Why doesn't splitting by even and odd IDs produce cohorts of users with the same behavior (in A/A testing)?

Question

Google Analytics

I'm trying to do A/A testing: I check how users respond to the same form, so that I can show them two different forms and choose the one that gives the most conversions.
Split users by their ID. I record separately the number of form submissions for even users, and separately for odd ones.
The test has been going on for 10 days, about 50 events have been collected. And all these days are constantly "winning" the odd ones! And not by a few percent, but in the proportion of two to one, that is, twice. There was not a single day when the even ones won or even equaled the odd ones.
How to be in such a situation, what kind of conspiracy is this? (=

Answer 1

Can you explain about the use of ID?
I understand you are passing UserID?