CAPE Analytics Launches Roof Condition Rating Version 5

Our newest model delivers a new level of insight, explainability, and stability in AI-driven roof assessments

Today, we’re excited to announce the launch of CAPE Roof Condition Rating (RCR) version 5 (v5), a groundbreaking advancement in automated roof condition assessments. For this newest model, our team leveraged years of innovation and market adoption in order to bring new enhancements designed to boost model performance, enhance stability over time, and reduce uncertainties due to tree cover or shadow. RCR v5 offers insurance carriers, underwriters, actuaries, and property stakeholders CAPE’s most performant model yet.

At CAPE Analytics, a Moody’s company, we pioneered broad roof condition assessments from aerial imagery in 2016 and 2017, laying the foundation for leveraging computer vision and geospatial imagery to provide comprehensive and timely property insights. By 2018, this capability was available across the continental United States, and by 2019, we had shown that CAPE’s AI-based roof condition rating could predict insurance losses with greater fidelity than homeowner- or agent-reported roof age alone. Since then, CAPE RCR has been adopted by more than 100 carriers, with over 300 approved rating and underwriting filings across over 40 states.

Over multiple iterations, our team has continuously and rigorously improved the RCR model. Key innovations have included:

• Confidence Scores & Reason Codes: Users gain insight into the model’s determination, including the identification of the specific defect leading to a particular rating.
• Interoperable Across Imagery Vendors: CAPE supports and maintains model functionality across multiple imagery sources and resolutions.
• Broad Coverage: With 98% U.S. coverage, plus high coverage across Canada and Australia, carriers, underwriters, and actuaries can leverage RCR data across diverse regions.
• Deep Support Structure: CAPE offers extensive implementation options, rate filing and State DOI support, user training, underwriter education, and ongoing carrier resources.

Now, with the launch of RCR v5, CAPE is introducing updates that enhance loss predictiveness across peril categories, such as wind and wind/hail, and further increase model accuracy (as defined by precision and recall). This newest iteration enhances stability over time, which can provide more consistent results with changing imagery and seasonal conditions. Finally, RCR v5 reduces cases in which the model determines the condition of a roof to be classified as “unknown” due to tree cover or shadows obscuring the roof. In totality, these improvements result in the most sophisticated model in CAPE’s history.

“With RCR v5, CAPE is pushing the boundaries of what’s possible in automated roof assessments,” said Mike McCormick, Product Manager at CAPE Analytics. “Our commitment to innovation and the success of property carriers has never been stronger. We look forward to offering insurance carriers and property stakeholders our highest-quality insights to make informed risk decisions swiftly and confidently.” 

Much more information is available to CAPE clients, including white papers, product overviews, and more. To learn about Roof Condition Rating v5, contact your CAPE Analytics team today.