I did a few tests to really find out what that "inconsistency" was about.
Take this example:
Lewis Hamilton,16370,100,16240,100
So, the basic performance is 16240 for quali, but the 100 is a variable to be ADDED to that number, so the bigger the better, just that you can't be sure about the real performance (sometimes will be 16260, others 16300...), but it will be within the range of 16240 to 16340.
So, some comparisons:
- (16000, 100) is better than (16000,0) but
- (16000, 100) is worse than (16100,0)
If someone is more interested in these results I have a couple of xsl sheets with test data with performances and power... It is interesting to give the right dimension to your performance files.