The Altares Score

Anticipate bankruptcy on a 12-month horizon thanks to the Altares default score, updated daily


The Altares-D&B Insolvency Score measures the risk of entering insolvency proceedings (safeguard, reorganization, judicial liquidation and European insolvency) in the next twelve months for French companies.

Altares takes the best of both worlds to build its score. On the one hand, it draws on historical expertise in corporate risk, proprietary data such as Paydex, and expert segmentation. On the other hand, it uses all the power of statistical and artificial intelligence algorithms to obtain the best possible learning of the failure.

Scoring Methodology

On an ongoing basis, we improve and adapt our rating taking into account long-term changes in market structure and economic outlook in order to maintain the performance and relevance of the score.

Our financial analysis has notably evolved on liquidity and solvency indicators in order to adapt to the context linked to the Covid-19 crisis.

Our membership in the Dun & Bradstreet global BtoB data network allows us to track corporate failures worldwide and thus to assess their potential impact on the French market (when they hold stakes in French companies or conversely if they are held by French parent companies).


Users of the score in France


Of companies with an active failure score in France


Daily update of the Altares default score

D&B's statistical model development process includes the following steps:

  • Segmentation of companies: entities are grouped according to their risk profile, their size, their nature and the availability of balance sheet information.
  • For each segment, we identify the relevant information (both from a statistical and a business point of view) to be selected and analyzed in order to predict entry into collective proceedings. This choice is made within a large panel of more than 600 variables mixing :
    • Structural data: administrative and identity-related
    • Relational data: statutory and operational managers, and capital links
    • Payment data: payment behaviors collected by our proprietary DunTrade program
    • Solvency data: financial data
    • Ad hoc industry and commercial data
  • The selected variables are then analyzed and transformed (cleaning of outliers, treatment of missing values, etc.) in order to extract the maximum information and thus optimize their predictive power
  • From the transformed variables, we build a scoring algorithm for each segment of companies, seeking to achieve the highest level of accuracy in predicting failures.

Keeping the score up to date with optimal reliability for our customers is the daily mission of our Data Scientists.


Age, legal form, number of employees, sector of activity,...

Stable over time (few modifications) allow to define a company typology

Legal events

Previous insolvency proceedings, recent legal opinions (change of head or location,...)

Be aware of ongoing changes and keep track of previous failures


Privileges, Payment behavior, Paydex

Useful for score responsiveness (weekly update and coverage in addition to MSDS information)


Static (last FS) and dynamic financial ratios and aggregates, age of last balance sheet.

An accurate credit analysis and structure can be obtained when the FS is available

Corporate Linkages

Local and foreign majority shareholder, local and foreign subsidiaries

Understands the financial strength of the group

Data Signals

Activity indicators (D&B reports, contact information, etc.)

New data can be built from the signals: identify inactive and very active companies

What is score performance?

The performance of the Altares score assesses the reliability of predictions made one year ago, based on multiple indicators, including two main ones: the Gini index*, and the share of failing companies among the 20% of companies with the lowest score.

* The Gini index transcribes the ability of the Altares score to give low scores to companies that will fail: 100% means a perfect score (all failing companies get the lowest scores) and 0% means a model that is unable to sort out healthy companies from risky ones. A score with a Gini index greater than 60% is considered a high quality score.

Using the score

Our statistical score has 4 risk levels: from “very low risk” for the safest companies to “high risk” for the riskiest.

The Failure Score helps companies by providing an assessment:

  • Financial risk: default is directly linked to insolvency
  • The risk of non-payment: monitoring of major overdue payments and liquidity problems
  • Operational risk: measuring inactivity and anticipating downtime

The combination of the Altares score and a human analysis (financial, for example) allows you to optimize your decision-making.

  • Credit decision support: partially automate the credit decision by rejecting credit applications from the riskiest companies and accepting those from the safest entities, take more collateral for companies whose creditworthiness is not certain.
  • Accounting provisioning: record anticipated losses taking into account the risk of non-payment associated with each invoice (for example, for a receivable of €10,000 with a risk of default of 2%, a provision of €200 is made for anticipated losses).

Prenez rendez-vous avec un expert