on this page

SCOTUS Predict v2

model

An enhanced Supreme Court prediction model achieving 70.2% accuracy across 200 years of decisions (1816-2015), analyzing over 240,000 justice votes

period: 2016-2017
tech:
Legal AnalyticsMachine Learning
โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•โ•

The second version of the Supreme Court prediction model, significantly expanded to cover nearly 200 years of Supreme Court history. Published in PLoS ONE, this work represents a major advance in quantitative legal prediction.

Key Improvements

  • Extended coverage from 60 years to 200 years (1816-2015)
  • Analyzed 240,000+ justice votes and 28,000+ case outcomes
  • 70.2% accuracy at the case outcome level
  • 71.9% accuracy at the justice vote level
  • Outperforms baseline models by nearly 5% across the past century

Technical Innovation

The model employs a time-evolving random forest classifier with sophisticated feature engineering:

  • Temporal adaptation to account for changing legal landscapes
  • Generalized approach applicable to any time period
  • Out-of-sample validation ensuring robust predictions
  • Uses only pre-decision data for all predictions

Research Impact

Published as: Katz DM, Bommarito MJ II, Blackman J (2017) A general approach for predicting the behavior of the Supreme Court of the United States. PLoS ONE 12(4): e0174698

The model and data are fully open source, enabling:

  • Reproducible research
  • Further academic study
  • Practical applications in legal analytics

Authors

Developed in collaboration with:

  • Daniel Martin Katz (Illinois Tech - Chicago Kent College of Law)
  • Josh Blackman (South Texas College of Law Houston)
on this page