Exchange rate forecasting - LabMazurokCom/Blockchain GitHub Wiki

Objective:

To review forecasting methods for the crypto currency exchange rate.

What would we like to get?

1. The trend of exchange rate in the nearest time interval or for several intervals (the number is set in advance);
2. The numerical value of the exchange rate.

Methodology:

1. Technical analysis;
2. Fundamental analysis.

Literature:

• Article «Forecasting models for the dynamics of the exchange rate on the basis of fundamental indicators»;

• Book «Computer analysis of the futures market»;

• PhD (math) thesis «Behavioral models of the foreign exchange market»;

• Сайт BERG.

Approaches:

1. Multiple Regression (linear, non-linear, piece-wise);
2. Discriminate analysis (standard or generalized);
3. Time series analysis (ARIMA);
4. Neural networks.

Input Data:

We can use both indices of technical analysis (if necessary, transform them according to a selected approach) and the indicators (time series) that are used in fundamental analysis. So we will synthesize two methods!

Findings:

• We should provide a market strategy and take it into account developing and diagnosing our forecasting models;
• The prediction problem can be transformed to the classification task, but we also can get a numerical forecast of the exchange rate for the nearest time interval;
• We should find and explore appropriate indices and indicators;
• We can merge technical analysis and fundamental analysis into one model.