Time Series Synchronous Validity Analysis of Oil Exchange Rate Changes Predicted by Neural Network Based on Keyword Research

Abstract

Oil is one of the most important raw materials in the world economy. It is used in almost every segment, logistics, production, delivery of services or delivery of service providers to the service location, etc. In short, it involves the entire economy, which is why the market reacts very sensitively to any volatility or price increase. In the case of large users, the importance is clear, but it also has a great influence in the residential segment, as it affects the elements of the consumer basket, as well as changes the cost of commuting to work, etc. Thus, it is important for actors to be able to prepare for change, if there is a larger fluctuation, they can make their economic decisions in light of this, thus avoiding vulnerability and minimizing losses resulting from volatility. The oil distributors, or more precisely the producers, operate in an oligopoly market, where in many cases they coordinate their decisions and change their production volumes, which increases the uncertainty of the forecast. In the course of our study, we analyse the articles of leading economic journals by keyword research, and we apply several other variables during the forecasting with the Neural Network, and then we compare the co-movement of the resulting time series with the volatility of the real oil price, thus estimating how predictable the changes are for actors with non-confidential information. The time series analysis is carried out using several methods, thereby avoiding partiality, and inferring how this system is suitable for predicting real and not necessarily expected market movements in time, thus for timing our economic decisions, so that the resulting negative impact is as small as possible.