How does
Rebel Fund
Use AI?
Achieved a net portfolio IRR of 44-59% on mature investments, outperforming the S&P 500
Project Overview
Predicting long-term success of Y Combinator startups using a proprietary dataset and non-linear machine learning models.
Layman's Explanation
The system uses AI to assess early-stage startups and predict whether they’ll grow big, stay small, or fail—helping investors make smarter bets.
Details
Rebel Theorem 3.0 is a non-linear machine learning algorithm developed by Rebel Fund to classify early-stage YC startups into outcome categories—"Success," "Zombie," or "Dead"—based on structured and inferred founder and company attributes. It uses a highly granular dataset comprising millions of data points, including founder education, past startup experience, inferred personality traits, and company metadata. The model leverages over 150 decision trees and accounts for nuanced interdependencies between variables, such as ideal co-founding experience combined with industry sector and personality alignment. Trained on historical YC data from 2012–2020, the model demonstrated backtested internal rates of return (IRRs) up to 59%, significantly outperforming both random investment strategies and public market benchmarks.
Analogy
It's like having an advanced fantasy football algorithm that drafts startup "players" by analyzing their personality profiles and weird correlations instead of only relying on their stats.
Machine Learning Techniques Used
More Use Cases in
Finance
Novelty Justification
ML-based outcome prediction for early-stage startups is innovative, with few at-scale deployments.
Project Estimates
