Active investment strategies are built on signals that specify active positions in financial assets or markets. When extracting these alpha signals from data (viz. time series), the question is what the optimal sample frequency (horizon) is. A new strand of literature in finance explicitly focuses on signal processing and investigates the cyclical behavior of the alpha signals and the corresponding alpha (out-) performance. Amongst those, Chaudhuri and Lo (2016) consider the use of spectral analysis, the decomposition of time in the frequency domain, to extend portfolio theory in this direction. Lopez de Prado and Rebonato (2016) introduce the kinetic component analysis, the decomposition of signals into hidden components associated with position, velocity and acceleration concepts.
These new techniques can provide new insights for quantitative investing, with applications ranging from portfolio construction and risk management to signal improvement. This research project aims at shedding light on the potential for signal processing techniques to bring new insights. You will conduct a literature review, replicate the reference papers and extend their work to the analysis of alpha factors. Specific focus will be put on evaluating the benefits of frequency diversification, the identification of market states and the potential for frequency-based timing of factors.
 Chaudhuri, Lo, 2016, “Spectral Portfolio Theory”, Working Paper. Available at SSRN: https://ssrn.com/abstract=2788999
 Lopez de Prado, Rebonato, 2016, “Kinetic Component Analysis”, The Journal of Investing, Fall, 25 (3) pp.142-154