Recently we decided to add speed ratings to our repertoire. Speed ratings are notoriously inaccurate when based upon the dawdle and sprint characteristic of turf racing. However all weather racing tends to be run at an all out pace so maybe speed ratings would help. For completeness we elected to generate speed figures for all types of racing even if merely to provide a comparison when assessing all weather speed ratings.
For each race distance at each course we produced a linear regression to determine a par time for each type of going. We also felt that races featuring highly rated horses should be faster run than lower class affairs. However we did not have enough data to determine this for each race distance at each course. Instead we determined the effect of race class on par time from a regression of class considering all races at all courses and all distances.
This regression generated a surprising result. It confirmed that higher class races are indeed run at faster pace but not by as much as one would think. This rather suggests that speed figures would prove pretty unuseful as the small difference due to quality of horse is dwarfed by the going effect and indeed the tactics of the race. The limited going variances on all weather may limit this problem particularly as they are also generally run at a true pace.
Having determined a par time for race conditions we then compare the actual race time and use the class regression to determine a speed rating for the race. Eg if the race were run at 1 second faster than par then the previously mentioned class regression would determine a class of horse that would be expected to run at this pace. This resulting class is used as the speed rating for the race and has the added advantage of being transposed onto the same scale as official and our own handicap ratings.
The scale is on pounds so it is simple to correct for weight carried. As only the race time of the winner is provided, the speed rating for trailing horses is determined by the same ratings decrement that we determine for our handicap rating depending upon finishing lengths behind the winner.
The absolute speed ratings proved to be of marginal worth in determining winners. They unsurprisingly prove to be most useful for all weather racing. However more interestingly they seem to provide a useful indicator for chases too.
If we relied on speed ratings to compare horses we would be deeply disappointed. However the speed ratings do provide one very useful tool. They seem to provide a useful indicator that a horse is on good form. So rather than use them to compare horses we look at the trend in speed ratings for each horse to determine whether that horse is improving or declining.
One very worrying observation about race times is that they seem to be little different for professional riders and amateur riders. Its as if the extra couple of stones carried by a typical amateur rider does little to slow down the horse as far as the stopwatch is concerned. We think there is still a little research to be done.
Dave Taylor, January 13th, 2002