A couple of weeks ago we commented unfavourably on the use of PDF versions of SEC filings as inputs to Large Language Models (LLMs). We are much more interested in how these tools can be leveraged to examine structured data.
XBRL Calculations can play an important role in ensuring the quality and accuracy of a financial report, and can flag up errors in both the XBRL tagging and the underlying numbers. Unfortunately, until now, calculation validation reports have been hampered by false positives — validation messages that don’t reflect real issues in the report.
One item in particular over the holiday break caught our attention. Research by Patronus AI highlighted apparent challenges faced by large language models (LLMs), such as OpenAI’s GPT-4, in analysing financial data contained in US Securities and Exchange Commission (SEC) filings.
This week Europe published agreed-upon regulation to establish the European Single Access Point (ESAP). Designed to provide centralised access to publicly relevant data from across EU member states, ESAP has the potential to be a game-changer for digital data in Europe.
As we have written here before, the Basel III requirement for banks to publish a range of safety and soundness data — a summary of the prudential data they disclose to central banks and regulators — has largely been a failure.