Tradytics for Python

The official Python client for the Tradytics API: options flow from the full OPRA tape, dark pool prints ranked against a year, dealer gamma, delta, vanna and charm, and the whole market around them. 79 endpoints, one method each, no dependencies.

pip install tradytics

What you get

• Options flow: every print of $25K+ from the full OPRA tape (every US options exchange), live and back to March 2025, with side, sweep or block, and whether it opened a position.
• Dark pool: block and off-exchange prints as they happen, each ranked against a year of its kind, and the levels where size traded.
• Dealer exposure: gamma, delta, vanna and charm by strike for about 2,500 symbols, every five minutes, with the change since yesterday, a week and two weeks.
• The market: quotes and bars from 1 second to 1 month, breadth, sectors, movers, the options market, research, the full-market scanner, Signals, Kalshi and Polymarket odds and the AI Fund.
• Built for scripts and services: retries on rate limits, paging and live streaming built in, typed and documented methods, and plain JSON back.

Quickstart

The API is part of Tradytics Pro. Make a key at tradytics.com/developer/keys and pass it in, or export it as TRADYTICS_API_KEY. A key you pass wins over the environment.

from tradytics import Client

client = Client("tdy_...")   # or Client(api_key="tdy_...")
client = Client()            # or set TRADYTICS_API_KEY and pass nothing

flow = client.options.flow(symbol="NVDA", min_premium=1_000_000, opening=True)
for p in flow["data"]:
    print(p["time"], p["symbol"], p["type"], p["strike"], p["expiry"], p["premium_usd"])

Objects come back as {"data": ..., "meta": {...}} and lists as {"data": [...], "next_cursor": ..., "next_url": ..., "meta": {...}}, the same JSON as the REST API.

Examples

# The year's biggest dark pool prints for one stock, ranked against its own year
client.darkpool.prints(symbol="AAPL", from_="2026-01-01", max_rank=10)

# Dealer gamma by strike with the call wall, put wall and flip; 0DTE only
client.options.exposure(symbol="SPY", greek="gamma", dte=0)

# Every quarter Apple has filed, and its 8-Ks as they land
client.fundamentals.income_statement(symbol="AAPL", period="quarter", limit=40)
client.filings.list(symbol="AAPL", form="8-K")

# Insider buys over $1M, and Tesla's deliveries quarter by quarter
client.disclosures.insiders(type="buy", min_value=1_000_000)
client.kpis.list(symbol="TSLA", kpi="vehicle_deliveries")

# Screen the market on any field
client.screener.stocks(min_relative_volume=2, max_rsi=30, sort="change_pct", limit=25)

Every page

for p in client.paginate(client.options.flow, symbol="TSLA", min_premium=500_000, from_="2026-01-01"):
    print(p["time"], p["premium_usd"])

New prints as they happen

for p in client.stream(client.options.flow, min_premium=250_000):
    print(p["symbol"], p["type"], p["strike"], p["premium_usd"])

stream works with client.options.flow and client.darkpool.prints, polls every two seconds and yields each print once, oldest first.

Errors, retries and limits

from tradytics import Client, TradyticsError

try:
    client.options.flow(symbol="NVDA", limit=5000)
except TradyticsError as e:
    print(e.status, e.code, e.message, e.request_id)  # 400 invalid_parameter ...

Each key gets 300 calls a minute and 50,000 a day. A 429 or 5xx answer is retried up to three times, waiting as Retry-After says; the daily quota isn't retried. Client(api_key=..., timeout=30, max_retries=3) changes the defaults, and client.get("/any/path", **params) calls any endpoint by path.

Reference

Parameters and fields for each are in the method's docstring and on its docs page.

More

• API docs, with a page per endpoint and a live console
• JavaScript SDK: npm install tradytics
• MCP server for Claude, ChatGPT and Cursor
• Support: hello@tradytics.com