This dataset requires a paid license. Please purchase access at the link below and provide your Hugging Face username. Once purchased, submit an access request on this page. Access is granted manually.

Polymarket Prediction Market Data for Sports and Event Research

This dataset packages public Polymarket market metadata into a single analysis-ready table for prediction-market research, sports-betting workflow support, content production, and dashboard building. It is designed for buyers who do not want to page the public Gamma API, flatten nested market structures, normalize event context, or derive league and market-style features by hand before they can do useful work.

The current release contains 28,364 rows across 96 columns. Each row represents one Polymarket market enriched with its parent event metadata, market status flags, volume and liquidity fields, outcome-price snapshots, tags, series information, sportsbook-style text heuristics, and source links back to the official public surface. Coverage in this build runs from 2020-10-02 through 2029-01-20, and the table intentionally includes both active and closed events so a buyer can compare live markets with recent historical structure in the same file.

The target buyer is specific. A prediction-market researcher can use this file to compare category mix, liquidity, and market density across sports, politics, crypto, and culture. A sports-betting analyst can filter player props, spreads, totals, and moneyline-style questions without starting from raw nested JSON. A dashboard builder can load the parquet file directly into pandas, DuckDB, or a warehouse and build screens around event counts, active-versus-closed status, tag families, or volume trends. A content operator or newsletter writer can scan the data to identify crowded market themes, multi-market events, and segments worth covering. These are concrete workflows, not vague "insight generation" promises.

The value of the dataset is not that the underlying source is secret. The value is the packaging. Polymarket exposes a rich public surface, but the raw API is inconvenient for commercial research use because events and markets arrive as nested structures, active and closed coverage must be reconciled, fields vary in completeness, and a buyer often needs derived labels before the data is pleasant to explore. This product turns that collection and cleaning work into a reusable research asset. The output is already column-named, typed, deduplicated, and aligned to a consistent schema with matching column descriptions.

The file is ready for everyday tools. You can load it in pandas for notebook work, query it with DuckDB for fast filtering, use it in BI dashboards, or join it with your own schedules, portfolios, watchlists, and event annotations. Because the format is parquet, it is efficient to store and straightforward to process. Buyers who want to validate fit before purchasing can inspect the free Hugging Face sample, confirm the schema, and test loading behavior without authentication.

The free sample is intentionally proof-first. It demonstrates that the dataset is real, public, viewer-compatible, non-empty, and aligned to the same schema as the paid version. The preview is not a watered-down export with fake columns. It is a proportional slice of the exact structure a buyer receives after purchase, which makes it easier to validate whether the category, pricing, liquidity, and event fields match the intended workflow.

Limitations are important. This dataset does not promise profitable trading, betting success, or edge on any specific event. It reflects the public Polymarket surface at collection time, so field completeness depends on what the platform exposed in the API for each market. Some heuristic columns, such as sports or player-prop signals, are derived from titles and tags and should be treated as workflow helpers rather than as official source labels. The dataset is best used for research, comparison, monitoring, and development.

The full product adds more rows, more coverage, and less manual setup than the free preview. Instead of validating a few hundred example records and then rebuilding the rest yourself, you receive the complete analysis-ready table plus a launch surface that already points to the paid checkout, the public sample, and bundle-adjacent related datasets in the same vertical.

Access Requirements (Paid Dataset)

This dataset is behind manual gated access.

To obtain access:

Purchase the dataset here:
https://thearticulated.gumroad.com/l/dwwemm?utm_source=hf_paid_readme&utm_medium=referral&utm_campaign=polymarket-prediction-markets-enriched
Provide your Hugging Face username at checkout.
Return to this Hugging Face page and click:
"Request Access"
Your access will be approved within 1-12 hours.

Once approved, you can use the Python snippet at the bottom of this README to load the dataset.

Related Datasets

Dataset Structure

Total rows: 28,364

Total columns: 96

Splits

data: 28,364 rows

Data Files

data: data/data.parquet

Data Dictionary

The table below describes the columns included in this dataset.

