Skip to content
All library documents

Neural Networks for Joint Time-Series and Cross-Sectional Momentum

Article arXiv papers · Author: Wee Ling Tan et al.
Read the original

Summary

This document presents spatio-temporal momentum strategies that combine an asset’s own momentum history with its momentum relative to other assets. The approach uses neural networks to learn trading signals across a portfolio, treating time-series and cross-sectional momentum as related inputs rather than separate strategies.

Backtests cover actively traded US equities and equity index futures. A single fully connected layer is reported to retain an advantage over benchmarks with transaction costs as high as the levels tested. Least absolute shrinkage and turnover regularization perform best across the stated cost scenarios. The evidence is limited to the described portfolios and backtests; the document does not provide further detail on evaluation periods, benchmark definitions, or live trading results.

Key ideas

  • The strategy combines each asset’s historical momentum with its momentum relative to other portfolio assets.
  • A neural network can learn trading signals for multiple assets jointly.
  • A single fully connected layer is reported to perform well in the tested portfolios.
  • Least absolute shrinkage and turnover regularization improve results across the stated transaction cost scenarios.

Tags

EquitiesFuturesMomentumMachine learningBacktesting
Full text
# Spatio-Temporal Momentum: Jointly Learning Time-Series and Cross-Sectional Strategies


# Spatio-Temporal Momentum: Jointly Learning Time-Series and Cross-Sectional Strategies









We introduce Spatio-Temporal Momentum strategies, a class of models that unify both time-series and cross-sectional momentum strategies by trading assets based on their cross-sectional momentum features over time. While both time-series and cross-sectional momentum strategies are designed to systematically capture momentum risk premia, these strategies are regarded as distinct implementations and do not consider the concurrent relationship and predictability between temporal and cross-sectional momentum features of different assets. We model spatio-temporal momentum with neural networks of varying complexities and demonstrate that a simple neural network with only a single fully connected layer learns to simultaneously generate trading signals for all assets in a portfolio by incorporating both their time-series and cross-sectional momentum features. Backtesting on portfolios of 46 actively-traded US equities and 12 equity index futures contracts, we demonstrate that the model is able to retain its performance over benchmarks in the presence of high transaction costs of up to 5-10 basis points. In particular, we find that the model when coupled with least absolute shrinkage and turnover regularization results in the best performance over various transaction cost scenarios.

Shown in full with attribution under the source's licence. Licence: abstract CC0

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.