>If, for example, a race is run with a very fast pace, ALL the horses will have slower final time numbers than expected.<

I agree with you except for the above on occasion. I will use my typical example:

Assume there are the 5 horses in a race. To keep it simple these are their typical and expected figures:

a = 0
b = +2
c = +2
d = +4
e = +5

Now let's assume b, c, and d get into a 3 way duel in 44 flat in a 7F race with a final time of 123+. Obviously a smoking pace.

Assume 'a' was 8 lengths off the early pace and won by 5 lengths.

'b' and 'c' finshed half length apart in 2nd and 3rd.

'd' finished 3 1/4 lengths behind 'b' and 'c'

'e' never got into contention and finished well beaten by 9 lengths.

One could easily interpret this race as 'a' improving sharply and the rest of the field more or less running close to their typical figures.

IMO, it is more likely that:

'a' ran his typical figure.

'b' and 'c' ran a few lengths slower than usual because of the duel (an equal performance but slower).

'd' was impacted even more by the duel than 'b' and 'c' because he's not as good as them, thus the pace was even more extreme for him and effected him more.

'e' ran slightly subpar.

Now, if you break this race out from the rest of the day because you assume the pace didn't have an impact on 'b', 'c', and 'd' (the duelers), you will be overrating 'a' and 'e' and probably underrating 'd' a little.

Futhermore, if you then adjust their performances for pace at a later date, you will be double counting the impact.

The problem is that when the pace (and other unknowns) are baked in, it solves a lot problems, but it ceates new ones because when it comes to pace, not all the horses are impacted equally in the same race. The impact is dependent on position, ability, and other attributes of the individual horse.

I do not have a perfect solution for this problem. I just know it exists because I've seen it and exploited it for many years. No figure maker I know of has a perfect solution for these complexities because there is no perfect formula for pace/time.

However, "you have to know" whether the pace was fast or slow enough to impact the time in a measureable way. In those instances, depending on who made the speed figure, there will different kinds of errors.

These are my current beliefs:

1. The Rags rarely break races out so they will sometimes give out speed figures that won't make sense based on the prior and subsequent performances of the horses, but they will make sense in light of the pace if you are aware of it and adjust accordingly.

2. Beyer is very inconsistent. Sometimes he bumps races up and/or down when they come up unusually fast or slow and sometimes he doesn't regardless of the the pace. Without going back to the DRF and seeing what track variant he used, you can be really lost in the races where the pace had an impact.

3. TG is giving out figures that "BEST" represent the abilities of the horses, but IMHO, he occasionally overrates horses that won big because of pace related issues (example above) and underrates others (like RIM in his Belmont race).

Naturally, if I had only one choice I would go with TG and sort of ignore the pace. However, for myself, I try to understand what does into each figure makers' numbers because I am certain that some of the high profile differences are related to methodology and pace issues.