Insider trading data

Q: What's the single most useful insider trading data filter?

Transaction code P (open-market purchase), minimum dollar value of $250k, by an officer. That subset is small enough to read individually and most filings in it carry real signal.

Insider trading data is the public record of stock transactions by company officers, directors, and large beneficial owners. The data comes from a handful of specific SEC filings, and using it well requires knowing what each filing does and does not contain.

Want this data via API instead of reading about it? Get a free API key →

What counts as "insider trading data"

In the regulated, public-disclosure sense (not the criminal-law sense), insider trading data is everything in these filings:

Form 3, initial statement when someone first becomes an insider.
Form 4, every Section 16 insider transaction. The core dataset.
Form 5, annual cleanup of Form 4-exempt transactions.
Form 144, pre-sale notice for restricted/control securities.
Schedule 13D, activist 5%+ position disclosure.
Schedule 13G, passive 5%+ position disclosure.

Form 4 is by far the largest and most-used. There are typically several thousand Form 4 filings per week across all U.S. public companies.

What the data tells you

A Form 4 line gives you, for each transaction:

Who: the insider's name, role, and CIK.
What: the security (typically common stock), the transaction code, the share quantity.
When: the transaction date (not the filing date) and the filing date.
Why (somewhat): the transaction code distinguishes open-market purchases from option exercises from tax withholdings from gifts.
Price.
Post-transaction holding.
Optional footnotes: 10b5-1 plan references, indirect ownership explanations, etc.

That's enough for a meaningful screening workflow but not enough for full attribution. You don't get the insider's motivation. You get the form, not the reason.

What the data does not tell you

Common misreadings:

It does not tell you why the insider traded. A sale can be planned (10b5-1), tax-driven, diversification, divorce, charitable giving, or directional. A buy can be conviction, signaling, or required (some companies have minimum ownership requirements for executives).
It does not tell you about non-insiders. Most short-term trading by hedge funds is not reported as insider data. The 13F system covers institutional holdings on a 45-day lag, not real-time flows.
It does not include shorts. Long-only.
It does not include international markets. U.S. only.

The three signals people actually care about

1. Open-market purchases

Transaction code P. The insider used their own cash to buy stock at market prices. The strongest single-filing bullish signal in the entire Form 4 universe.

Filter by code P and a meaningful dollar threshold (typically $100k+) to remove noise.

2. Cluster buying

Three or more distinct insiders at the same issuer placing open-market purchases within a 7-14 day window. Cluster patterns have a stronger track record in academic studies than individual buys.

To detect clusters via API:

curl "https://api.edgarkit.com/v1/filings?form_type=4&transaction_code=P&min_value=50000&since=2026-06-05&limit=500" \
  -H "Authorization: Bearer YOUR_API_KEY"

Then group by issuer_cik and count distinct reporter_cik values per 14-day window.

3. 10%-owner accumulation

A holder crossing 5% files a 13D (activist) or 13G (passive). When the holder is already an insider on the company's books, their Form 4 activity adds to that picture. Combine 13D/G signals with Form 4 to build position-tracking on major holders.

Sales: noisier signal

Code S (open-market sale) is the inverse of P but with much weaker predictive value. Insiders sell for many non-informational reasons. Filter S by:

Whether the transaction is referenced as part of a 10b5-1 plan (visible in footnotes), discount these heavily.
Whether multiple officers are selling in the same window (cluster sells are a stronger negative signal than single sells).
Whether the sale is a meaningful fraction of post-transaction holdings.

A CEO selling 5% of their position is routine. A CEO selling 80% of their position is a signal.

How to access the data

Direct from SEC EDGAR

The raw filings are on EDGAR. You poll the Atom feed, fetch XML, parse, and store. The catches are detailed in SEC EDGAR rate limits and the schema changes over the years.

Expect to spend 2-4 engineer-weeks plus ongoing maintenance.

Via EdgarKit

EdgarKit's API returns insider transactions as normalized JSON, with the heavy lifting (parsing, CUSIP mapping, ticker resolution, amendment tracking) done for you. The free tier covers 1,000 requests per month, which is enough to prototype most screening strategies.

# Most recent open-market purchases above $500k
curl "https://api.edgarkit.com/v1/filings?form_type=4&transaction_code=P&min_value=500000&limit=25" \
  -H "Authorization: Bearer YOUR_API_KEY"

Via web dashboards

OpenInsider, Form4Oracle, and similar dashboards present Form 4 data to humans. Useful for one-off lookups; not useful if you're building anything programmatic. See EdgarKit vs OpenInsider.

Historical depth

EDGAR has Form 4 filings going back to the early 2000s. Pre-2003, the deadline was 40 days (changed to 2 business days by Sarbanes-Oxley). Pre-1996, filings were paper-only and not in EDGAR.

For backtesting, the practical clean window is 2003-present.

Common research strategies

Form 4 cluster buys + earnings surprise filter. Look for cluster buying within 30 days of a positive earnings 8-K.
Insider buying at 52-week lows. Insiders buying near the bottom of the recent range tend to time better than insiders buying after a run-up.
CFO purchases specifically. CFOs have the cleanest line of sight to forward financials; their P codes are particularly informational.
Following 13D filers' subsequent Form 4 activity. An activist who builds a position via 13D and then keeps adding via Form 4 is signaling conviction.

FAQ

What's the difference between insider trading data and insider trading (the crime)?

The data is the public disclosure of trades by company officers, directors, and 10% owners. The crime is trading on material non-public information. The disclosure system exists in part to discourage and detect the crime, but the data itself is fully legal trading by insiders that has been publicly reported.

How fast is insider trading data available?

The SEC accepts the filing immediately and the document is public the moment it's accepted. Real-time delivery (within 30 seconds) is possible by polling EDGAR; batched daily indexes are 12-24 hours behind.

Can I backtest insider trading strategies?

Yes. The data goes back to 2003 in clean form. Use filing date, not transaction date, as the data-availability date in your backtest to avoid look-ahead bias.

Are 10b5-1 plan sales weaker signal than discretionary sales?

Yes. 10b5-1 sales are pre-scheduled, often months in advance, when the insider was not in possession of material non-public information. Footnotes reference the plan adoption date.

What's the single most useful insider trading data filter?

Transaction code P (open-market purchase), minimum dollar value of $250k, by an officer. That subset is small enough to read individually and most filings in it carry real signal.