ImpliedData/prediction-markets

Prediction Markets Dataset (Sample)

ImpliedData is the cross-platform data layer for prediction markets: it matches the same real-world event across Polymarket, Manifold & Myriad — polarity-aligned so YES means the same thing on every venue — on top of the Polymarket on-chain tick history (404M fills, 2023–present).

This is a free sample: market metadata and 1-hour OHLCV candles from Polymarket and Manifold. Prices are implied probabilities (0–1): a close of 0.73 means the market priced the event at 73% probability at that hour. The cross-platform event links, intraday candles, and full history are in the complete dataset at implieddata.com.

Files

File	Rows	Description
markets_metadata.parquet	231,662	Market catalog — titles, categories, statuses, dates (Polymarket + Manifold)
polymarket_ohlcv_1h_sample.parquet	14,021	1h OHLCV — 70 top-volume Polymarket markets, 2026-05-05 → 2026-05-19
manifold_ohlcv_1h_sample.parquet	26,168	1h OHLCV — 60 top-volume Manifold markets, 2025-11-19 → 2026-05-16
sample_markets_lookup.parquet	130	Join table: candle market_id → title, platform, category, status, resolution

No Kalshi price or trade data is included in this release.

Schema

markets_metadata.parquet

Column	Type	Description
platform_market_id	string	Platform-native identifier
title	string	Market question text
platform	string	polymarket or manifold
category	string	Topic category (politics, crypto, sports, …; sparse)
status	string	open, resolved, closed, or cancelled
resolution_result	string	Outcome if resolved
opened	date	Market creation date
closes	date	Resolution date, if known
trader_count	int	Unique traders (null for a large share of markets)

*_ohlcv_1h_sample.parquet

One row per market per hour: timestamp (UTC, candle open), platform_id (1 = Polymarket, 3 = Manifold), market_id, open, high, low, close, volume, trade_count.

sample_markets_lookup.parquet

Joins the candle files to human-readable market info: market_id, platform_market_id, platform, title, category, status, resolution (Yes/No when resolved), opened, closes.

Loading the data

import pandas as pd

candles = pd.read_parquet("polymarket_ohlcv_1h_sample.parquet")
lookup  = pd.read_parquet("sample_markets_lookup.parquet")
df = candles.merge(lookup[["market_id", "title", "status", "resolution"]], on="market_id")

# Price path of the most-traded market in the sample
top_id = df.groupby("market_id")["volume"].sum().idxmax()
m = df[df.market_id == top_id].sort_values("timestamp")
print(m[["timestamp", "title", "close", "volume"]].tail(10))

# Via the HuggingFace datasets library
from datasets import load_dataset
ds = load_dataset("ImpliedData/prediction-markets", data_files="polymarket_ohlcv_1h_sample.parquet")

Why ImpliedData

Single-platform prediction-market data is a commodity. The hard part is knowing that a market on Polymarket and a differently-worded one on Manifold are the same event, and aligning them so YES means the same thing on both — ~1,800 links across ~1,360 events, matched by embeddings + LLM review and polarity-aligned. Read the research: real money vs. play money · how the event links work.

Notes & limitations (read before using)

• This is a sample: 130 top-volume markets, 1h resolution only, limited windows (Polymarket: 2 weeks; Manifold: ~6 months). Low-volume markets are not included.
• Manifold uses play money (Mana) — prices are crowd probability estimates, but the incentive structure differs from real-money markets. Volume is denominated in Mana, not USD.
• category and trader_count are null for a significant share of markets.
• The metadata file is a snapshot (May 2026); the live ImpliedData catalog has grown since.

What the full dataset adds

The live ImpliedData dataset: markets across Polymarket, Manifold & Myriad, the Polymarket on-chain tick history (404M fills, 2023–present; metadata back to 2020), ~1,800 cross-platform event links, intraday candles (1m / 5m / 15m / 1h / 1d), resolved-outcome labels, and daily updates — exportable as Parquet or CSV per platform, category, or individual market.

→ implieddata.com — browse the catalog and download free samples of any market.

Citation

@dataset{implieddata2026,
  title     = {ImpliedData Prediction Markets Dataset},
  author    = {ImpliedData},
  year      = {2026},
  url       = {https://huggingface.co/datasets/ImpliedData/prediction-markets},
  note      = {Cross-platform historical prediction-market data (Polymarket, Manifold, Myriad)}
}

License

Free for non-commercial research (CC BY-NC 4.0 terms). Underlying market data originates from Polymarket and Manifold Markets and is subject to each platform's terms of service. The schema, documentation, and normalisation methodology are the work of ImpliedData. Attribution: "Data: ImpliedData (implieddata.com)". Commercial licensing → [email protected].