It is a bit strange for me but my basketball under over research shows similar results. I use 6 methods for picking matches and all of them give results around 50%. This of course is 100% unsatisfying, but my research will continue.
Here are the results after 9 matches on WNBA:
When I use my old method with all previous matches – 4 from 9;
When I use my new method with all previous matches from the sector where bookie line is – 5 from 9;
When I use Gauss theory for less square approximate with the last 10 matches for each team – 4 from 9;
The same method used with the last 10 mutual games – 5 from 9;
When I use average scores from data above both are 4 from 9.