Probalytics Probalytics
Back to home
For Researchers & Data Scientists

Cross-platform
prediction market analytics

Compare forecasting accuracy across Polymarket, Kalshi & PredictIt. One unified dataset, no data wrangling. Query with SQL, export to Parquet.

Be first to know when we launch. No spam, ever.

150K+

Markets indexed

5+

Years of history

3

Platforms unified

12B+

USD volume tracked

The Power of Unified Data

Which platform is better calibrated?

Run one query to find out. No manual data cleaning, no ETL pipelines, no schema mapping. We've already normalized everything.

  • Same events tracked across platforms with unified IDs
  • Resolution outcomes normalized for accurate comparison
  • Cross-platform price history at every timestamp
  • Event metadata and categorization included
Calibration ComparisonSample data
Polymarket0.92 accuracy
Kalshi0.89 accuracy
PredictIt0.85 accuracy

Cross-platform calibration comparison - one query, instant results

SQL Access

Research query examples

Direct SQL access to cross-platform data via ClickHouse

Calibration by platform

Compare forecasting accuracy across Polymarket, Kalshi, and PredictIt

SELECT platform,
  FLOOR(probability * 10) / 10 as bucket,
  AVG(resolved_yes) as actual_rate,
  COUNT(*) as n
FROM unified_markets
WHERE resolved_at IS NOT NULL
GROUP BY platform, bucket

Data Access

Export in your preferred format

Parquet

For Pandas, Spark, DuckDB

CSV

For Excel, R, general use

SQL

Direct ClickHouse queries

API

REST & streaming endpoints

Use Cases

What researchers build with Probalytics

Forecasting Calibration

Study how well prediction markets are calibrated. Do 70% predictions resolve "yes" 70% of the time?

Market Efficiency

Analyze how quickly markets incorporate new information. Compare across platforms.

Crowd Wisdom

Research information aggregation dynamics. How do prediction markets outperform polls?

Cross-Platform Comparison

Compare the same events across Polymarket, Kalshi, PredictIt. Which is most accurate?

Python SDK

Cross-platform analysis in Python

Run SQL queries directly from Python. Get results as DataFrames. No data wrangling required.

  • Direct SQL queries to ClickHouse
  • Results as pandas DataFrames
  • Async support for large queries
research.py
from probalytics import Probalytics

client = Probalytics(api_key="pk_...")

# Compare calibration ACROSS platforms - one query
df = client.sql("""
  SELECT
    platform,
    FLOOR(probability * 10) / 10 as bucket,
    AVG(resolved_yes) as actual_rate,
    COUNT(*) as n
  FROM unified_markets
  WHERE resolved_at IS NOT NULL
  GROUP BY platform, bucket
  ORDER BY platform, bucket
""")

# Instant cross-platform comparison!
print(df.pivot(index='bucket', columns='platform'))
Access the Dataset

Ready to start your research?

Join the waitlist for early access to the most comprehensive cross-platform prediction market dataset.

Be first to know when we launch. No spam, ever.