column	pandas_dtype	dataset_type	description
record_id	object	string	Text column. Stable unique identifier for the dataset row derived from the Polymarket market id.
market_id	object	integer-like string	Stored as text, but values appear to represent whole numbers. Polymarket market identifier from the public Gamma API.
condition_id	object	string	Text column. On-chain condition identifier associated with the binary market.
market_slug	object	string	Text column. Public slug string for the market in the Gamma API.
question	object	string	Text column. Primary market question text shown to traders.
question_word_count	Int64	integer	Whole-number numeric column. Word-count length of the market question after normalization.
outcomes	object	string	Text column. JSON-array string containing the ordered list of outcome labels returned by the market payload.
outcome_count	Int64	integer	Whole-number numeric column. Number of outcome labels available for the market.
yes_price	Float64	float	Decimal numeric column. Implied Yes-side price when the market exposes Yes and No outcomes.
no_price	Float64	float	Decimal numeric column. Implied No-side price when the market exposes Yes and No outcomes.
favorite_outcome	object	string	Text column. Outcome label with the highest exposed price at collection time when prices were available.
favorite_price	Float64	float	Decimal numeric column. Highest exposed outcome price at collection time when prices were available.
price_sum	Float64	float	Decimal numeric column. Sum of all parsed outcome prices for the market snapshot.
market_spread	Float64	float	Decimal numeric column. Difference between best ask and best bid when those values were available.
last_trade_price	Float64	float	Decimal numeric column. Last trade price reported by the public market payload.
best_bid	Float64	float	Decimal numeric column. Current best bid from the public market payload.
best_ask	Float64	float	Decimal numeric column. Current best ask from the public market payload.
one_day_price_change	Float64	float	Decimal numeric column. One-day price change reported by the market payload.
one_week_price_change	Float64	float	Decimal numeric column. One-week price change reported by the market payload.
one_month_price_change	Float64	float	Decimal numeric column. One-month price change reported by the market payload.
one_year_price_change	Float64	float	Decimal numeric column. One-year price change reported by the market payload.
volume_usd	Float64	float	Decimal numeric column. Cumulative market volume reported in dollar terms by the public payload.
liquidity_usd	Float64	float	Decimal numeric column. Current market liquidity reported in dollar terms by the public payload.
volume_24h_usd	Float64	float	Decimal numeric column. Trailing 24-hour market volume in dollars when available.
volume_1wk_usd	Float64	float	Decimal numeric column. Trailing 1-week market volume in dollars when available.
volume_1mo_usd	Float64	float	Decimal numeric column. Trailing 1-month market volume in dollars when available.
volume_1yr_usd	Float64	float	Decimal numeric column. Trailing 1-year market volume in dollars when available.
order_min_size	Float64	float	Decimal numeric column. Minimum order size reported by the market payload when available.
order_price_min_tick_size	Float64	float	Decimal numeric column. Minimum tick size for the order book when available.
market_status	object	string	Text column. Normalized market-status label derived from archived, closed, active, and accepting-orders flags.
market_active	bool	boolean	True/false column. Boolean flag indicating the market payload marked the market as active.
market_closed	bool	boolean	True/false column. Boolean flag indicating the market payload marked the market as closed.
market_archived	bool	boolean	True/false column. Boolean flag indicating the market payload marked the market as archived.
market_restricted	bool	boolean	True/false column. Boolean flag indicating the payload marked the market as restricted.
market_featured	bool	boolean	True/false column. Boolean flag indicating the market payload marked the market as featured.
accepting_orders	bool	boolean	True/false column. Boolean flag indicating whether the market was accepting orders at collection time.
enable_order_book	bool	boolean	True/false column. Boolean flag indicating whether the market exposed order-book support.
fees_enabled	bool	boolean	True/false column. Boolean flag indicating whether fees were enabled on the market.
neg_risk	bool	boolean	True/false column. Boolean flag indicating whether the market payload flagged negative-risk support.
competitive_score	Float64	float	Decimal numeric column. Competitive score value reported by the market payload when available.
created_at	datetime64[ns, UTC]	datetime	Date or timestamp column. Timestamp when the market was created according to the market payload.
updated_at	datetime64[ns, UTC]	datetime	Date or timestamp column. Timestamp when the market was last updated according to the market payload.
start_date	datetime64[ns, UTC]	datetime	Date or timestamp column. Market start timestamp when available.
end_date	datetime64[ns, UTC]	datetime	Date or timestamp column. Market end timestamp when available.
closed_time	datetime64[ns, UTC]	datetime	Date or timestamp column. Timestamp when the market closed when exposed by the payload.
days_since_creation	Int64	integer	Whole-number numeric column. Number of days between market creation and collection time.
days_since_update	Float64	float	Decimal numeric column. Number of days between market last update and collection time.
days_until_end	Float64	float	Decimal numeric column. Number of days between collection time and the scheduled market end timestamp.
event_id	object	integer-like string	Stored as text, but values appear to represent whole numbers. Parent Polymarket event identifier.
event_ticker	object	string	Text column. Parent Polymarket event ticker when available.
