polymarketdata-sdk

Official Python SDK for the PolymarketData API — historical data for Polymarket prediction markets.

Installation

pip install polymarketdata-sdk

With optional pandas DataFrame helpers:

pip install "polymarketdata-sdk[dataframe]"

Quickstart

from polymarketdata import PolymarketDataClient

# api_key can also be set via the POLYMARKETDATA_API_KEY environment variable
with PolymarketDataClient(api_key="YOUR_API_KEY") as client:
    health = client.utility.health()
    print(health.status)          # "ok"

    usage = client.utility.usage()
    print(usage.plan)             # e.g. "pro"
    print(usage.limits.requests_remaining)  # requests left in the current window

Discovery

List and search markets

from polymarketdata import PolymarketDataClient

with PolymarketDataClient(api_key="YOUR_API_KEY") as client:
    # List the 10 most-recently-updated series
    series_page = client.discovery.list_series(limit=10)
    for s in series_page.data:
        print(s.id, s.slug, s.title)

    # Cursor-based next page
    next_cursor = series_page.metadata.next_cursor
    if next_cursor:
        next_page = client.discovery.list_series(limit=10, cursor=next_cursor)

    # Search within a series
    events_page = client.discovery.list_events(series_slug="us-elections-2024")
    for e in events_page.data:
        print(e.id, e.title, e.tags)

    # Markets for a specific event
    markets_page = client.discovery.list_markets(event_slug="us-presidential-election-2024")
    for m in markets_page.data:
        print(m.id, m.question)
        for token in (m.tokens or []):
            print(f"  token: {token.id} ({token.label})")

    # Fetch a single market by ID or slug
    market = client.discovery.get_market("will-trump-win-the-2024-presidential-election")
    print(market.market.question)
    print(market.market.status)   # "resolved", "open", etc.
    print(market.market.resolved_token_label)   # e.g. "Yes" if resolved

    # All available tags
    tags = client.discovery.list_tags()
    print(tags.data[:10])   # list[str]

Auto-paginating iterators

The iter_* methods handle cursor pagination automatically, yielding one item at a time across as many pages as needed:

from polymarketdata import PolymarketDataClient

with PolymarketDataClient(api_key="YOUR_API_KEY") as client:
    # Stream every series (or stop early with max_items)
    for series in client.discovery.iter_series(search="election", max_items=50):
        print(series.slug, series.title)

    # Nested discovery: series → events → markets
    for series in client.discovery.iter_series(tags=["politics"]):
        for event in client.discovery.iter_events(series_slug=series.slug):
            for market in client.discovery.iter_markets(event_slug=event.slug):
                print(f"{series.title} / {event.title} / {market.question}")

History

Historical data requires a time range (start_ts, end_ts) and a resolution. Timestamps can be passed as Unix ints, ISO-8601 strings, or datetime objects.

Resolutions

Enum	String	Interval
Resolution.ONE_MINUTE	"1m"	1 minute
Resolution.FIVE_MINUTES	"5m"	5 minutes
Resolution.FIFTEEN_MINUTES	"15m"	15 minutes
Resolution.ONE_HOUR	"1h"	1 hour
Resolution.SIX_HOURS	"6h"	6 hours
Resolution.ONE_DAY	"1d"	1 day

Market metrics (volume, liquidity, spread)

import time
from polymarketdata import PolymarketDataClient, Resolution

with PolymarketDataClient(api_key="YOUR_API_KEY") as client:
    result = client.history.get_market_metrics(
        "will-trump-win-the-2024-presidential-election",
        start_ts="2024-01-01T00:00:00Z",
        end_ts="2024-11-05T00:00:00Z",
        resolution=Resolution.ONE_DAY,
    )
    print(f"market_id={result.market_id}, resolution={result.resolution}")
    for dp in result.data:
        print(f"  t={dp.t}  volume={dp.volume:.2f}  liquidity={dp.liquidity:.2f}  spread={dp.spread:.4f}")

