Each year, RapidRatings measures how well the Financial Health Rating (FHR) separates companies likely to default from companies likely to survive. This article summarizes that analysis from the 2025 Annual Default Review.
What This Measures
To test classification accuracy, RapidRatings compares two groups as of January 1, 2025:
2025 Default Cohort: the 40 U.S. non-financial companies that went on to default during the year.
US Coverage: the broader population of U.S. non-financial companies RapidRatings rates.
If the FHR is working well, defaulters should be heavily concentrated in the higher-risk categories at the start of the year, while the general population should be spread more evenly across the risk spectrum.
Risk Distribution: Defaulters vs. the General Population
Going into 2025, the two groups looked very different:
Risk Category (FHR) | 2025 Defaulters | US Coverage |
Very High Risk (0–19) | 25% | 2% |
High Risk (20–39) | 63% | 21% |
Medium Risk (40–59) | 10% | 34% |
Low Risk (60–79) | 3% | 27% |
Very Low Risk (80–100) | 0% | 16% |
88% of 2025 defaulters were rated High or Very High Risk at the start of the year, compared with just 23% of the general population.
Lift: How Much More Likely to Default
Lift measures how concentrated defaulters are in a risk category compared to the general population. A Lift of 1× means a category defaults at the same rate as a random company; higher than 1× means that category is over-represented among defaulters.
Risk Category (FHR) | Lift |
Very High Risk (0–19) | 12.5× |
High Risk (20–39) | 3.0× |
Medium Risk (40–59) | 0.3× |
Low Risk (60–79) | 0.1× |
Very Low Risk (80–100) | 0× |
A company rated Very High Risk was 12.5 times more likely to default in 2025 than a randomly selected company in RapidRatings' US coverage, and 125 times more likely than a company rated Low Risk.
False Alarms and Missed Defaults
An accurate model should rarely flag healthy companies as high risk (a false alarm) and rarely miss a company that goes on to default (a missed default).
Only 20% of surviving companies were ever rated at or below an FHR of 40, a low false alarm rate.
Only 19% of companies that defaulted were rated above an FHR of 40 at the time of default, a low missed default rate.
Together, these two figures show that the FHR separates defaulters from survivors with a high degree of precision.
Seen in Practice: 2025 Examples
Several 2025 defaulters carried High Risk ratings years before they filed:
iRobot Corp was rated High Risk three years before its December 2025 filing.
Wolfspeed Inc was rated High Risk three years before its June 2025 filing.
Danimer Scientific Inc was rated High Risk three years before its March 2025 filing.
Why This Matters
A High Risk or Very High Risk rating identifies companies whose financial profile closely resembles that of historical defaulters.
Low Risk and Very Low Risk companies rarely default, and the model rarely misclassifies them.
This level of separation gives clients confident, actionable early warning with minimal false alarms.