Abstract:

In this study, gold prices are estimated by using KNEA algorithm, which is extended from ARPS decline curve model. ARPS decline curves that are used in predicting oil production quantities  provides a good indication of the nonlinear relationship between production capacity and influencing factors, KNEA provides input data in non-linear combination, and inevitable (extreme small or large) deficiencies in time series can be estimated. The study focused on gold, silver, platinum, palladium, Brent Petrol, Natural Gas, 5, 10 and 30-year bonds, S&P500, Nasdaq, DowJones, FTSE100, DAX, CAC40, SMI, NIKKEI, HANH, SEND and Euro/Dollar data between January 4, 2010 and December 14, 2015 for analysis. For testing, 10 days of the 2015 December data were calculated as Mean Absolute Deviation 6.926, Mean Square Error 66.348, Root Mean Square Error 8.145 and Mean Absolute Percentage Error 0.65 from the 1447 days in total data.

Keywords: