When a director, officer or 10% owner of a US-listed company buys or sells its stock, they file a Form 4 with the SEC within two business days. The filings are public, but each one is an XML document somewhere in EDGAR, and a single filing can hold several transactions. Our SEC Insider Trading API on Apify reads them and gives you one row per transaction.
One stock, last 90 days
Say you want the open-market buys and sells by NVIDIA insiders over the last 90 days. The input:
{
"tickers": ["NVDA"],
"transactionCodes": ["P", "S"],
"lookbackDays": 90,
"maxResults": 3,
"includeClusterSignals": false
}
P is an open-market purchase and S an open-market or private sale. Leave transactionCodes out to get every Form 4 code, including grants and option exercises.
One of the rows we got back on 17 November 2026, trimmed:
{
"ticker": "NVDA",
"issuer_name": "NVIDIA CORP",
"insider_roles": ["officer"],
"officer_title": "EVP, General Counsel and Sec",
"transaction_date": "2026-09-21",
"transaction_code": "S",
"transaction_type": "Open-market or private sale",
"shares": 4499,
"price": 223.75,
"value_usd": 1006651.25,
"shares_owned_after": 2687660,
"plan_10b5_1": true,
"filing_date": "2026-09-23",
"source_url": "https://www.sec.gov/Archives/edgar/data/1045810/000169684126000014/0001696841-26-000014-index.htm"
}
The full row also names the insider and their SEC CIK, as the filing does.
Reading the row
-
plan_10b5_1says the trade was made under a Rule 10b5-1 trading plan, set up in advance. A planned sale says much less about what the insider thinks than a sale they chose to make that week. -
filing_dateandtransaction_datediffer: the trade happened on the 21st and was filed on the 23rd. If you backtest a signal, use the filing date (the row also hasusable_from, the first day you could have acted on it). -
source_urlopens the filing on sec.gov, so you can check any number against the original.
The whole market, and clusters
Leave tickers empty to read every Form 4 in the date range. Set clusterBuysOnly to true to keep only open-market purchases where several insiders of the same company bought within a short window (clusterWindowDays); those rows say how many insiders and how much in total. Data goes back to 2006.
On a daily schedule, set onlyNewSinceLastRun to get, and pay for, only the transactions you haven't seen.
Run it
On the Store page, click Try for free, paste the input and press Start. From Python, with Apify's official client (pip install apify-client):
from apify_client import ApifyClient
client = ApifyClient("<APIFY_TOKEN>")
run = client.actor("bosco_labs/sec-insider-trades").call(run_input={
"tickers": ["NVDA"],
"transactionCodes": ["P", "S"],
"lookbackDays": 90,
})
for row in client.dataset(run["defaultDatasetId"]).iterate_items():
print(row["transaction_date"], row["transaction_code"], row["value_usd"])
From an AI agent, add this server URL to your MCP client (Claude, Cursor, VS Code):
https://mcp.apify.com?tools=bosco_labs/sec-insider-trades
The agent then has the Actor as a tool and reads the rows itself.
Price
$0.0018 per transaction row ($1.80 per 1,000) and $0.02 per cluster signal, through Apify's pay per event. The same data is on our own HTTP API if you'd rather not use Apify.
Top comments (0)