What is "data-driven financial materiality?"

Sustainability reporting / strategy / investing starts with materiality. That is, what's important AND what isn't. From a board's point of view, which sustainability drivers move the value of the company?

A data-driven approach uses well-known mathematical methods to pinpoint those material drivers. There are four steps:

🧮 Statistical screening. Mutual Information measures statistical dependency, and Area Under the Cure evaluates classification power. This ranks the 12 most material drivers.

🌲 Random Forest. This uses 100 trees and prioritises drivers based on their multivariate interactions. This reduces from 12 to 8 most material drivers.

🕰️ Validation over Time. The stability of the selected drivers is tested across rolling 26 week windows. Driver rankings can shift based on economic or social regime changes.

🪑 Industry benchmarking. The process is repeated across a peer group and industry benchmark on a weekly basis. This provides context and insight for reporters, assurers and investors.

For a detailed discussion, see Majid Jangani, Paresh Date and Andrew Tucker. 2026. A New Approach to Identifying Key Financial Materiality Drivers from Sentiment Data

Previous
Previous

What's wrong with financial materiality assessments for sustainability reporting?

Next
Next

Trustworthy as a feature, not a bug