top of page

Frequently Asked Questions 

What Is Legal Predictive Analytics?

Legal Predictive Analytics is the science of analyzing patterns, correspondences, and irregularities in large quantities of legal data in order to predict legal outcomes.

How Do Predictive Analytics For Insurance Claims Help?

Settlements vs. Litigation: understand what claims should be disputed or settled by evaluating the likelihood of success for your case (win rate) and the expected quantum granted (recovery rate) to create cost-effective strategies for closing a claim.

Optimize Payouts: anticipate your costs and get the right preparation for difficult cases while saving on mock trials expenses, other legal fees, and increasing your confidence against aggressive plaintiffs.

Anticipate Devastating Verdicts: detect early on latent outlying cases and protect your company from devastating verdicts while exploring the influence of specific litigation micro-variables such as venues, judges, jury compositions, attorneys, and experts to mitigate social inflation.

Obtain a High Return on Investment: insurance carriers, law firms, legal funds, and corporations reduce their loss ratio up to 10 points over time through optimized payouts, reduced legal spending, lowered litigation rates, and eliminated devastating verdicts.

Where Does Your Data Come From?

Our predictions come from a combination of public and private claims litigation data.

We never communicate any client data: ethically we are fully committed to data privacy and technically our databases are securely hosted in the US by the most powerful service providers. These standards were recently affirmed by our Soc 2 Type II Certification. 

What Does 'The Path From Data To Prediction' Mean?

It's a summary of our technology development. First, we start with legal theory in understanding the relevant law in any claims situation. Second, from a small sample, we development a hypothesis of the legal and economic variables which predict the outcome in those particular cases. Third, using either public or private sources, we automatically extract and curate the relevant predictive variables from a large data set. Finally, we validate the hypothesis on this larger sample.


Our system iterates the process as many times as necessary to optimize the predictive algorithm in terms of both accuracy and interpretability. Finally, those algorithms are embedded into an intuitive user-interface called AGATHA.

bottom of page