Academic Research with Polymarket Historical Order Book Data
How researchers and academics can use Polymarket historical data for market microstructure and prediction market studies.
Why Polymarket Data Matters for Research
Prediction markets are a rich source of data for studying information aggregation, market microstructure, and behavioral finance. PolyHistorical provides the granular order book data that most academic studies require but rarely have access to.
Research Areas
| Field | Research Questions | Data Needed |
|---|---|---|
| Market Microstructure | How do spreads form? What drives depth? | Order book snapshots |
| Information Efficiency | How fast do markets incorporate news? | Time-series prices + events |
| Behavioral Finance | Do biases exist in probability pricing? | Historical prices near resolution |
| Forecasting Accuracy | Do market probabilities match outcomes? | Prices + resolution data |
| Liquidity Studies | What determines market maker participation? | Order book depth over time |
Data Access for Academics
PolyHistorical's free Starter tier provides enough data for initial research and paper prototyping. For comprehensive studies requiring full historical depth:
- Pro plan ($11/mo): Unlimited history for BTC, ETH, and SOL markets
- Bulk export: Download datasets in CSV/JSON for offline analysis in R, Stata, or Python
- API access: Programmatic queries for reproducible research pipelines
Example: Calibration Study
# Fetch resolved markets and compare final price to outcome
import requests
resp = requests.get(
"https://api.polyhistorical.com/v1/markets",
headers={"X-API-Key": "your_key"},
params={"resolved": "true", "coin": "BTC", "limit": "100"}
)
for market in resp.json()["data"]:
final_up = float(market["price_up"])
winner = market["winner"] # "Up" or "Down"
actual = 1.0 if winner == "Up" else 0.0
print(f"Predicted: {final_up:.2f}, Actual: {actual}")Citation
When using PolyHistorical data in publications, please cite the data source and API version used. Consistent data provenance strengthens reproducibility.