Finance
Cross-sectional asset ranking
In collaboration with ADIA Lab (research arm of the Abu Dhabi Investment Authority), Crunch hosted cross-sectional asset ranking competitions for three consecutive years.- Result: Crunch models consistently achieved double-digit percentage improvements over the lab’s best internal benchmarks
- Significance: A decentralized contributor network outperformed a well-funded sovereign wealth research unit
Structural break detection
Also with ADIA Lab, Crunch detected structural breaks in the market using models from the Structural Break Crunch.Mid-market pricing and volatility prediction
Mid+One™ is a high-frequency mid-market pricing engine built on Crunch. It bridges community research and institutional use cases, turning collective intelligence into a live inference engine operating in under 60 microseconds. It is currently live and being tested by major institutions for:- Mid-market pricing — models from the Mid+One Crunch
- Volatility prediction — models from the Falcon Crunch
Science
Autoimmune disease research
In a competition organized by the Broad Institute of MIT and Harvard (Eric and Wendy Schmidt Center), Crunch models outperformed internal benchmarks on biomedical vision tasks — demonstrating that the protocol generalizes beyond finance.Metabolic disease research
Together with the Broad Institute, Crunch is crowdsourcing new approaches in the Obesity ML competition, where Crunchers tackle obesity prediction using real-world data.Crypto
Powering Bittensor miners
A Crunch Node powers miners on the Synth Bittensor Subnet, enabling any ML engineer to participate in the Synth subnet through the Crunch Protocol.
Each of these use cases proves a fundamental pattern: when diverse modeling approaches combine, individual limitations offset each other, creating solutions no single team could discover alone.
Start building
Ready to create your own Crunch? Follow the getting started guide.