Earlier in the week someone from the Wall Street Journal rather cleverly put the Belmont Stakes performances of Secretariat and American Pharoah together to show exactly when the two of them would have finished from the time that the gates opened to the finish.
I am not really one that revels in comparing great racehorses between one another. My feeling is that once they are in the top 5% of the population, they are good, and the difference between them is so nuanced that comparison ruins its relevancy. For example, I'd counter any comparison between American Pharoah and Secretariat by saying that Secretariat had the benefit of being full of steroids (take it easy....they all were at that time) and less stringent drug testing, while American Pharoah had to deal with a track that has significantly more cushion than what Secretariat was running on.
The video did give me an idea though. Some of the more interesting papers on stride mechanics in thoroughbreds were completed by MIT professor Dr George Pratt in the mid 1970's and Secretariat was one of his subjects (along with Riva Ridge). Using high speed camera's Dr Pratt measured Secretariat's stride at 24.3 feet long. Since then there have been a variety of statements claiming Secretariats stride measured such and such, ranging from 22 to 25.25 feet. At Kentucky Horse Park , a scale exists that allows visitors to compare the stride lengths of some of the finest Thoroughbreds foaled in America. There, Secretariat's stride is listed at 24 feet.
There is of course the question of when this stride was taken. The work of Pratt described how Secretariat's stride length varied depending on the period of the race. We know that Secretariat got the last two furlongs in the Belmont in 25 seconds flat (2:24 finish - 1:59 at the quarter pole). Using an open source video software program Kinovea we downloaded the video of the Belmont from Youtube and ran the tape forward to the winning post, or more specifically the stride right after the post where Secretariat had his off foreleg down. We clipped the end of the video off at that point and using the stopwatch feature of the software, reversed the video 24 seconds back to get the point on the film at which Secretariat started the final quarter.
Secretariat took 55 strides to finish the final quarter. That works out at an average stride length of 7.27m or 23.8ft. His Belmont fractions were however 23 3/5, 22 3/5, 23 3/5, 24 2/5, 24 4/5 and 25. He sustained an average 7.27m/23.8ft stride in his slowest furlong when his jockey was looking around to see what was happening and easing up on the horse. His fastest furlong was his second, when Sham blasted out to press him at the turn and they ripped off a 22 3/5 furlong. Again, using Kinovea we edited the video and counted the strides from when he hit that pole to the next to get the quarter. He peeled off his fasted furlong in just 53.5 strides which works out at 7.47m or 24.5ft. If we look at his fastest furlong in the Kentucky Derby we find that it is right at 24.5ft also so its probably fair to say that Secretariat's peak average stride was 24.5ft, which is pretty much what George Pratt came up with (24.3ft). Given that he could generate a 22 3/5 furlong with this stride length and number of strides, he was more of a horse that could really turn them over (good extension/low time of feet on the ground) rather than having a pure stride length advantage over his competitors.
As a comparable, Champion Sprinter Black Caviar had an average peak stride length of 8m/26.25ft (measured from wins in the Newmarket and Lightning Stakes) which is the same average peak stride length of Frankel (measured from the 2000 Guineas and 2012 Sussex Stakes). These two horses had a longer average peak stride and they could string a couple of furlongs together taking just 50 strides a furlong to get it done. That made them impossible to beat in a normally run race. As a comparable for them, even in two brilliant performances by Kingman in the Sussex Stakes and St. James's Palace Stakes he never broke a peak average stride length 7.69m/25.23ft, despite breaking 40mph at one point in the race itself.
So what about American Pharoah? I took the Arkansas Derby and the Kentucky Derby and measured separate peak quarters in each of these races and again counted strides. In both races his best effort was to take 52 in the Arkansas Derby and 53 in the Kentucky Derby, which works out at 7.69m/25.23ft for the Arkansas Derby and 7.54m/24.73ft in the Kentucky Derby. It was interesting that he never seemed to reach what I now know to be his peak average stride length in the Kentucky Derby. Maybe something was bugging him that day.
So here is where it gets really interesting. In the Belmont Stakes American Pharoah completed his final quarter in 24.39 and took 52 strides, so that makes his average stride length 7.69m/25.23ft. His final half was done in 48.71 (37.5mph) and took 104 strides, exactly the same 7.69m/25.23ft. You can watch the video below and count if you want.
