DataVents Open SDK

Unified Python SDK for Kalshi and Polymarket.

What’s inside

• Unified, no‑auth REST façade over Kalshi and Polymarket public endpoints.
• Normalization helpers and Pydantic models for events, markets, search, and history.
• A multiplexed WebSocket client that streams from Kalshi and Polymarket concurrently.

Supported Python: 3.9+

Installation

From PyPI (preferred):

pip install datavents

From a local wheel (e.g., CI artifact):

pip install /path/to/datavents-0.1.0-py3-none-any.whl

Editable install for contributors:

pip install -e .[dev]

---

Quickstart

List events (both providers) and print a few titles:

from datavents import DataVentsNoAuthClient, DataVentsProviders, DataVentsOrderSortParams, DataVentsStatusParams

client = DataVentsNoAuthClient()
events = client.list_events(
    provider=DataVentsProviders.ALL,
    limit=5,
    page=0,
    status_params=DataVentsStatusParams.OPEN_MARKETS,
    order_sort_params=DataVentsOrderSortParams.ORDER_BY_TRENDING,
)
for payload in events:
    print(payload["provider"], "→ keys:", list(payload["data"].keys()))

Normalize provider search into one schema:

from datavents import normalize_search_response, Provider, OrderSort, StatusFilter

raw = events[0]["data"]  # e.g., Kalshi payload
normalized = normalize_search_response(
    Provider.kalshi,
    raw,
    q="election",
    page=1,
    limit=5,
    order=OrderSort.trending,
    status=StatusFilter.open,
)
print("Normalized results:", len(normalized.results))

Explore more complete runnable samples under examples/.

---

Unified Client

The DataVentsNoAuthClient exposes a small set of high‑leverage methods. All network calls hit live provider public APIs (no credentials needed for REST). Each method returns a list of provider‑tagged dictionaries like:

[{"provider": "kalshi", "data": {...}}, {"provider": "polymarket", "data": {...}}]

search_events

search_events(
    provider: DataVentsProviders,
    query: str,
    limit: int,
    page: int,
    order_sort_params: DataVentsOrderSortParams,
    status_params: DataVentsStatusParams,
    **params
) -> list[dict]

• Provider‑specific mapped sorts and statuses (see Enums below).
• When provider=ALL, Kalshi and Polymarket are called in parallel.
• Returns raw provider search payloads (use normalization helpers to unify shape).

Common extra params

• kalshi_params: dict – forwarded to Kalshi search. If scope="series" with a status other than ALL, client falls back to Kalshi events search (series endpoint doesn’t support status).
• polymarket_params: dict – forwarded to Polymarket public‑search call.

list_events

list_events(
    provider: DataVentsProviders,
    limit: int = 50,
    page: int = 0,
    status_params: DataVentsStatusParams = DataVentsStatusParams.ALL_MARKETS,
    query: str = "",
    order_sort_params: DataVentsOrderSortParams = DataVentsOrderSortParams.ORDER_BY_TRENDING,
) -> list[dict]

• Thin wrapper around provider search tuned for event discovery.
• Use with normalize_search_response for unified typing.

list_markets

list_markets(
    provider: DataVentsProviders,
    limit: int = 50,
    page: int = 0,
    status_params: DataVentsStatusParams = DataVentsStatusParams.OPEN_MARKETS,
    query: str = "",
    order_sort_params: DataVentsOrderSortParams = DataVentsOrderSortParams.ORDER_BY_TRENDING,
) -> list[dict]

• Kalshi path currently uses events search (markets are nested under events).
• Polymarket path uses public‑search with type="markets" and exposes a top‑level markets list (flattened if needed).

get_event / get_market

get_event(provider, kalshi_event_ticker=None, polymarket_id=None, polymarket_slug=None, *, with_nested_markets=False, include_chat=False, include_template=False) -> list[dict]
get_market(provider, kalshi_ticker=None, polymarket_id=None, polymarket_slug=None, *, include_tag=False) -> list[dict]

• Supply the proper identifier for each provider (Kalshi uses tickers; Polymarket supports id or slug).
• With provider=ALL, only providers with identifiers provided are called.

Orderbook (Kalshi, lazy auth)

While the client is "no‑auth" by default, it can lazily spin up a signed Kalshi REST client for auth‑only routes:

from datavents import DataVentsNoAuthClient, DataVentsProviders
from datavents.providers.config import Config as ProviderConfig

dv = DataVentsNoAuthClient()

# Direct helper (returns raw provider JSON)
ob = dv.get_kalshi_market_orderbook("ABC-24-XYZ-T50", depth=50, env=ProviderConfig.LIVE)

# Unified facade (list with provider tag)
res = dv.get_market_orderbook(DataVentsProviders.KALSHI, kalshi_ticker="ABC-24-XYZ-T50", depth=50)

Set environment variables for Kalshi auth:

• LIVE: KALSHI_API_KEY, KALSHI_PRIVATE_KEY (PEM path)
• PAPER: KALSHI_API_KEY_PAPER, KALSHI_PRIVATE_KEY_PAPER

Provider‑specific helpers

• get_event_metadata(event_ticker: str) – Kalshi event metadata.
• get_event_tags(event_id: int), get_market_tags(market_id: int) – Polymarket tags.

---

Normalization Layer

Convert raw provider payloads into consistent Pydantic models located in datavents.schemas.

