Probalytics Python Client

Python client for Probalytics market data. Use it to query markets, fills, and orderbook snapshots directly from ClickHouse and return results as Polars or pandas dataframes.

Installation

pip install probalytics

The client requires Python 3.11 or newer.

Polars support is included by default. To use frame="pandas", install the pandas extra:

pip install "probalytics[pandas]"

Connect

from probalytics import ProbalyticsClient

client = ProbalyticsClient.from_clickhouse(
    host="clickhouse.probalytics.io",
    username="your_username",
    password="your_password",
)

By default, the client connects securely to the probalytics database. If your account uses a custom host, port, or database, pass those values when creating the client.

client = ProbalyticsClient.from_clickhouse(
    host="clickhouse.probalytics.io",
    port=9440,
    database="probalytics",
    username="your_username",
    password="your_password",
    secure=True,
    frame="polars",
)

You can also configure the client from environment variables:

export PROBALYTICS_CLICKHOUSE_USERNAME="your_username"
export PROBALYTICS_CLICKHOUSE_PASSWORD="your_password"
export PROBALYTICS_CLICKHOUSE_HOST="clickhouse.probalytics.io"
export PROBALYTICS_FRAME="pandas"

client = ProbalyticsClient.from_env()

Use the client as a context manager when you want the connection closed automatically:

with ProbalyticsClient.from_env() as client:
    markets = client.markets(platform="POLYMARKET", status="ACTIVE")

Query Markets

Markets are returned as typed Pydantic models.

markets = client.markets(
    market_platform_id="0xmarket",
    platform="POLYMARKET",
    status="ACTIVE",
    limit=100,
)

market = markets[0]
print(market.title)
print(market.platform)
print(market.platform_id)

You can also return markets as a dataframe.

markets_df = client.markets_frame(
    platform="KALSHI",
    status="ACTIVE",
    frame="polars",
)

Filters such as platform and status accept either one value or a list of values.

markets = client.markets(
    market_platform_id=["0xmarket", "KXBTC-26JUN-T50000"],
    platform=["POLYMARKET", "KALSHI"],
    status=["ACTIVE", "PAUSED"],
)

Query Fills

Fills return as a Polars dataframe by default.

fills = client.fills(
    market=market,
    start_time="2026-03-15T00:00:00Z",
    end_time="2026-03-16T00:00:00Z",
)

Use market, market_id, or market_platform_id to scope fills to one or more markets. A full market object is the most convenient option when you already loaded markets first.

fills = client.fills(market=market)
fills = market.fills()
fills = client.fills(market_id=market.id)
fills = client.fills(market_platform_id=market.platform_id, platform=market.platform)

market_id, market_platform_id, platform, taker_side, and trader_id also accept lists.

fills = client.fills(
    platform=["POLYMARKET", "KALSHI"],
    market_platform_id=["0xmarket", "KXBTC-26JUN-T50000"],
    taker_side=["BUY", "SELL"],
    start_time="2026-03-15T00:00:00Z",
)

Use a market selector or a bounded time range when querying fills. This keeps queries fast and avoids scanning more data than you need.

Filter by participant or taker side:

fills = client.fills(
    market=market,
    trader_id="0x1234...",
    taker_side="BUY",
    start_time="2026-03-15T00:00:00Z",
)

If you need typed fill models instead of a dataframe:

fill_models = client.fills_models(
    market=market,
    start_time="2026-03-15T00:00:00Z",
)

fill_models = market.fills_models(
    start_time="2026-03-15T00:00:00Z",
)

Query Orderbook Snapshots

Orderbook snapshots are available for accounts with orderbook access.

book = client.orderbook_snapshots(
    market=market,
    start_time="2026-03-15T00:00:00Z",
    end_time="2026-03-15T00:01:00Z",
)

book = market.orderbook_snapshots(
    start_time="2026-03-15T00:00:00Z",
    end_time="2026-03-15T00:01:00Z",
)

Each snapshot is a complete replacement book for one outcome. Rows include indexed_at, hash, state, continuity, and path_index in addition to the market identifiers, outcome, bids, asks, and source timestamp.

Use state="VERIFIED" when you only need books reconstructed without gaps. Use continuity separately to distinguish rows connected to the preceding published state from fresh starting points.

Choose Polars or pandas

All dataframe methods return Polars by default. Set frame="pandas" when creating the client to use pandas globally. pandas support is optional; install it with pip install "probalytics[pandas]".

client = ProbalyticsClient.from_clickhouse(
    host="clickhouse.probalytics.io",
    username="your_username",
    password="your_password",
    frame="pandas",
)

You can also override the frame for a single call.

fills_pd = client.fills(
    market=market,
    start_time="2026-03-15T00:00:00Z",
    frame="pandas",
)

Invalid frame values fail early with a clear error. Supported values are "polars" and "pandas".

Run Custom SQL

Use query() for read-only ClickHouse queries when the convenience methods do not cover your use case.

df = client.query(
    """
    SELECT platform, count() AS fills
    FROM fills
    WHERE timestamp >= %(start_time)s
    GROUP BY platform
    ORDER BY fills DESC
    """,
    parameters={"start_time": "2026-03-15T00:00:00Z"},
)

Always pass user-provided values through parameters instead of formatting them directly into SQL strings.

Supported Filters

markets() and markets_frame() support:

• start_time
• end_time
• status or list of statuses
• platform or list of platforms
• market_id or list of market IDs
• market_platform_id or list of market platform IDs
• limit
• max_rows

fills() and fills_models() support:

• start_time
• end_time
• platform or list of platforms
• market
• market_id or list of market IDs
• market_platform_id or list of market platform IDs
• taker_side or list of taker sides
• trader_id or list of trader IDs
• limit
• max_rows

orderbook_snapshots() supports:

• start_time
• end_time
• platform or list of platforms
• market
• market_id or list of market IDs
• market_platform_id or list of market platform IDs
• state or list of states
• continuity or list of continuity values
• limit

Local Development

uv sync --extra test
uv run --extra test pytest

Run the ClickHouse integration tests with Docker:

PROBALYTICS_RUN_INTEGRATION=1 uv run --extra test pytest tests/test_clickhouse_integration.py

Use PYTHONPATH=src if running tests without installing the package into the active environment.