I used pandas.tools.plotting.autocorrelation_plot in Python to plot autocorrelation functions of the same time series: for its first 100 and 1000 entries, respectively (code below).
from pandas.tools.plotting import autocorrelation_plot autocorrelation_plot(X[:100]) plt.savefig(graph_path+"/acf1.png",format='png') plt.show() autocorrelation_plot(X[:1000]) plt.savefig(graph_path+"/acf2.png",format='png') plt.show()
These pictures clearly aren't consistent. For example, in the first one the first lag with negative autocorrelation is about 30, while in the second one about 180. Do you know how to explain it?