A Word is Worth A Thousand Dollars: Adversarial Attack on Tweets Fools Stock Predictions

TL;DR

This paper demonstrates that social media-based stock prediction models are vulnerable to adversarial attacks, which can manipulate predictions and cause financial losses by subtly altering tweets.

Contribution

It introduces a novel adversarial attack method on social media-based stock prediction models, considering semantic and budget constraints, and evaluates its effectiveness.

Findings

Attack achieves high success rates in fooling models.

Adversarial tweets cause significant monetary loss in simulations.

Method works across multiple stock prediction models.

Abstract

More and more investors and machine learning models rely on social media (e.g., Twitter and Reddit) to gather real-time information and sentiment to predict stock price movements. Although text-based models are known to be vulnerable to adversarial attacks, whether stock prediction models have similar vulnerability is underexplored. In this paper, we experiment with a variety of adversarial attack configurations to fool three stock prediction victim models. We address the task of adversarial generation by solving combinatorial optimization problems with semantics and budget constraints. Our results show that the proposed attack method can achieve consistent success rates and cause significant monetary loss in trading simulation by simply concatenating a perturbed but semantically similar tweet.