Introduction Differential Machine Learning (DML), as introduced in the recent arXiv paper (Differential Machine Learning for 0DTE Options with Stochastic Volatility and Jumps), extends supervised learning by incorporating not only function values but also their derivatives. In financial contexts, this often means sensitivities such as Greeks. However, when direct derivatives are unavailable, we can approximate […]

R-bloggers

Differential Machine Learning (DML) is adapted for Bitcoin price forecasting in R using a twin-network architecture built with Keras3. Instead of mathematical derivatives, RSI, MACD, and Bollinger Bands serve as volatility proxies. A primary network learns price/trend dynamics while an auxiliary network captures volatility signals. Their outputs are combined via a linear regression stacking ensemble, reducing RMSE from ~76,000 (individual networks) to ~3,030 and MAPE from ~99% to ~3.65%. Residual-based 95% confidence intervals are added for uncertainty quantification, and results are visualized with ggplot2.

Differential Machine Learning with Twin Networks in R: Forecasting Bitcoin with Volatility Proxies

Why Volatility Variables Instead of Derivatives?

Keras3 in R: Flexible Deep Learning for Financial Forecasting