Speed ratings - driving me mad!

[Faxy Fowler]

I have spent the last few weeks putting together a standard time for all courses in the UK and Ireland. I am absolutely delighted with the results, they are based on races on good ground, 2 year old races excluded, all calibrated to the median performance of a runner over the course at Class 5. The numbers are not the issue. What is however now happening is, when I try and rate a meeting, particularly National Hunt, you will get 6 of the races run to say around 99-100% of my standard, then a single race sticks out at 94-95%. My hunch is that perhaps a fence or two may have been missed in that race. How would you adjust that race without buggering up the others? I could do each race individually and adjust it, but I am trying to backrate races one month at a time from 2009, to get four years worth of ratings in time for the forthcoming flat season. One way I thought of was to get a flavour for how far from Official Ratings the race is and adjust accordingly, so if you get OR SPEED 102 148 105 144 113 141 103 140 101 138 Clearly then the race has been run at the false pace - presumably it is wise to bring each rating down by about 40 or so, or by an amount relative to what is achieved over and above OR in other races? Any thoughts welcome. Presumably the default answer would be these things simply take time, and that's fine, but I wondered if any excel whizzes or people with experience of ratings adjust those races that are clearly falsely run?

[The Saint]

Good Post and good luck with the speed ratings. I can't think of a way to back date personally but u have pointed out one of the many hazards with speed ratings over the sticks. For me, they are a much more effective tool on the level. Not very helpful I'm sure. Hope to be proven wrong in time.

[Faxy Fowler]

Re: Speed ratings - driving me mad! Thanks TS. I am glad its not just me. As you are touching on, I am more than happy with what comes out for flat, all weather and bumpers, the predicament is with other races. I suppose you could have a closer look and find that fences are omitted, what I am trying as mentioned is to avoid the 7 race card that gives percentages: 101%, 99%, 100%, 98%, 93%, 101% and 100%. In the above, the 93% sticks out like a sore thumb, so either it was run at a ridiculous pace, which is almost implausible, or something was done to change the course. If you decide on the latter, change that to 100% and everything else is moved out with it. If I change that to an average of the others it could be unrepresentative, I suppose the only thing I can think of is to take the average of the median of the others and the standard time for the race, then "asterisk" the race so the rating can be confirmed as sketchy. I'd probably do it for any race which is 5% under or over the median, (though it is clearly the unders that are more likely to be the problem). Hope I'm making sense...