AIS Adaptive Trend Smoothing

In order to isolate long-term and non-random components, it is necessary to know not only how much the price has changed, but also how these changes occurred. In other words - we are interested not only in the values ​​of price levels, but also in the order in which these levels replace each other. Through this approach, one can find long-term and stable factors that influence (or may influence) the price change at a given point in time. And knowledge of these factors allows you to make a more or less accurate prediction.

Let's determine what specific results we want to get from the analysis of financial series in time:

In order to fulfill the specified conditions, we use a regression analysis of relative price changes, and make an indicator based on this model. The algorithm of this indicator is based on learning from historical data, and its work is under the complete control of the trader.

• LH is a parameter that determines the number of bars used to smooth the financial series. Its allowable value is between 0 - 255.
• UTS is a parameter that affects the speed and depth of learning. Its value also lies within 0 - 255. If the value of UTS is zero, then learning occurs throughout the history. In all other cases, the training set is updated from time to time. This update occurs the more often the smaller the UTS value.

The main disadvantage of this indicator is that when smoothing the financial series are considered as stationary, which may cause some delay.