ASSESSMENT OF THE QUALITY OF THE STATISTICAL FORECAST FOR THE SHARE PRICE

Анотація

The stock market is an important part of the world's economy, but the values of stocks change at a significant speed and finding an effective method of forecasting the price of stocks will be a significant victory for investors. The article is devoted to the use of statistical methods in the stock market, evaluating their effectiveness on the example of determining the forecast share price and comparing it with the real share price. In addition, work was carried out to determine the relevance of the research topic by means of bibliographic analysis based on the results of searching a selection of publications in the Scopus database with the subsequent use of the VOSviewer software toolkit. A time series covering the period of daily values of stock prices of Apple Inc. was chosen as the input test sample for developing the forecast. from December 12, 2016 to December 12, 2021. At the same time, the median values of the prices of the monthly financial shares of the Apple com corporation were used to build forecast models, since there is no information that fell on weekends for the studied time period. Forecasts are published using Statgraphics 19 software, which generates 17 forecast models: random walk, random walk with drift, constant mean, linear trend, quadratic trend, exponential trend, S-shaped trend, simple moving average, simple exponential smoothing with a smoothing coefficient α = 0.9999, linear exponential Brownian smoothing with smoothing factor α= 0.5705, linear exponential Holt smoothing with smoothing coefficients α = 0.9999 and β = 0.0359, quadratic exponential Brownian smoothing with smoothing coefficients α = 0.3046, ARIMA model. Akaike's information criterion was used to assess the quality of the models. As a result of the analysis, the ARIMA(1,0,0) model has the lowest value according to Akaike's criteria. On the basis of ARIMA(1,0,0), a forecast of stock market exchange rates was built 6 months ahead of the age of the final value of the input array. In the process of comparison with real data, a significant discrepancy between the forecasting results and the real share price in the forecasted period was found, however, the obtained models were tested for statistical significance. The conducted research showed that it is possible to use classical statistical methods to develop a forecast on the stock market only for a short-term period, while it is necessary to take into account the risk factors inherent in the stock market (the frequency of share price changes during the day, liquidity risks, operational risks, political situation in the world).

Класифікація

Ідентифікатори

Схожі роботи