Time series prediction using tools from Chaos Theory

This is a PhD thesis from 1997 that became a seminal work in Quantitative Analysis. It explains the process of using Chaos theory to analyse time series, assess their predictability, detect chaos, and to make limited (if there is chaos) predictions.
It is very accessible for a thesis and surprisingly has not been overtaken by time. The results and techniques described are applicable to a range of disciplines, Finance, of course, but notably biotech and climate studies.
Despite being an academic work, it explains how the results were achieved in practice with pseudocode supplied for the algorithms employed.