A retail investor in a cobweb of social networks

Plos One. 2022. Vol. 17. No. 12. Article e0276924.

Teplova T., Tomtosov A., Sokolova T.

In this study, using AI, we empirically examine the irrational behaviour, specifically attentiondriven trading and emotion-driven trading such as consensus trading, of retail investors in an emerging stock market. We used a neural network to assess the tone of messages on social media platforms and proposed a novel Hype indicator that integrates metrics of investor attention and sentiment. The sample of messages, which are written in Russian with slang expressions, was retrieved from a unique dataset of social network communication of investors in the Russian stock market. Applying different portfolio designs, we evaluated the effectiveness of the new Hype indicator against the factors of momentum, volatility, and trading volume. We found the possibility of building a profitable trading strategy based on the Hype indicator over a 6-month time horizon. Over short periods, the Hype indicator allows investors to earn more by buying stocks of large companies, and over «longer» periods, this indicator tends to perform better for illiquid stocks of small companies. As consensus trading tends to produce negative returns, the investment strategy of ‘Go against the crowd’ proves rewarding in the medium term of 3 months.

Keywords: социальные сети, российский фондовый рынок, искусственный интеллект, social networks, neural networks, Russian stock market, частные инвесторы, нейросети, artificial intelligence, retail investors, сентимент инвесторов, investor sentiment, stock portfolios, портфели акций