Financial time series such as stock prices, exchange rates and crypto currencies are generally non-linear and non-fixed variables. To date, many researchers have used statistical models, wavelet models and Bayesian models to predict stock prices, exchange rates and crypto currencies. Statistical models assume that time series are stationary and linear, which brings with them large statistical errors. Wavelet models have recently been used as an alternative to time series analysis.

The market value of the crypto currencies is $ 200 billion and the daily trading volume is $ 9 billion. Crypto-currencies, whose market values ​​are increasing day by day, are in the financial sector as a significant gaining tool with their risks. From the crypto currencies called Bitcoin, the changes in the market value and transaction volume of the 3 major currencies (BTC, ETH, XRP) in USD terms were analyzed. In the financial sector, investors are taking risks and taking positions by selecting an investment tool type. The estimation of crypto currency has a significant place in this investment instrument, which involves other security risks of investors.

In this study, Wavelet transforms and Bayesian approach have been utilized in the estimation of crypto currencies as investment instruments. Wavelet-based regression model of cryptocurrencies is obtained by applying Discrete Wavelet Transform (DWT) and Maximal Overlap Discrete Wavelet Transform (MODWT). In this model, Bayes optimization was performed with Markov Chain Monte Carlo (MCMC) simulation and a price estimate was made with a hybrid approach. The comparison of the estimates and the error analysis and the success of the hybrid approach were observed.