Gold subscribers have access to TextScore®️ included in their subscription. Navigate to the 'Financial Health Assessment' menu item and select 'Textscore'
TextScore® is currently available for UK quoted companies and UK large & medium private companies.
Based on annual/interim financial reports for publicly listed and large private companies, the Company Watch TextScore® does not consider fundamental financial information directly, but is based on a natural language processing (NLP) approach, looking only at the text in a company’s financial accounts.
By looking at a large database of historical financial reports and utilising advanced AI techniques, it is possible to identify the patterns of words/phrases that may be indicative of financial distress. We have put thousands of financial reports through this process to help our machine-learning models learn to differentiate between companies that went on to become financially distressed and those that did not. The result is a financial health score that is complementary to the H-Score® .
The chart on this page (above) shows the H-Score® graph (the blue line), the TextScore® graph (orange line) and the Combined score graph (green line). The Combined Score is a weighted combination of the H-Score® and TextScore®. The exact weighting used is determined by the industry sector in which a particular company operates. The resultant combined score is more accurate at predicting financial distress than the TextScore® or H-Score® alone as it combines the two different distinct views of a company’s financial report.
All the scores are ranked and take a value from 0 (weakest) to 100 (strongest). Companies with a score of 25 or less are described as being in the warning area: Almost all companies that go into distress were flagged in the warning area in advance of the distress event.
It is possible to explore the data by clicking on any TextScore® point on the chart to view more details on the source document, including:
- Filing details - you can select to view the original source documents
- Phrase count - a chart showing the word frequencies contained in the source document
- Implied Industry - Sometimes it isn’t possible to know exactly what industry a company is operating in directly because of inaccuracies in the reported SIC code. The implied Industry category shown is an attempt to solve this problem. By looking at historical filings we have built a statistical model that infers the likelihood of a company operating within one of 13 industry sectors from the words used. The bar chart shows the percentage confidence that the model has for each industry category.
Please note that the quality of the published financial reports could have an impact on the TextScore® - certain words that are indistinct may not be correctly picked up.