Computing Trading Strategies Based on Financial Sentiment Data Using Evolutionary Optimization

Abstract

In this paper we apply evolutionary optimization techniques to compute optimal rule-based trading strategies based on financial sentiment data. The sentiment data was extracted from the social media service StockTwits to accommodate the level of bullishness or bearishness of the online trading community towards certain stocks. Numerical results for all stocks from the Dow Jones Industrial Average (DJIA) index are presented and a comparison to classical risk-return portfolio selection is provided.

A BERT based Sentiment Analysis and Key Entity Detection Approach for Online Financial Texts

“…The existing techniques of sentiment analysis are mainly divided into three types: the sentiment dictionaries based methods [5,6], machine learning based models [7,8], deep learning based models [9]. Among them, deep learning methods for sentiment analysis have become very popular.…”

Fintech Risk Management: A Research Challenge for Artificial Intelligence in Finance

“…Typically, the Telegram groups are characterized by many members and detailed discussions about the value of the individual projects, as well as, by the expectations of the communities concerning the success of the ICO and the company. By collecting data from the Telegram ICOs (including the corresponding white papers) and discussions on Telegram chats relating the value and prospects of the projects in question, we can build, train, and test supervised models to discriminate and classify ICOs by their probability of fraud, using for example the methods shown in Hochreiter ( 2015 ).…”

News-Based Trading Strategies

“…Its trading engine recommends trades, such as "buy stock X and hold it until the stock prices hit the +d % barrier". Similarly, a trading strategy can be built around social media data [28] or the Google query volume [29] for search terms related to finance. The latter variable is inserted into a simple buy-and-hold strategy (without transaction costs) to buy the Dow Jones index at the beginning and sell it at the end of various holding periods.…”

“…Different data sources might provide further insights into the costs of additional noise. For example, innovative news sources [15,28], such as social media like Twitter, offer a massive volume of new textual materials, which are published quickly and without quality control. Related approaches also incorporate Google query volume for search terms related to finance [29] and Internet stock message boards [11].…”