A
Digital Filters
Digital filters are intended for signal processing (filtration) presented as time series. We will exaime only line digital filters (hereinafter DF), as thier methods of calculating exist for a long time, both of filters themselves and their characteristics. Besides, they are mostly often used by traders, although traders do not know this themselves.
I will give some examples:
Simple moving average (SMA). As it is known, it is calculated as an average value of final N elements of time series. This is an example of DF with final impulse response (FIR-filter).
Triangular and weighted average (Triangular & Weighted). These moving averages are calculated in the same way as simple average, but members of time series are taken with different weight. For triangular average weight the triangle is made, so middle elements have the largest weight, and last elements - the smallest. For weighted average the last row elements are the most important, unimportant - the most old elements. They are examples of FIR-filters, but with other characteristics.
Here we can add such well-known indicators and their combinations as Momentum, ROC, MACD, TRIX, difference of two moving averages with different periods or taken in different moments of time and many others. I will not describe them in details now. All of them are digital filters and come to a sum (finite or infinte) of input row members taken with some weight or to finite sum of some last members of input row and it is possible that some members of output row taken with some weight. The set of weight defines filters qualities. So we came to a formular of line digital filter, it is written in the following way:
Y = B[0] X + B[1] X[i-1] + .... - A[1] Y[i-1] - A[2] Y[i-2] - ....
Here X input time series, Y output row, A and B sets of weight coefficients.
Digital filters are intended for signal processing (filtration) presented as time series. We will exaime only line digital filters (hereinafter DF), as thier methods of calculating exist for a long time, both of filters themselves and their characteristics. Besides, they are mostly often used by traders, although traders do not know this themselves.
I will give some examples:
Simple moving average (SMA). As it is known, it is calculated as an average value of final N elements of time series. This is an example of DF with final impulse response (FIR-filter).
Triangular and weighted average (Triangular & Weighted). These moving averages are calculated in the same way as simple average, but members of time series are taken with different weight. For triangular average weight the triangle is made, so middle elements have the largest weight, and last elements - the smallest. For weighted average the last row elements are the most important, unimportant - the most old elements. They are examples of FIR-filters, but with other characteristics.
Here we can add such well-known indicators and their combinations as Momentum, ROC, MACD, TRIX, difference of two moving averages with different periods or taken in different moments of time and many others. I will not describe them in details now. All of them are digital filters and come to a sum (finite or infinte) of input row members taken with some weight or to finite sum of some last members of input row and it is possible that some members of output row taken with some weight. The set of weight defines filters qualities. So we came to a formular of line digital filter, it is written in the following way:
Y = B[0] X + B[1] X[i-1] + .... - A[1] Y[i-1] - A[2] Y[i-2] - ....
Here X input time series, Y output row, A and B sets of weight coefficients.