arnechos/time-series-momentum

🚂💨 Deep Momentum Networks for Time Series Strategies

Time Series Momentum: Deep Momentum Networks

Overview

This project repository implements the deep momentum network timeseries momentum factor proposed by Lim, Zohren and Roberts (2019).

For a description of the full results as well as a summary of time series momentum strategies, please confer the pdf report.

The plot below summarizes the performance of the TSMOM-DMN factor (with and without transaction cost) against the AQR cross-sectional momentum factor and a baseline time series momentum strategy (SIGN).

Data

The data originates from the Pinnacle Data Corp CLC database of continuously linked futures contracts.

Context

This report was created as part of the Dynamic Asset Management course at the Berkeley MFE.