Bolagsanalys
Methodology

How we assess companies

A plain guide to our scores - simple first, more detail below.

In short

We read the digital annual report, compute several risk measures and show a composite 0-10 score with a risk band. You see both the whole and the parts.

Not a sample - the full digital market

Most tools cover a few thousand listed names. Bolagsanalys is built on the full Swedish register of companies that filed digital annual reports (iXBRL via Bolagsverket). That is the difference between a sample and a population - and our biggest advantage.

3.2 million+

annual-report periods analysed

570,000+

companies with a current risk score

Where do the numbers come from?

The full digital AB register

Public iXBRL annual reports filed with Bolagsverket. We process the population of digital filings - income statement and balance sheet - not a small sample.

Millions of reporting periods

Over 3.2 million analysed periods (many companies have multiple years). That supports training, history and comparability over time.

Real outcomes + current scores

The default model is trained on known outcomes with temporal splits. In the product you typically see the latest filing per company - 570,000+ companies with a ready score.

Score 0-10

10 = stronger picture. 0 = very weak. Risk bands are the same everywhere in the product:

8–10

Low risk

6–7.9

Moderate

4–5.9

Elevated

2–3.9

High risk

0–1.9

Critical

Parts of the analysis

You don’t need to calculate anything. Here’s what each part means in practice.

Swedish PD (default risk)

Probability-style signal trained on Swedish firms. About a 24-month horizon. Higher PD = lower score.

Altman Z''

Classic financial-stress model using book values - suited to private firms.

Zmijewski

Three simple ratios: profit/assets, liabilities/assets and liquidity.

Nordic / SME signals

Rule-based flags for smaller firms: leverage, cash, interest, tax liabilities and more.

Quality flags

Warnings such as negative equity, razor-thin profit or missing key items.

Composite score

Average of available component scores. Missing data is simply left out.

How to use the result

  • Start with risk band and composite - often enough as a first filter.
  • Open the parts that drag the score down - that’s the explanation.
  • Combine with common sense and other info (industry, orders, management).
  • Use as input for review - not as the only decision.

What we don’t do

  • We do not give buy, sell or hold advice.
  • We are not a statutory credit report.
  • We don’t see everything: future orders, hidden disputes or management intent.
  • Models err. Low risk is not zero risk.

Listed companies (add-on)

For some listed names we also offer Bolagspoäng: 17 parameters in four pillars. It complements the Swedish AB registry.

For those who want more depth

Academic roots (selection):

  • Altman (Z-score / Z'' för bokförda värden)
  • Zmijewski (1984) — probit-baserad distressmodell
  • Nordisk/SME-forskning om konkursprediktion (t.ex. Yazdanfar m.fl.)
  • Egen svensk PD-modell tränad på bokslut + konkursfacit

All information is a forensic signal from public accounts. No guarantees. Business, credit and investment decisions are your own.