Co-integration test in every combination is returning '3'.

[ex1dreamer]

Tobias, you wrote in comments:
"These techniques work better with a longer time period, so you would rather want to test this with daily or even monthly data, but it is highly unlikely you would find something at a minute resolution as the return distribution is more likely to be random at this time interval."
I cannot get any results except "3" for RUSSEL2000 (I mean all tickers from the index, which is > 1900 of individual securities) vs SPY on daily timeframe. I also quick test JPM against JPM, for example, and co-integration test still returns 3. Should not it be 1 in case of identical time series?

[vindberg]

Hi @ex1dreamer did you ever find a solution to this? Or another project example?

Thanks.

[ex1dreamer]

to @vindberg I haven't. Apparetly it's strange, no c# implementation exists. there are couple of commerical packages available only.

[vindberg]

Thanks for getting back. I wonder if @Tobias-Mann has any insight into this ;)