Speed Ratings

CompuBet Speed Ratings are the best in the world. Why? We analyze over 75 North American harness tracks and approximately 15,000 to 17,000 race lines every day.

Speed ratings require three very important ingredients, and any attempt to develop accurate speed ratings without all three would be futile and misleading. Here’s how CompuBet tackles the daunting task of producing accurate speed ratings:

1. We start with the base mile time for each contestant. Obviously if the race is contested over a distance other than a mile, it must be adjusted. To do this we use standard formulas developed using a number of similar distance races and comparing the individual contestants’ times against those same contestants’ times derived from one mile events. As is always the case when comparing times in this fashion, we only use clean, competitive lines. We’ll talk more about that later.

2. We then find the Daily Track Variant. To do this, we compare each horse’s actual time (using only those horses beaten less than five lengths, and excluding qualifying lines) to those same horses’ “standard times”. These standard times are based on each horse’s average mile time over the past three weeks at that track, again using only clean performances with beaten lengths less than five, and for the purposes of the standard time calculations, only fast tracks are used.

As an example, let’s say a horse has an average time in the last three weeks at today’s track of 120 (that means 120 seconds, a two minute mile). Remember that number was derived by averaging any and all lines at that track for that horse, in the last three weeks, when the track was “fast” and the horse was beaten less than five lengths. Now if the horse goes the mile on today’s track in the time of 121 (2 minutes, 1 second), then based on that single horse, we would say the track is “off” or slower today by one second.

For each day’s race results, we repeat this process for every available horse who meets the criteria outlined above. We hope to find at least 30 horses who do meet the criteria and then average their individual “variants” to come up with a final variant for that day. Remember, we cannot use horses that have not raced at today’s track in the previous three weeks, there would be no “standard”. And we cannot use trotters, we cannot use qualifiers, and we cannot use horses beaten more than five lengths. Now you understand why 30 to 40 horses is a good and acceptable number. It is worth noting that there are days when the track condition changes during the race card. We have provisions for making adjustments as needed for this circumstance.

And so now we have the second ingredient, the Daily Track Variant. This is added to each horse’s actual base mile time, and the result will serve as the horse’s speed rating for that track.

But how do we convert a speed rating from one track to find out how fast the horse is expected to go at a different track?

3. We use our “Inter-Track” Variant to determine the speed a horse having raced at one tracks will achieve at a new track.  The way this is done is to compare as many horses’ times as possible at one track to those same horse’s times at another track. If you can find a hundred horses who each raced at two particular tracks within the last 6 to 8 weeks (always trying to avoid seasonal weather fluctuations), and they all were beaten less than 5 lengths, and they were all pacers, and you exclude qualifying races, and you only use fast tracks, then you will get a pretty accurate Inter-Track Variant.

As an example, if you find that a group of 100 horses average 120 at one track, and 119.5 at another track, then it is safe to say that the second track is a half second faster than the first. Therefore, if you have a horse who has been racing at the first track, and has an adjusted speed rating (after the Daily Track Variant has been computed) of 116 (1 minute 56 seconds), that horse would be expected to perform at a 115.5 rating at the second track.(*)

This procedure is a bit more complex when there are not sufficient data samples (horses) for an accurate comparison. In that case we must use relative comparisions, such as comparing track “a” with track “b” and then track “b” to track “c”, and combing those variants to arrive at a comparative variant from track “a” to track “c”. The more indirect our analysis the less reliable the resulting speed projections will be and although we feel this is the best information available, we do recognize it is not perfect and our Predictability Index will be reduced accordingly.

Obviously these fundamental concepts are well known to handicappers. The difficult thing for the human handicapper is to make useful estimates as to the actual numbers involved for every horse in every race. In fact it would be impossible to do it accurately.

CompuBet does all the above for you as well as hundreds of additional calculations per horse involving the many other handicapping factors, such as post position, driver, form, class, etc., etc. It takes millions of calculations to produce a report. Our customers do not have to spend hours with the Program: They spend their time on betting strategies and money management. And CompuBet even makes those jobs easier by providing our New and Powerful “Predictability Index” for every horse, and our “Betting Patterns” reports which show you the success rates of 18 different types of bets at each track using the CompuBet Selections (based on all reports and all races over the recent few weeks) 

footnote (*) The speed rating we publish generally is determined by taking the two most recent lines when possible, and averaging them after performing the three steps outlined here. Obviously we will not use lines that are considered “poor” since those lines would not provide an accurate assessment of the horse’s true speed capability, and we use up to 3 recent lines when necessary. We also use up to ten recent lines to determine base-line speed and class ratings.