Deep Learning Stock Volatility with Google Domestic Trends

TL;DR

This paper demonstrates that a Long Short-Term Memory neural network, using Google domestic trends as indicators, can significantly improve the prediction accuracy of S&P 500 volatility over traditional models.

Contribution

It introduces a novel approach combining LSTM neural networks with Google trend data to model stock volatility, achieving superior prediction accuracy.

Findings

LSTM model achieves 24.2% mean absolute percentage error.

Outperforms Ridge/Lasso and GARCH benchmarks by at least 31%.

Optimal data normalization maximizes mutual information between trends and volatility.

Abstract

We have applied a Long Short-Term Memory neural network to model S&P 500 volatility, incorporating Google domestic trends as indicators of the public mood and macroeconomic factors. In a held-out test set, our Long Short-Term Memory model gives a mean absolute percentage error of 24.2%, outperforming linear Ridge/Lasso and autoregressive GARCH benchmarks by at least 31%. This evaluation is based on an optimal observation and normalization scheme which maximizes the mutual information between domestic trends and daily volatility in the training set. Our preliminary investigation shows strong promise for better predicting stock behavior via deep learning and neural network models.