Calculation moving averages trend

Calculation trend

The trends are identifying the value movements of individual time series sequences. The trend in time series is not linear, but not obvious. The trends are calculated by relating the current value of a phenomenon in the previous value. Several techniques are proposed to identify trends [7, 10]. It is important that everyone has a special filter tech trend. Some methods, such as moving averages, incorporate a trend indicator in writing technique. Others are trying to anticipate the inevitable change in the trend and therefore enters when the trend is against them at that time.

A simple approach to identify trends to find total Directional movement during the time period you selected for measurement. If it is positive, the trend is, if negative, the trend is down, and if 0, this trend is neutral. More sophisticated approaches can use mathematical equations and methods of massage past values ​​precisely determine whether the trend is up or down.

Because our main goal is to focus on the overall technique of determining the dependencies between time series, comparing the different approaches of calculating the trend indicators is beyond the scope of this paper, we
favor any specific approach for identifying indicators trend. In this paper, we adopt a simple approach for identifying indicators trend. The approach is modified to predict more or less correct. A trend scale is used to determine the amount of space used to set the values ​​of the trend indicators. For example, if the trend is equal to 4 scale, this trend is an indicator space {2, 1, 0, -1, -2}, where 2, 1, 0, -1, -2 and means very high high, neutral, low and very low, respectively. In figure 2.4, we show that this trend indicators of the prices of oil data and S & P500 stock prices data.