top of page

PREDICT. ANALYZE. DEPLOY.

The Prediction Engine

Cutting-edge research, made accessible

Python

R

Your Tools

CSA prediction engine

yhat

Composite prediction

fit

task-specific reliability

insights.relevance

observation relevance

by variable

impact of fit / prediction

yhat_linear • fit • adjusted_fit • kfit • weight • relevance • similarity agreement • asymmetry • info_theta • info_x • contribution_to_prediction • contribution_to_conviction • component_contribution_to_prediction • outlier_influence • r_star • r_star_percent • ysolo_sigma • ysolo_skewness • ysolo_kurtosis

01  THE METHOD

A new way to reason about uncertainty. 

Relevance-Based Prediction (RBP) is a novel, model-free prediction methodology developed by the founders of Cambridge Sports Analytics. Grounded in information theory, RBP treats every prediction as a distinct information problem, weighting historical observations according to their relevance to the prediction at hand.

Most prediction systems deliver a number. RBP reveals the reasoning behind it.

By making the formation of each prediction transparent, RBP reveals which historical observations matter, how much they contribute, which variables are most important, and how much confidence to place in the result.

No black boxes. Just transparent, explainable predictions.

  01

Advanced insights

Model-free predictions that weight historical observations by their relevance, grounded in the principles of information theory.

  02

Transparent by design

See which historical observations and variables matter, how they influence each prediction, and how much confidence to place in the result.

  03

Flexible Deployment

Integrate RBP into your analytics environment through licensed software, a hosted API, or locally deployed solutions.

Multiple solutions,
one
engine.

data.csv

prediction lab

Add your data

/01  ANALYST

Lab

Predict. Explore. Understand.

An interactive, no-code environment for Relevance-Based Prediction. Use your own data to generate predictions, explore the factors that shape each result, and assess its reliability.

Generate predictions without writing code

Assess reliability with prediction-specific fit

Explore which variables influence each prediction

Examine the most relevant historical observations

rbp-engine

python

R

Your Tools 

/02  ENTERPRISE

Engine

$ rbp-engine  --license <key>

Deploy the Engine directly within your own computing environment. Integrate RBP into existing workflows while maintaining complete control over your data and infrastructure.

Single compiled binary with no runtime dependencies

Integrate with Python, R, and custom applications

Keep your data secure in your own computing environment

High-performance computing without API rate limits

Built for professionals.

02 SOLUTIONS

/01

Quantitative researchers

Develop and evaluate predictions using the Python SDK. Examine relevance weights, test hypotheses, and assess prediction reliability with complete transparency into the underlying calculations.

/02

Engineering teams in regulated industries

Deploy the licensed Engine within your own computing environment, from individual workstations to enterprise servers. Maintain control over data security, privacy, and infrastructure while integrating advanced prediction capabilities into existing systems.

/03

Analyst & domain experts

Generate predictions through the Lab’s interactive, no-code interface. Explore alternative prediction scenarios, identify influential variables, and evaluate the reliability of individual predictions.

/04

Sports & media organizations

Deliver transparent, evidence-based forecasts that engage and inform audiences. Explain which historical observations drive each prediction and provide meaningful context for interpreting the results.

03 IN PRACTICE

Your First RBP Prediction in Just Three Lines of Code.

The Python SDK makes it easy to integrate Relevance-Based Prediction into your analytical workflow. Provide your data and receive predictions with complete transparency into which historical observations matter and how much each contributes to the result.

predict.py - Python 3.11

from rbp_engine import predict_grid, GridOption, PredictionResults

# Submit y, X, theta - predictions come back with full attribution
yaht, output_details = predict_grid(y=y, X=X, theta=theta, options=GridOptions())

# Every insight packaged in one results object
Results = PredictionResults(output_details)

Start building with the Prediction Engine

©2026 Cambridge Sports Analytics, a division of Cambridge Prediction Analytics, LLC. All rights reserved.      

Cambridge, Massachusetts

bottom of page