    # Iterate all data points across pages without manual cursor management
    for dp in client.history.iter_market_metrics(
        "will-trump-win-the-2024-presidential-election",
        start_ts="2024-01-01T00:00:00Z",
        end_ts=int(time.time()),
        resolution=Resolution.ONE_DAY,
    ):
        print(dp.t, dp.volume)

Token prices

Each Polymarket market has two tokens (e.g. "Yes" and "No"). The market-level endpoint returns prices for all tokens keyed by label; the token-level endpoint returns a single price series.

from polymarketdata import PolymarketDataClient, Resolution

with PolymarketDataClient(api_key="YOUR_API_KEY") as client:
    # All tokens for a market
    result = client.history.get_market_prices(
        "will-trump-win-the-2024-presidential-election",
        start_ts="2024-10-01T00:00:00Z",
        end_ts="2024-11-05T00:00:00Z",
        resolution=Resolution.ONE_HOUR,
    )
    print(result.tokens)  # {"Yes": "<token_id>", "No": "<token_id>"}
    for label, points in result.data.items():
        print(f"{label}: {len(points)} data points")
        for dp in points[:3]:
            print(f"  t={dp.t}  p={dp.p:.4f}")

    # Single token — useful for streaming a specific outcome's price history
    yes_token_id = result.tokens["Yes"]
    single = client.history.get_token_prices(
        yes_token_id,
        start_ts="2024-10-01T00:00:00Z",
        end_ts="2024-11-05T00:00:00Z",
        resolution=Resolution.ONE_HOUR,
    )
    print(f"token_label={single.token_label}")
    for dp in single.data:
        print(f"  t={dp.t}  p={dp.p:.4f}")

    # Auto-paginating iterator for a single token
    for dp in client.history.iter_token_prices(
        yes_token_id,
        start_ts="2024-01-01T00:00:00Z",
        end_ts="2024-11-05T00:00:00Z",
        resolution=Resolution.ONE_DAY,
    ):
        print(dp.t, dp.p)

Order book snapshots

Order books are returned as lists of [price, size] pairs.

from polymarketdata import PolymarketDataClient, Resolution

with PolymarketDataClient(api_key="YOUR_API_KEY") as client:
    result = client.history.get_market_books(
        "will-trump-win-the-2024-presidential-election",
        start_ts="2024-10-01T00:00:00Z",
        end_ts="2024-10-02T00:00:00Z",
        resolution=Resolution.ONE_HOUR,
    )
    for label, snapshots in result.data.items():
        print(f"Token: {label}  ({len(snapshots)} snapshots)")
        for snap in snapshots[:2]:
            print(f"  t={snap.t}")
            print(f"  bids (top 3): {snap.bids[:3]}")   # [[price, size], ...]
            print(f"  asks (top 3): {snap.asks[:3]}")

    # Iterate all book snapshots for a single token
    yes_token_id = result.tokens["Yes"]
    for snap in client.history.iter_token_books(
        yes_token_id,
        start_ts="2024-10-01T00:00:00Z",
        end_ts="2024-11-05T00:00:00Z",
        resolution=Resolution.ONE_HOUR,
    ):
        best_bid = snap.bids[0][0] if snap.bids else None
        best_ask = snap.asks[0][0] if snap.asks else None
        print(f"t={snap.t}  bid={best_bid}  ask={best_ask}")

