I know TGJB. Trust me. I was just giving an easy example. I\'ve seen many days where when you are trying to figure out what the variant looks like you see this for a 10 race sequence, and let\'s assume they are all 1 turn sprints just to keep it simple:
+4,+6,+3,+18,+4,+7,+20,+3,+1,+4
So all the race times seem fast for the day, but what happened in races 4 and 7? Did the winner really run that fast? If so then you would expect to see one horse win by 10 lengths and everyone else be far behind. But what do you do when it turns out both of those fast races were decided by a photo and most of the rest of the field were only a length or two behind? Was the whole field that fast that day? (not likely, but possible).
So that\'s the problem you have when you try to come up with a universal variant for the day and then apply it to all races. Because if you just average all those out you come up with a variant of +7. But if you split out the two fast races and do the others by themselves you come up with a variant of +4, which seems like a much better fit.
I say you have to split them out because otherwise you adjust those 8 other races that seemed to fit a variant of +4 down by +7 instead. So you make all those horses in 8 races look slower just because the two other races happened to be very fast that day. That doesn\'t seem right to me. And that\'s the problem with taking such a dogmatic, the number is the number and the track doesn\'t change at all approach.
It can\'t be right.
I say you go off the horses and try to come up with a variant (this is important because you need it as a data point), then see if it seems to make sense for the day. Some days it will. Easy day at the office. Some days it won\'t and we have to do some thinking...
