I put together a spreadsheet of Groups race in England for 2012 and pulled the Racing Post Rating for each winner. I sorted the races by age and sex and came up with an average rating for all Group 1, Group 2 and Group 3 open company races (no restrictions to 2yo, 3yo, or females).
I'd like to add races from France and Ireland into this analysis but that will take a lot longer to update, so we'll just go with England for now.
For North American races, I engaged in the same process - sort all graded races by age and sex and then average the Beyer, Bris and Equibase speed figures for each grade.
For now, I've included all surfaces with each grade with no separation between turf, dirt and synthetic. If we were to break things out simply by surface the sample sizes get pretty small when looking at a single year. Additionally, I'm trying to get a gauge of overall figures at each level and not into the granularity of surface, distance, etc. As the amount of information in the database grows in the future, the sample size will allow us to take a better look at more specific conditions and performance.
There is one big caveat to the Group 1 races this year in England: Frankel (GB). When you've got a horse that's routinely pumping out 138, 142 ratings in race after race, it has a tendency to skew the numbers. In order to make a distinction between Frankel and non-Frankel races, I've averaged the Group 1 ratings from this year's races with and without Frankel's numbers. In those cases, I included the rating of the runner-up. The overall difference between the Frankel/non-Frankel races was 5 points: Group 1 average with Frankel is a 129, without it's a 124.
Below is a summary of the average ratings for each classification.
|Grade 1/Group 1||104||105||114||124|
|Grade 2/Group 2||101||103||111||117|
|Grade 3/Group 3||98||101||109||113|
A couple of things to note:
The speed figures are quite bunch across all three systems; the average Grade 1 Beyer is only 3 points higher than a Grade 2, and so on and so forth. The Racing Post Ratings are much more spread out, especially at the top end of the spectrum. The differences between each system aren't really a concern (i.e., the average Grade 1 Bris is 105, while Equibase is 114), since the use different scales in their computations. But the relationships between the figures do provide some insight into how each system regards a specific race.
If we were to use median figures instead of an average, the results are pretty much the same. At the Group 1 level the median figures are 103 (Beyer), 105 (Bris), 116 (Equibase), 124 (Racing Post; no Frankel).
The highest figures under each system during 2012:
Beyer: 117 (Wise Dan, G3-Ben Ali, Keeneland, Synthetic)
Bris: 117 (Flat Out, G1-Jockey Club Gold Cup, Belmont, Dirt)
Equibase: 129 (Nates Mineshaft, G2-New Orleans Handicap, Fair Grounds, Dirt)
Racing Post Rating: 143 (Frankel, G1-Queen Anne Stakes, Ascot, Turf; G1-Juddmonte International Stakes, York, Turf)
The is the old adage that if you subtract 15 points from a Timeform rating you get an approximate Beyer rating. I've never subscribed to that notion since Timeform (and Racing Post Ratings) aren't speed figures but performance ratings (with speed as one element of the computation). While I don't think the figures can be extrapolated to give us a "theoretical speed number" for a specific horse, I do think we can use them to determine whether or not a horse running at the Group 2 or Group 3 level is good enough to step up and win a Grade 1 race in North America. In other words, I think the analysis is useful for race-to-race comparisons in terms of depth and strength, but note a helpful in determining what a European shipper will "run to" in the U.S. This is even more true is we're talking about horses making a transition from turf in Europe to dirt in the U.S.