Why did it took so long?

It's been a while since we wrote something. Some of us have been busy with work, finishing master's and finding a job. Finally we are close to summer vacation, which means vacation. This also means that we will have more time to write or blogs on topics.

Which forecasting method are we talking about?

Well lets pick up where we left at. In our last blog we talked about forecasting with exponential smoothing and moving average. Those two methods are really easy to use. So we definitely recommend to check our previous blog which explains these two extensively. Today we will talk about the one an only the regression model!

Regression my old friend

If you ever had statistics or basics in time-series, you have definitely heard of the regression model. We will be talking about the simple linear regression model. Linear regression is used quite often for large run forecasts and it is popular among corporate firms.

SLR (Simple linear regression)

Using regression models for forecasting can be done in three stages. Number one specified the dependent variables. Secondly, create a model indicating the form of the relation between the independent and dependent variables. Once the first two steps have been clarified then, the model can be estimated by means of the regression analysis. Once you have the data you basically can run a linear regression analysis. A regression analysis can be done in almost in any program and software. You can even code and make your own linear regression analysis from scratch using java , C++, python ect. If you have excel, in excel you can fin the analysis tool pack. Once you click on the analysis tool pack you can find the regression analysis option.

Developing regression models

We will now describe the basic steps that are used to develop a regression. Keep in mind there's a lot of theory behind regression, and also behind the simple linear regression. SLR can become sometimes difficult to understand because there are certain rules you have to follow. *Nevertheless, you don't have to follow all the rules and theorems, you follow the rules to make your regression model more accurate. The less rules you follow the less accurate your model will be. Here we just mention the four basic steps. In our next part theorems of SRL are discussed and in our last part we touch up on the econometrics behind the SLR.

1. Plot prices over time (in this case bitcoin prices, this is also known as a time series)
2. Consider the variables that are relevant to predict the prices (normally this is done, but for bitcoin we only focus on the price variable, we will talk in the next part about trends, seasonality, residuals ect.)
3. Collect data (you can do this by going to coinmarketcap website and download the bitcoin prices of the last three months, last year, our last five years and put it on a excel file).
4. Analyze the data (this is done after you let excel or your chosen program run the regression, you will obtain a table and a graph that you will have to interpret. We will help you in the next parts on how to interpret a regression output)

Stay tune

We will describe the rules, theorems and analysis in the next upcoming posts. The Simple linear regression is not so simple, but do not worry we will try to explain everything as simple as possible. We will show in the end after all the basics have been explained how we can currently regres the bitcoin prices.

P.S Just hold on in these bear markets and stay strong for now!