Consider the Cross Correlations for Australian end of day data for the combined:
- ASX Equities
- Australian indices
- Commodities
- Continuous futures
- Economic indicators,
- AU forex
- World indices
These together form a set of 7000 EOD series.
A simple cross correlation would involve 7000*7000/2 cross correlations =2.45*10^7 correlations
Then add in a variable window size for the correlation going form 5 days to 2 years. Roughly increasing correlations to be performed by 500, leaving 2.45 * 10^7 = 1.2*10^10 correlations.
Then add a variable lag from 0 days to 180 days, again increasing the number of correlations by 180
Leading to 180* 1.2*10^10 correlations or approximately 2*10^12 correlations
or roughly a 2 trillion correlations.
Then check to see if there is any statistical significance with null hypothesis testings.
I was wondering if there are any of the shelf finance/trading platforms that can do this or if anyone has done that.
And what did they find in terms of significance.
Also what platform/tool did they use.
And if they did this did the use this in a predictor matrix.