DataFrame helpers

If you have pandas installed (pip install "polymarketdata-sdk[dataframe]"), you can convert responses directly to DataFrames. Timestamps are automatically parsed to timezone-aware datetime64[ns, UTC]:

from polymarketdata import PolymarketDataClient, Resolution
from polymarketdata.dataframe import to_dataframe_metrics, to_dataframe_prices, to_dataframe_books

with PolymarketDataClient(api_key="YOUR_API_KEY") as client:
    metrics_resp = client.history.get_market_metrics(
        "will-trump-win-the-2024-presidential-election",
        start_ts="2024-01-01T00:00:00Z",
        end_ts="2024-11-05T00:00:00Z",
        resolution=Resolution.ONE_DAY,
    )
    df = to_dataframe_metrics(metrics_resp.data)
    print(df.dtypes)
    # t             datetime64[ns, UTC]
    # volume                    float64
    # liquidity                 float64
    # spread                    float64
    print(df.head())

    prices_resp = client.history.get_market_prices(
        "will-trump-win-the-2024-presidential-election",
        start_ts="2024-10-01T00:00:00Z",
        end_ts="2024-11-05T00:00:00Z",
        resolution=Resolution.ONE_HOUR,
    )
    # to_dataframe_prices accepts either the full response or a list of PriceDataPoint
    yes_df = to_dataframe_prices(prices_resp.data["Yes"])
    print(yes_df.head())

Error handling

All SDK errors inherit from PolymarketDataError:

from polymarketdata import (
    PolymarketDataClient,
    AuthenticationError,
    NotFoundError,
    RateLimitError,
    PolymarketDataError,
)

with PolymarketDataClient(api_key="YOUR_API_KEY") as client:
    try:
        market = client.discovery.get_market("nonexistent-market-slug")
    except NotFoundError as e:
        print(f"404 — not found: {e.detail}")
    except AuthenticationError:
        print("Check your API key")
    except RateLimitError as e:
        print(f"Rate limited — status {e.status_code}, request_id={e.request_id}")
    except PolymarketDataError as e:
        print(f"SDK error {e.status_code}: {e.detail}")

Exception hierarchy

Exception	HTTP status
BadRequestError	400
AuthenticationError	401
PermissionDeniedError	403
NotFoundError	404
RateLimitError	429
ServerError	5xx
NetworkError	— (network failure)
RequestTimeoutError	— (timeout)

Retry configuration

The client retries on 429 Too Many Requests and 5xx errors with exponential back-off + jitter. Adjust via the constructor:

from polymarketdata import PolymarketDataClient

client = PolymarketDataClient(
    api_key="YOUR_API_KEY",
    max_retries=4,           # default: 2
    retry_backoff_base=1.0,  # default: 0.5 s
    retry_backoff_max=30.0,  # default: 8.0 s
    timeout=60.0,            # default: 30.0 s
)

Rate-limit awareness

from polymarketdata import PolymarketDataClient

with PolymarketDataClient(api_key="YOUR_API_KEY") as client:
    usage = client.utility.usage()
    print(f"Plan: {usage.plan}")
    print(f"Requests/min allowed: {usage.limits.requests_per_minute}")
    print(f"Requests remaining:   {usage.limits.requests_remaining}")
    print(f"Max history days:     {usage.limits.max_history_days}")
    print(f"Resets at:            {usage.reset_at}  (Unix timestamp)")

HTTP metadata

Every response carries .meta (HTTP metadata) and .raw (unmodified response body):

from polymarketdata import PolymarketDataClient

with PolymarketDataClient(api_key="YOUR_API_KEY") as client:
    result = client.utility.health()
    if result.meta:
        print(result.meta.status_code)    # 200
        print(result.meta.request_id)     # "abc-123"
    print(result.raw)                      # RawPayload with all raw fields

Development

# Install all dev deps
uv sync --group dev

# Run the full check suite
uv run ruff check .
uv run mypy src scripts
uv run pytest

# Run the live smoke test (requires POLYMARKETDATA_API_KEY)
uv run python scripts/smoke_test.py

# Build a distribution
uv build

OpenAPI model generation

Pydantic models are auto-generated from the pinned OpenAPI snapshot at openapi/openapi.json.

# Regenerate after updating openapi.json
uv run python scripts/update_openapi.py

# Check for drift (also runs in CI)
uv run python scripts/update_openapi.py --check

License

Apache 2.0 — see LICENSE.