My rating (or comparison) method uses (nearly) like for like comparisons as much as possible. I'm more than happy for others to provide evidence that they've found something which supercedes that technique.

Meanwhile I'd like to explore other interesting issues mentioned here:


"There are just as many races run at very fast pace which produce times that are rarely repeated as there are races at slow pace that don't provide and accurate indication of the fields ability."


Given my suspicion that contesting a very fast race is a terrific way of never winning again, I'd infer that there are far more slow races than fast ones. And so maybe osulldj's assertion was uncharacteristically intemperate.


But can anyone volunteer a percentage figure for such slow pace races which are useless for rating?

I don't have one, but just ran this experiment to try and collect my thoughts.


Consider 200 Rosehill 1200m races on officially GOOD tracks.

For this exercise define Pace as Final 600 time/ First 600 time, so that relatively fast early sectionals get the highest figure.

The fastest decile (here the 20th fastest sample) reads:

1.01 Pace Ratio
35.58 Final 600
35.08 1st 600
0.50 Final - 1st (Call this Diff)

If anything with a Diff <= -0.50 (i.e. over 1 second worse) is considered unsatisfactory then the failure rate is 66.5%.

And 29% for a pathetic over 2 seconds worse.


However the 66.5% sample, might mean the least-slow horses spent 6 seconds taking 9 strides 1 metre shorter than I'd like them to. That feels like a lot of energy being saved, aka loafing.



(Assuming ball park 18 meters/second, 12 meters/stride hence 3 strides/ 2 seconds).

http://www.equimost.com/stride-ocala-march02.html