If you have really done enough tests to prove statistical significance, I suggest you team up with a researcher or consultant, verify your results and publish it. There is a chance you may have found something really interesting. However, you might also like to understand (or ask here on the community for) the theoretical explanation before this undertaking in case you do not know that already. That might save you some time.
I think we have done enough explanation here on this, and none of us, including Mr Carlos, will pursue this issue no further here.