event_slug	object	string	Text column. Parent Polymarket event slug.
event_title	object	string	Text column. Parent Polymarket event title.
event_category	object	string	Text column. Event-level category string from Polymarket when available.
event_active	bool	boolean	True/false column. Boolean flag indicating the parent event was marked active.
event_closed	bool	boolean	True/false column. Boolean flag indicating the parent event was marked closed.
event_archived	bool	boolean	True/false column. Boolean flag indicating the parent event was marked archived.
event_restricted	bool	boolean	True/false column. Boolean flag indicating the parent event was marked restricted.
event_comment_count	Int64	integer	Whole-number numeric column. Comment count reported on the parent event when available.
event_market_count	Int64	integer	Whole-number numeric column. Number of market rows attached to the parent event payload.
event_volume_usd	Float64	float	Decimal numeric column. Cumulative event volume reported in dollars.
event_liquidity_usd	Float64	float	Decimal numeric column. Current event liquidity reported in dollars.
event_open_interest_usd	Float64	float	Decimal numeric column. Open interest reported on the event when available.
event_created_at	datetime64[ns, UTC]	datetime	Date or timestamp column. Timestamp when the event record was created on Polymarket.
event_updated_at	datetime64[ns, UTC]	datetime	Date or timestamp column. Timestamp when the event record was last updated on Polymarket.
event_start_date	datetime64[ns, UTC]	datetime	Date or timestamp column. Parent event start timestamp when available.
event_end_date	datetime64[ns, UTC]	datetime	Date or timestamp column. Parent event end timestamp when available.
series_titles	object	string	Text column. JSON-array string containing the list of series titles attached to the parent event.
series_slugs	object	string	Text column. JSON-array string containing the list of series slugs attached to the parent event.
primary_series_title	object	string	Text column. First attached series title when available.
primary_series_slug	object	string	Text column. First attached series slug when available.
tag_labels	object	string	Text column. JSON-array string containing the list of event tag labels attached to the parent event.
tag_slugs	object	string	Text column. JSON-array string containing the list of event tag slugs attached to the parent event.
tag_count	Int64	integer	Whole-number numeric column. Number of event tags attached to the parent event.
is_multi_market_event	bool	boolean	True/false column. Boolean flag indicating the parent event contains more than one market.
is_binary_market	bool	boolean	True/false column. Boolean flag indicating the market exposed exactly two outcomes.
is_sports_market	bool	boolean	True/false column. Boolean flag indicating sports-oriented event, series, or title language.
sports_league	object	string	Text column. Normalized sports league label derived from series, tags, category, and title heuristics when detected.
is_crypto_market	bool	boolean	True/false column. Boolean flag indicating crypto-related market language or category.
is_politics_market	bool	boolean	True/false column. Boolean flag indicating politics or election language or category.
is_current_affairs_market	bool	boolean	True/false column. Boolean flag indicating current-affairs style categorization.
is_pop_culture_market	bool	boolean	True/false column. Boolean flag indicating entertainment or pop-culture language.
has_player_prop_signal	bool	boolean	True/false column. Boolean flag indicating a player-prop style betting market.
has_spread_signal	bool	boolean	True/false column. Boolean flag indicating spread-style language.
has_total_signal	bool	boolean	True/false column. Boolean flag indicating totals or over-under language.
has_moneyline_signal	bool	boolean	True/false column. Boolean flag indicating moneyline-style win or beat language.
has_price_target_signal	bool	boolean	True/false column. Boolean flag indicating price-target or threshold-style language.
has_election_signal	bool	boolean	True/false column. Boolean flag indicating election-oriented language.
has_award_signal	bool	boolean	True/false column. Boolean flag indicating award-oriented language.
has_culture_signal	bool	boolean	True/false column. Boolean flag indicating entertainment, music, movie, or release-oriented language.
resolution_source_domain	object	string	Text column. Domain extracted from the market resolution source URL when available.
source_api_events_url	object	string	Text column. Base events endpoint used to collect the source data.
source_api_event_url	object	string	Text column. Event-specific public API URL for the parent event.
source_api_market_url	object	string	Text column. Market-specific public API URL for the market.
source_directory_url	object	string	Text column. Top-level Polymarket website URL for the source platform.
source_domain	object	string	Text column. Source platform domain for the dataset row.
last_collected_at	datetime64[ns, UTC]	datetime	Date or timestamp column. UTC timestamp when the row was collected for this dataset build.

Intended Use

This dataset is intended for research, experimentation, analysis, and model prototyping.

Loading the Dataset


import os
from datasets import load_dataset

HUGGINGFACE_API_KEY_KARMANE = os.environ.get("HUGGINGFACE_API_KEY_KARMANE")

dataset = load_dataset(
    "Karmane/polymarket-prediction-markets-enriched",
    token=HUGGINGFACE_API_KEY_KARMANE,
)

print(dataset)
print(dataset[list(dataset.keys())[0]][0])

# getting the DataFrame itself
# df = dataset[list(dataset.keys())[0]].to_pandas()

Karmane. (2025). Polymarket Prediction Market Data for Sports and Event Research. Hugging Face. https://huggingface.co/datasets/Karmane/polymarket-prediction-markets-enriched

Downloads last month: 188

Total file size:

6.05 MB