optimal lag length in VAR and Johansen test

[AURA1112]

Hi everyone,

I have got some problems in my cointegration test. I am testing the market efficiency in futures market. I select the data spot prices and futures prices in datastream. The data and my eviews files are in the attachment. When I estimate the VAR, in the lag criteria, the optimal lag length for f1, f2, f3 are 2, 3, 4 respectively. However, in the Johansen test the optimal lag length is all 1 when I choose 'option 6' to show all results. I am not sure which one I should use in Johansen test and all these time series data are I(1). So if the lag length in VAR is p, I think the lag length use in JJ test should be p-1. That is what I am confuse about.
The other question is about the Johansen test resul, as you can see in the EViews file, if I choose 'option 3' the test result are all reject it. I think it is unusual because the test should be completed when we accept the null hypothesis. But if I choose 'option 4', it reject the 'none' and accept 'at most 1'. So should I choose 'option 4' to analyse it or there is some problem with the data I selected?
I am doing my dissertation now and I have already be sucked for a week about these questions. I am really thankful if some one can help me to fix this problem.