A Survey of Long-Term Memory in the Digital Currency Index

Subject Areas : Financial engineering

Shima Alizadeh ¹ (Department of Business Management, Faculty of Management, Central Tehran Branch, Islamic Azad University, Tehran, Iran)
hossein safarzadeh ² (Department of Business Management, Faculty of Management, Central Tehran Branch, Islamic Azad University, Tehran, Iran)

Received: 2018-11-20 Accepted : 2019-03-17 Published : 2019-09-23

Keywords: Time series, Crypto currency, Long-term memory, Autoregressive Fractionally Integrated Moving Average,

Abstract :

Long-term memory, also referred to as long-range dependence, explains the correlation structure of values of a time series at long intervals. According to the efficient market hypothesis, prices follow a randomized step process, so returns on assets can not be predicted based on past price changes. Long-term memory is a weak point of the business-market hypothesis. Long-term processes have important implications for asset yields and play a crucial role in time series analysis. This study examines the existence of long-term memory in the price index of crypto currencies equals $ 1 and lower for the period from September 1, 2015 to September 1, 2018. To estimate the parameter d, the OLS method is used in the EVIEWS software package. The ARFIMA model is used to test hypotheses. The results indicate that long-term memory is in the currencies of DIGIBYTE, Dodge Coin, EMER Coin, BITSHARES, MAIDSAFE COIN, XEM, Redd Coin, NXT, Verge and Ripple, and on the other hand, three currencies of Byte Coin, SIA Coin, STELLAR lacks long-term memory, and therefore these currencies are among the most efficient market products.