Beware this is a 3 to 4 hours Workshop!
Who is Crunch: We are a decentralized intelligence platform where Financial institutions tap into thousands of data scientists to produce and crowdsource predictive models on live financial problems.
This workshop is led by Crunch Founder and CEO Jean Herelle.
Each session is designed to get you from zero to a competitive baseline by end of workshop.
[Intro] - Platform introduction and fundamentals - 15min + 10min Q&A
How does Crunch really works in production? How does rewards, scoring, tokenomics work at model node and protocol level? What’s the platform direction and roadmap for coming year?
[Session 1] - Datacrunch New Dataset Release - ⍺H Long / Short Equity - 30min
Datacrunch’s most significant dataset upgrade to date, rebuilt from the ground up based on the fund’s first eight months of live trading. Dataset structure, Target definition, Live EDA on extended history and larger cross section, Modeling implications of the 28 day forward return., Live walkthrough and submission
[Session 2] - Falcon Game - Institutional low frequency eFX - 30min
An anonymized high frequency dataset provided by one of the largest institutional banks to improve their Forex Market Making. Designed by Peter Cotton, this dataset reflects real production constraints including a 20ms maximum inference time. How to handle high resolution time series data? Live EDA. Latency aware modeling Open Discussion / Q&A, Live walkthrough and submission.
[Session 3] - ADIA Lab Time Series Structural Break Modeling Challenge - 30min
Built with Marco Lopez de Prado’s team at ADIA Lab. Structural break concept, dataset and scoring, Top 2025’ Submission, Open discussion and Q&A, Live walkthrough and submission