What is Loughran-McDonald Dictionary?
The Loughran-McDonald dictionary is a finance-specific sentiment lexicon developed by Tim Loughran and Bill McDonald (Notre Dame) that classifies words used in 10-K and 10-Q filings as positive, negative, litigious, uncertain, modal-strong, or modal-weak. It is the de facto standard input for financial NLP sentiment scoring.
Why Loughran-McDonald Dictionary matters
Generic sentiment dictionaries (like Harvard IV-4 or AFINN) misclassify finance terms. The word "liability" is neutral in finance but flagged negative in general English. The word "tax" is neutral in finance but flagged negative everywhere else. Loughran-McDonald corrects for these misclassifications, which is why every credible financial NLP system uses it as a foundation. If your IR vendor scores filings using a non-financial sentiment dictionary, the scores are systematically biased and the recommendations are unreliable.
How AxonIR measures it
AxonIR uses the latest Loughran-McDonald master dictionary (updated periodically by the original authors) as the base sentiment layer. We extend it with a SPAC-specific overlay to catch trust-language and redemption-related vocabulary, and a sector overlay for biotech, mining, and real-estate-specific terminology. Every AxonIR readability report includes a Loughran-McDonald breakdown by category — positive, negative, litigious, uncertain — so issuers can see exactly which word classes are dragging the composite down. The litigious bucket is usually the highest-leverage one to fix.
The litigious-language category is particularly important for small-cap issuers. Excessive use of words like "litigation," "indemnification," "damages," and "settlement" in disclosure prose (as opposed to clearly delineated risk-factor sections) can cause algos to mark a filing as litigation-elevated, which triggers risk-off signals and can suppress liquidity for days.
Loughran and McDonald also publish modal-strong and modal-weak lists ("must" vs "may"), which AxonIR uses to score forward-looking-statement assertiveness. Filings that lean too heavily on weak modal language are marked by algos as low-confidence, which suppresses positive sentiment scoring even when the underlying business news is good.
See your company's loughran-mcdonald dictionary score
Free for any NYSE/NASDAQ ticker. 5-business-day delivery.
Get Free Algo Score