ISYE6402 Time Series Analysis Project -- Cryptocurrencies: Pricing, Volatility and Trading Strategy Using Time Series Analysis
Team Member : Nirmit Chetwani, Tianyi Liu, Minghan Xu
Codes:
-
data_prep: input: all the raw csv files that we extract using Kaggle data source and other sources output: a standard, flat table that contains metrics related to all cryptocurrencies and external indices, csv called "combined_crypto_daily_data new.csv" inside "\data\inputs"
-
arima: input: takes the data prepared above what it does: performs ARIMA forecasts, and executes the trading strategy output: three csv files names btc_analysis, dji_analysis and ethereum analysis containing details like forecasts, s.e, optimal arima orders related to these currencies. The data from these files is merged into a csv "trend_analysis". All these files are in the folder "\data\outputs".
-
arimax_var: input: takes the data prepared in the first step output: two csv files named: arima_analysis and var_analysis located in folder "\data\outputs"
-
garch: input: takes input "garch_data.csv" present in "\data\inputs" output: graphs and analysis in the R-code itself.
-
xgb_model: input: btc_data, containing only the bitcoin data, but all fields like open, close, volume, etc. output: a file containing predictions from xg-boost, in the file called "btc_xgb_output.csv" inside the folder "\data\outputs". The trading strategy for this is implemented in the file "trend_analysis" in the folder "\data\outputs".