The purpose of the work is to establish a relationship between the enterprise's practice of providing financial data and the investors' opinion about the enterprise as well as to predict whether the enterprise will be inclined to cheat while providing its financial data in the future. To reflect the investors' opinion about an enterprise, the parameters of skew tdistribution and stable distribution calculated from stock data (close and volume) have been used. The obtained preference area, the parameters (calculated from close and volume stock price data) of the distributions and the indicators representing the present have been employed as Random Forest inputs in predicting the direction of the change in future Accounting & Governance Risk (AGR) rating, which defines the change in the risk of provision of financial data, i.e., whether the risk will increase or decrease. As it has been revealed by the selection of features, stable distribution parameters better reflect the amplitude of the change in stock prices, while the investors' preference area, drawn on the basis of skew t-distribution parameters, has reflected the discrepancy between the investors' expectations and the enterprise's actual value. The same most important selected features have been found to be equally well applicable in describing enterprises characterised by the tendency for AGR rating to rise as well as those characterised by the tendency for AGR rating to drop, or describing both those groups of enterprises collectively.

This paper provides a new methodology for company assessment besides other traditional assessment measures such as share price or forecasts of the analysts. It is suggested to assess the market reaction on change in share price via using graphical approaches. Investors buy shares with the expectation that its price will rise in the future. But sometimes expectations don’t coincide with reality and then shares are sold. This work has been taken into account in the asymmetry between expectations of investors and results. In order to identify the position of a company in 2D space, the paper uses classification algorithm of random forests with data on change in share price during the period of the year in the inputs, and the forecasts of analysts, i.e., whether a price will increase or decrease, for the same year in the outputs. Thus, two clusters of companies are seeking to represent: one of the companies whose changes in share price coincide with investors’ expectations, and another one – on the contrary. This method can be useful to investors, for whom it is important to identify the market reaction about companies from the whole industry or its branches and analyze its trend.

No abstract