The issue of successful investment has been addressed since the beginning of the financial markets. For investors and the academic community, there is a strong motivation to test the reliability of models for predicting the development of prices on these markets. Our goal is to determine which models are optimal for short-term predictions during periods of high market volatility, and what their reliability is. A number of studies indicate that methods of technical analysis are a suitable approach for their ability to be as effective as modern methods, user-friendliness, and their ability to be applied by basic software. The price prediction reliability for the first two weeks of March 2022 is tested on the case of copper based on three conventional forecasting strategies: a) linear regression, b) naive forecasting, and c) exponential smoothing, with the frequencies of data being the daily prices of the preceding month, the daily prices of the preceding three months, and the annual average monthly prices of the preceding 13 months. The results show that linear regression is the most suitable model with the frequency distribution of daily historical prices for the period of three months (MAPE = 4.16%) and average monthly data from the preceding year (MAPE = 4.37%). Predictions based on the daily frequency of data from the period of one previous month achieved the worst predictive results for all the models. The outcomes of this study, reflecting conditions of high market uncertainty, can help investors minimise risks and maximise returns. They demonstrate which method of technical analysis is suitable in terms of predictive power and the choice of data frequency under high market volatility.
Kuchar et al. (Sat,) studied this question.