Market data
Ask the market what actually happened.
Upload regulatory filings, price files or positions data, and ask the questions a spreadsheet makes painful. Crunch derives what isn’t in the files — who bought together, streaks, windows around events — joins it across datasets, and shows exactly how every number was counted.
Free to start · fully self-serve
Real SEC data · January–May 2026
When insiders buy together
The data, joined
- Filings108,302 Form 4 and 4/A filings — company, ticker, form type
- Insiders114,297 rows — an anonymous ID and the filed role, no names
- Transactions179,708 rows — every kind: purchases, awards, sales, gifts
The question
“Counting purchases (SEC transaction code P) with a transaction date from January 1 through May 31, 2026, by insiders whose filed relationship is Director, Officer, or Director,Officer: which five company-months had the most different insiders purchasing — and for those companies, how many purchased, in how many months, and on how many days?”
What Crunch did
- The Plan agent split the question into four asks and showed the plan before running anything.
- The Analyst picked real purchases out of every kind of transaction, joined each filing to its insiders and company, and counted different buyers per company and month.
- It wrote the results back into the plan — three asks solved, one (who bought in more than one month) left openly unresolved rather than guessed.
The answer
Five company-months stand out from 108,302 filings — and every count on screen checks out against the filings.
Source: U.S. Securities and Exchange Commission, Insider Transactions Data Sets (public domain). Insiders’ names and addresses were not loaded. A description of what happened — not investment advice.
Hundreds of thousands of rows, one question
Too big to eyeball, too fiddly for a quick formula. Upload the files and ask.
Get the files
Regulatory filings, daily price files, positions — whatever your regulator, exchange or data vendor gives you the right to use.
Upload them
CSV, Excel or Parquet. Each file becomes a dataset in your own workspace, next to the others.
Ask across them
Crunch works out how the datasets line up — by date, by symbol — and answers with the reasoning attached.
Questions it answers
What happened, measured
“On the days the index rose, how many stocks actually did?”
“Which sectors led in the months the market fell?”
“How does this quarter’s insider buying compare with the last four?”
“Which stocks traded far above their usual volume this quarter?”
“Which companies had several insiders buy within the same two weeks?”
“Which small caps beat their index, month by month?”
Crunch describes what the data shows. It doesn’t give investment advice.
Upload a file and ask
Sign up free, upload your first dataset, and ask the question you actually have.
Sign up free