Investigating the use of the method of "nearest neighbors" to predict the behavior of stock prices in the implementation of arbitrage
In this paper, it was considered one of the types of trading in the stock market - Implementation of arbitrage. The aim of the study was to examine the possibility of using the method of "nearest neighbors" with heuristic rules to predict short-term stock price behavior during arbitrage transactions. In a paper tests the hypothesis that the two parameters - TotalRise (percentage price change over the entire period of growth) and LastChange (percentage change in price over the last day) - are crucial for predicting the behavior of stocks after a sharp rise on positive news. Consequently, an investor might assume, how to behave in the price of the shares, if the result will analyze arbitrage other stocks have close to the value of the shares and TotalRise LastChange. For this action, the risk of loss was defined as the ratio of the neighbors with a loss to all of its neighbors (if 20 neighbors of the action 5 neighbors were unprofitable (i.e. the respective shares rose in price), the riskiness of the operation can be set equal to 25%). Win value was defined as the average of the gains (and losses) of all the neighbors. As a result, developed a model is plausible determines the behavior of the shares.