Beacuse the trend estimating is so complex, we use different methods and techniques of interpreting trend on stock market. One of those methods is machine learining model. In the first chapter we analyse more than 15 different moving average filters. Moving average filters are used to remove noise in time series data. The smoothness of the filter is defined by legnth of the period (window) it uses to calculate the averages. Therefore we choose window size that gives us the best accuracy and smoothness ratio. All filters are stored in bank that is later used as data instances of our model. In the second chapter we discuss the usufulness and properties of l1 trend filter (l1tf). L1tf is a good trend indicator due to its picewise linearity. How much will the l1tf follow the original data depends on the regularization parameter lambda which defines the number of slopes that l1 fitler has. By using large number of data on various values of lambda, we choose lambda that had the most correct future trend estimtes, L1tf will be used as labels of our machine learning model. The third chapter goes through the basics of regression analysis and when do we use linear regression on arranged pairs of dependent and independent variables. We create and train our linear regression model with the help of python library skicit-learn. By using the t-test on every of the filters, we reject those filters who are irrelevant to our regression model. Logistic regression is the subject of chapter four. For this regression we redefine our labels to series of ones and zeros, where ones represent the positive, and zeros the negative slope of each piece of the l1 filter. We use different metrics to represent the success of our model, such as confusion matrix, accuracy and f1 score. By adding more moving average filters and improving ther algorithm, and also by adjusting the values of parameter lambda and window size, we improve the results and estimates of our model. When we are satisfied with the accuary of our logistic regression model, we conclude the task finished.