Chart – Spread ACF
It is called Autocorrelation and is constructed so that each histogram represents the correlation value obtained by evaluating the close of the 1500 sample bars using this procedure:
the first histogram is the average of the correlation calculated between the first bar and the second, between the second and third, between 1499 and 1500
the second histogram skips 1 and so it becomes the first with the third, second with fourth, 1498 with 1500the third skips 2 … 20 skips 19 and so becomes the first with 20 the second with 21, the third with 22 the 1481 with the 1500.
Observing the histograms I will have to note low values (less than 0.5) and opposite values (negative and positive).
In the image we have the autocorrelation of a perfect couple.
What should I observe:
since this graph is calculated on the Z-Score, we would like to have:
- As in the figure if we are waiting for the Z-Score to return to zero;
- With histograms that often oscillate between positive and negative if the Z-Score exceeds the two threshold values of 2 standard deviations, because since there is no longer a trend it is time to open the position.