Key entry points

• normalize_search_response(provider, raw, q, order, status, page, limit) → SearchResponseNormalized
• normalize_event(provider, raw) → Event
• normalize_market(provider, raw) → Market
• normalize_market_history(provider, identifiers=..., start, end, interval, raw|points) → MarketHistoryResponseNormalized

Highlights

• Prices normalized to probability in [0,1] (accepts common provider units such as 0..1, percent, or bps‑like).
• Timestamps normalized to epoch milliseconds.
• status mapped to StatusNormalized (open/closed/settled/upcoming/unknown) while retaining status_raw.
• Provider snapshots preserved under vendor_raw with optional structured vendor_fields.

Example (market → binary outcomes):

from datavents import normalize_market, Provider

raw_market = {"id": 123, "slug": "will-something-happen", "bestBid": 0.47, "bestAsk": 0.49}
m = normalize_market(Provider.polymarket, raw_market)
print(m.mid_price, [o.name for o in m.outcomes])

---

WebSockets

Use DvWsClient to multiplex Kalshi and Polymarket streams with a single async loop.

import asyncio
from datavents import DvWsClient, DvSubscription, DvVendors

async def main():
    dv = DvWsClient()

    async def on_event(evt):  # evt is datavents.ws.NormalizedEvent
        print(evt.vendor, evt.event, evt.market)

    sub = DvSubscription(
        vendors=(DvVendors.POLYMARKET,),
        polymarket_assets_ids=["asset-id-1", "asset-id-2"],
    )

    await dv.run(sub, on_event)

asyncio.run(main())

Kalshi WS requires valid API credentials in your environment. Polymarket market WS is public and needs only asset ids.

Environment variables (Kalshi; see provider docs)

• KALSHI_API_KEY (live) or KALSHI_API_KEY_PAPER (paper)
• KALSHI_PRIVATE_KEY (live) or KALSHI_PRIVATE_KEY_PAPER (paper) – path to PEM file used for RSA‑PSS signing

---

Enums and Mappings

Order sorts (DataVentsOrderSortParams) map to provider‑specific semantics, e.g.:

• ORDER_BY_TRENDING → Kalshi trending, Polymarket volume_24hr (proxy).
• ORDER_BY_CLOSING_SOON → Kalshi closing, Polymarket end_date.

Status filters (DataVentsStatusParams):

• OPEN_MARKETS → Kalshi opened,unopened; Polymarket active.
• CLOSED_MARKETS → Kalshi closed,settled; Polymarket closed.
• ALL_MARKETS → no status filter.

Note: Kalshi series search does not accept a status filter; when you request scope="series" with a non‑ALL status, the client uses events search under the hood.

---

Configuration, Timeouts, and Rate Limiting

• HTTP timeout (seconds): HTTP_TIMEOUT_SECONDS env (default 15s).
• Basic pacing between requests via RateLimitConfig (default ~100ms between calls).
• Logging: the SDK uses the standard library’s logging module; enable debug logs by configuring handlers in your app.

---

Examples

See examples/ for small, runnable scripts:

• search_events.py – search across providers.
• list_markets.py – get Polymarket markets and Kalshi event‑nested markets.
• get_event.py / get_market.py – fetch single resources by id/ticker/slug.
• ws_multiplex.py – stream from Polymarket and/or Kalshi in one loop.
• normalize_examples.py – convert raw payloads into normalized Pydantic models.

Run like:

python -m venv .venv && source .venv/bin/activate
pip install -e .
python examples/search_events.py --q "election" --limit 5

---

Utilities

• extract_vendors(value, *, strict=False) → [DvVendors, ...]
  • Accepts flexible input such as "kalshi", "poly,kalshi", "all", Provider.kalshi, or a list of mixed tokens.
  • Returns a stable, de‑duplicated order: [DvVendors.KALSHI, DvVendors.POLYMARKET] when both.
  • Back‑compat alias: _extract_vendors(...).

from datavents import extract_vendors, DvVendors

assert extract_vendors("all") == [DvVendors.KALSHI, DvVendors.POLYMARKET]
assert extract_vendors(["k", "pm"]) == [DvVendors.KALSHI, DvVendors.POLYMARKET]

• build_ws_info(subscription) → dict summarizing a DvSubscription (WS URLs, channels, identifiers).
  • Back‑compat alias: _send_ws_info(subscription) for legacy callers.

from datavents import DvSubscription, DvVendors, build_ws_info

sub = DvSubscription(
    vendors=(DvVendors.KALSHI, DvVendors.POLYMARKET),
    kalshi_market_tickers=["SER-ABC-123"],
    polymarket_assets_ids=["asset-1"],
)
info = build_ws_info(sub)
print(info["kalshi"]["ws_url"], info["polymarket"]["ws_url"])  # ready for UIs/logging

---

Limitations & Notes

• This SDK targets provider public endpoints. Some features (e.g., private Kalshi WS) require valid credentials.
• list_markets on Kalshi returns markets nested inside events; flatten as needed.
• Provider APIs evolve; enums map to best‑available sorts/filters and may be updated across releases.

---

Development

python -m venv .venv
source .venv/bin/activate
pip install -e .[dev]
pytest

To publish artifacts:

make build   # dist/*.whl + dist/*.tar.gz
make check   # twine check dist
make publish # twine upload dist/*

---

License

MIT