This repository contains code used to collect coefficients describing the intensity at which market buy / sell orders are received as a function of mid-price. This data is collected over several thousand five minute periods from a cryptocurrency data source. During these minute periods, information on the spread and orderbook is also collected, and I fit a regressive model on this data to predict the collected coefficients. Refer to my personal website for further details. Ultimately, I find that a large portion in the variation of the model coefficients in question can be predicted as a non-linear function of spread, and as a linear function of the shape of the orderbook, to first order.