Comparing forecast accuracies of machine learning and financial analyst

Loading...
Thumbnail Image

URL

Journal Title

Journal ISSN

Volume Title

School of Business | Bachelor's thesis

Date

2022

Major/Subject

Mcode

Degree programme

Rahoitus

Language

en

Pages

17

Series

Abstract

In this study, I compare financial analysts’ stock price growth prediction accuracy to neural networks prediction accuracy. Predictions are made for the same US stocks from 2019 to 2021 and the accuracies are measured in R-squared as well as accuracy metric. Findings show that the neural networks outperform financial analysts for this period in both metrics. Additionally, the prediction statistics show that neural networks make more pessimistic predictions with lower variance. Analysts on the other hand make more optimistic predictions with a higher variance which seems to be the reason why the neural networks score better values in these metrics.

Description

Thesis advisor

Vihriälä, Erkki

Keywords

neural networks, analysts, stock growth predictions, accuracy

Other note

Citation