R-Portfolio Files

R-Portfolio - breakeven optimally diversified investment portfolio.  

As input files, quotations for the formation of R-portfolio format is CSV c delimited fields in the form of a semicolon and comma of digits in a decimal or as a point or a comma:  In the first line of the file should be placed the names of financial instruments. The first column in the file must be placed the date and time quotes (for the opening prices of trading sessions) - the format of any. Time must be sorted in descending order of precedence: the most outdated quotation should be placed in the rows above in relation to the most recent. In the first row and first column can be placed any information. At the intersection of the names of financial instruments and time should be placed in quotes as numerical values of prices opening bars of the timeframes. If price changes in a specified time there was no need to specify the price, made for a previous time, from the upper rows of the corresponding columns  Algorithm R-Portfolio converts quotes to the table yields the formula:  

next_profitability = (next_price - previous_price) / previous_price  

This table profitabilities a payoff matrix according to the mathematical theory of noncooperative games for antagonistic two-person game (meaning a trader - a player in the columns and the market - the player along the lines of) a zero-sum. Decision of the payoff matrix for player of the columns is optimally diversified portfolio. If the number of rows of the payoff matrix exceeds the number of columns, the oldest rows are removed so that the number of rows and columns was equal to.  To solve the payoff matrix used a fast algorithm for the Brown-Robinson, which has a high convergence and a small error.  

To avoid fitting to historical data, it is necessary to form portfolios on the basis of quotations of several hundred liquid financial instruments, and several hundred periods. A small number of financial instruments and time periods, as well as illiquid financial instruments may contain random correlation due to the high errors in quotations are not confirmed in the future.