Predicting Bitcoin Market Capitalization Using SmoothStep and ArcStep Regressions

A Multi-Factor Reverse Optimization Framework

• Associate Professor of Finance Ricky Cooper
• Associate Professor of Finance Benjamin E. Van Vliet (presenter)

Abstract:

This paper makes empirical and methodological contributions. I develop a new model for forecasting the market capitalization of Bitcoin. To do this, I develop new methods for non-linear regression using SmoothStep functions for sigmoid processes, and I define a new family of polynomial functions I call ArcStep for bounded exponential growth processes. These functions demonstrate comparable SSE performance versus the traditional polynomial, logistic, and bounded exponential regression models respectively. More importantly, the optimized parameters for these functions capture clear economic intuitions, unlike the traditional functions. Using these new techniques, my Bitcoin model is able to make precise predictions regarding both the maximum market capitalization and the time that maximum will be achieved.

 

All Illinois Tech faculty, students, and staff are invited to attend.

The Friday Research Presentations series showcases ongoing academic research projects conducted by Stuart School of Business faculty and students, as well as guest presentations by Illinois Tech colleagues, business professionals, and faculty from other leading business schools.

Join session on Blackboard Collaborate