RT Journal Article
SR Electronic
T1 Online Learning with Radial Basis Function Networks
JF The Journal of Financial Data Science
FD Institutional Investor Journals
SP jfds.2022.1.112
DO 10.3905/jfds.2022.1.112
A1 Gabriel Borrageiro
A1 Nick Firoozye
A1 Paolo Barucca
YR 2022
UL https://pm-research.com/content/early/2022/12/07/jfds.2022.1.112.abstract
AB We provide multi-horizon forecasts on the returns of financial time series. Our sequentially optimised radial basis function network (RBFNet) outperforms a random-walk baseline and several powerful supervised learners. Our RBFNets naturally measure the similarity between test samples and prototypes that capture the characteristics of the feature space. We show that the training set financial time series returns have low similarity with their test set counterparts, highlighting the challenges faced in particular by kernel-based methods that use the training set returns as test-time prototypes; in contrast, our online learning RBFNets have hidden units that retain greater similarity across time.