Using machine learning to forecast long-term equity price movement : an empirical study of the Finnish financial markets
Patrikainen, Tuomas (2019-05-12)
Using machine learning to forecast long-term equity price movement : an empirical study of the Finnish financial markets
(12.05.2019)
Julkaisu on tekijänoikeussäännösten alainen. Teosta voi lukea ja tulostaa henkilökohtaista käyttöä varten. Käyttö kaupallisiin tarkoituksiin on kielletty.
avoin
Julkaisun pysyvä osoite on:
https://urn.fi/URN:NBN:fi-fe2019052116456
https://urn.fi/URN:NBN:fi-fe2019052116456
Tiivistelmä
Predicting equity price movement is one of the fundamental challenges in finance, and even small improvements in prediction performance can be highly profitable for investors. Long-term investment is one of the popular investment strategies that investors follow. However, evaluating which companies are going to perform well in the future is difficult. This research presents machine learning aided approach to forecast long-term price movement of the stocks listed on the Helsinki Stock Exchange.
The purpose of the research is to find out which machine learning model performs the best in the Finnish financial markets and to understand what the key variables are, which have a major effect on the prediction accuracy of the models. The research is also testing whether the macroeconomic variables of Finland increase the accuracy of the machine learning models when forecasting long-term equity price movement. The following machine learning models are used in the research: logistic regression, support vector machine, decision tree, random forest, and k-nearest neighbors.
This research produced a number of key findings: the results from the models indicated that the best performance was achieved by the random forest model, which obtained classification accuracy of 65.3% and F1 score of 60.8%; the random forest model is able to give over 60% chance for an investor to pick a stock, which will have a 10% or higher return over the period of one year; the macroeconomic variables increased the prediction performance of every machine learning model used in the research.
The main conclusions drawn from this research are that the macroeconomic variables can provide new information, which is not explained by only using financial ratios in the models. Also, the equity prices in the Finnish financial markets are not equally random, meaning that they do not always follow a random walk process. Therefore, this research argues that the Finnish financial market is not highly efficient, thus stock prices are on some level predictable. These findings contribute to the financial theory of market efficiency.
The purpose of the research is to find out which machine learning model performs the best in the Finnish financial markets and to understand what the key variables are, which have a major effect on the prediction accuracy of the models. The research is also testing whether the macroeconomic variables of Finland increase the accuracy of the machine learning models when forecasting long-term equity price movement. The following machine learning models are used in the research: logistic regression, support vector machine, decision tree, random forest, and k-nearest neighbors.
This research produced a number of key findings: the results from the models indicated that the best performance was achieved by the random forest model, which obtained classification accuracy of 65.3% and F1 score of 60.8%; the random forest model is able to give over 60% chance for an investor to pick a stock, which will have a 10% or higher return over the period of one year; the macroeconomic variables increased the prediction performance of every machine learning model used in the research.
The main conclusions drawn from this research are that the macroeconomic variables can provide new information, which is not explained by only using financial ratios in the models. Also, the equity prices in the Finnish financial markets are not equally random, meaning that they do not always follow a random walk process. Therefore, this research argues that the Finnish financial market is not highly efficient, thus stock prices are on some level predictable. These findings contribute to the financial theory of market efficiency.