A Mobile Computing-Friendly Stock Price Trend Prediction Model

In recent years, stock price trend prediction has been a hot topic in the field of AI for finance. Meanwhile, with the constant fluctuations in financial markets, investors increasingly require real-time stock price trend prediction on their mobile devices. Existing stock mid-price prediction models based on limit order books are mostly deployed on servers or computers, with their complex network structures and large parameters unsuitable for mobile computing. In this paper, we design a lightweight stock mid-price prediction model based on the transformer and develop an application that can be deployed on Android phones to forecast stock price trends. Our model substantially achieved a maximum reduction of 82% in Parameters and over 90% reduction in memory consumption compared to state-of-the-art works.