered companies — such as constellations between Peer Score profile and price development that are statistically notable or indicate certain structural tensions. Detections are prioritized by selection relevance; the most frequently occurring pattern of a day is identified as the dominant daily pattern.

A particularity of this layer is its temporal dimension: detections are classified by freshness — new (detected today for the first time), recent (detected in the last few days), or continuing (already active for longer). This creates not only a picture of which patterns are visible today, but also an impression of which are newly appearing and which are consolidating.

The output is again descriptive: patterns are made visible, not actions proposed. A detected pattern is an invitation to look more closely, not a buy recommendation.

Market scanner

While single-company analysis, comparison, and portfolio build on known starting points, the market scanner works in the opposite direction. It searches the covered company universe along combinable conditions: Peer Score dimensions, structural tensions between valuation and quality, market-behavior features, membership in particular indices. The user defines which combination of features they are looking for — the market scanner returns the companies that match.

Methodologically, this is not a new layer but the systematic application of the existing peer-relative logic to a different entry point: not from known company to analysis, but from desired structure to the companies that exhibit it. Companies listed in multiple indices appear once, not multiple times — the membership ambiguity is resolved in the background.

What all layers share

Throughout all application layers, the underlying stance remains the same: the analysis describes states and makes patterns visible. It identifies structures, shows tensions, classifies. It does not interpret, does not recommend, does not forecast. Interpretation remains with the user.

Part 3: Limits and Self-Understanding

What the methodology does not attempt

An honest methodology names its limits not only at the margin but makes them visible. AssetNext deliberately refrains from the following:

No buy, sell, or hold recommendations. The analysis classifies; it does not recommend. Good classification is a prerequisite for a good decision — but not its substitute. Decisions require context that a data analysis system does not know: personal life situation, time horizon, risk capacity, tax framework.

No price targets, no trading signals. The methodology is not designed for forecasting. It describes present and past states. Anyone seeking price forecasts is in the wrong place at AssetNext — not because it would be less serious to deliver forecasts, but because the methodology is not built for that.

No allocation advice. Portfolio analyses show patterns and tensions in existing portfolios — they do not replace allocation advice. How much weight individual positions should carry, which risk budgets are appropriate, how a portfolio fits the personal situation — these are questions that presuppose life context a data analysis system does not know.

Limits of data coverage. For smaller companies or niche stocks, the peer base may be thin, fundamental data may be incomplete or delayed. AssetNext flags this transparently rather than concealing the gaps. A missing score is methodologically a stronger statement than a score filled with guesswork.

Structural comparability limits. The functional peer formation recognizes similar patterns but does not identify business model identity. For companies with highly idiosyncratic or hybrid business models, peer comparisons are accordingly to be interpreted with greater caution. This also applies to disruptive business models whose historical trajectories do not yet yield reliable peer similarity.

No pattern guarantee. Detected market patterns are descriptive. That a pattern has occurred in the past does not mean that a specific consequence can be derived from its occurrence today. Situations shows what is structurally notable — not what follows from it.

In summary

AssetNext is a tool for structured classification — not a recommendation system and not a forecasting instrument. The methodological foundation is peer-relative rather than absolute, separates perspectives rather than aggregating them, and makes its limits visible rather than concealing them. On this foundation, several analytical layers build — single-company analysis, direct comparison, portfolio analysis, market pattern detection — each with its own focus, all with the same stance: they are meant to serve judgment, not to replace it.