10X Document Dictionaries

This file provides a tabulated dictionary of all words from the LM dictionary for each filing, which means that you can work with word counts without parsing the underlying documents.

Loughran-McDonald_10X_DocumentDictionaries_1993-2025.txt (16.3 GB)

This file contains header information and word counts for all 10X filings.

• The header data is pipe-delimited from the word counts.
• The word counts appear as word-sequence-number:word-count.
• A Python function to decode the records.
• A sample program to produce word counts using the function.

Here is a truncated line containing Google's first 10-Q filing:

1288776,20040816,0001193125-04-141838,20040630,10-Q,Google Inc.|101:42,126:16,184:17,186:10,213:1,248:1, ...

The first six fields before the pipe are: (1) cik, (2) filing_date, (3) accession_number, (4) conformed_period_report, (5) form_type, and (6) company_name.

The subsequent fields contain the word sequence number in the master dictionary, followed by the count for that word. The first field (101:42) indicates that the word "ability "occurred 42 times in the filing.