So how does that compare to his competitors on the day? According to Trakus data and measuring off the video, Frosted did the last half mile in 49.25 (37mph) and takes 109 strides to do it. That gives him an average peak stride length of 7.33m/24.04ft (he might actually have a longer average stride as he was held up the first furlong which he did in 53 strides and tired over the second furlong which he did in 56). Keen Ice, who actually ran the second fastest average speed throughout the race of 37.1mph, took 108 strides for an average of 7.41m/24.31ft. Honor Code who impressively won the Metropolitan Mile (G1) on the same day took 104.5 strides 7.65m/25.09ft to complete the second half mile in 46.49 (38.4mph).
It is interesting to see that a horse like Secretariat (24.5ft) had a foot shorter average stride length than American Pharoah (25.23ft) but was able to win in faster time. How is this possible? Setting aside the track differences, which would play a part, it is a matter of turnover. As fluid a stride that American Pharoah has, and it is something of beauty, he doesn't break 23 seconds for a quarter in his races. He never has and he never will. His fastest furlong has been 23.43 seconds which means that if he did that in his average stride, he'd be doing 2.2 strides a second. This means that he is fairly consistent in his stride parameters and generally just grinds horses into the ground with a longer average stride (combined with a great cardio and genetics no doubt!). He has the perfect parameters for the modern American thoroughbred. Secretariat on the other hand was able to break 23 seconds, and did so in the Belmont itself. Secretariat had a shorter stride, but could turn them over quickly and keep doing it resulting in him taking closer to 2.4 strides a second.
So the really good horses will generally have an average stride length of 24-26 feet and be able to tun these long strides over efficiently at a fast speed. But these are good horses. What about the average horse?
There wasn't enough races run at Belmont over ten furlongs or more that I could get a video of to analyze, so I looked at races at mile and a sixteenth and more at Belmont over the past month that were non-stakes races. There were 15 races and I analyzed 2 horses in each race, the winner and the horse than ran the fastest quarter in the race.
So here is where the good is separated from the average, at least in North American dirt racing. Peak speed, as measured by Trakus, wasn't that different. Horses of lesser class could frequently run a quarter in 23.5 seconds, the same speed that American Pharoah has run his fastest quarter in, but on average it took them 60 strides (57-62) to do this. That is 8 more strides a quarter to achieve the same speed (38mph) and also means that their peak average stride was just 6.66m/21.85m, representing about a meter difference between the non-elite and elite on average.
Given that the stride length of a horse is linked to inspiration/expiration, that is, there is generally a 1:1 coupling of stride to breath, the more strides that a horse takes the more breathing and work that it has to do to get itself from point A to point B. This makes sustaining a high speed in a race against a horse with a longer stride an impossibility as the body is working harder to obtain the same speed. Taking an extra 8 strides every quarter just to keep up with America Pharoah at the same speed would take its toll very quickly.
So the first thing that separates the good from the bad is stride length. The better ones have a slightly longer average stride (please remember this is an average stride over 400m/2 furlongs, not their absolute maximal stride length). The catch also of course is that you can have a long stride, but not turn it over frequently enough to generate a sub 24s quarter and as it has been previously researched, horses with longer strides need to have a good stride frequency to be effective runners. It's a rare combination to find horses that can sustain 37mph or more, with a stride length greater than 24 feet, for more than two furlongs, which is why comparing elite's isn't relevant as getting over the bar in the first place is the hardest part.
P.S I was asked on twitter by @Rcapper if it is stride length or stride frequency that is more important depending on distance. There has been a fair few papers written on this in human terms, the best being Salo, et al which I have linked to here. Just to do a quick check I went and looked at the Breeders' Cup Sprint last year where from a standing start Work All Week did a quarter out of the gate in 43.34, had a peak speed of 45.5mph and an average speed of 41.3mph which is much faster than American Pharoah generates in his races. Work All Week does that quarter in 53.5 strides which equates to 7.48m/24.54ft, about the same stride length as Secretariat. So American Pharoah and Work all Week have longer than average stride lengths, and cover a quarter in similar stride frequency, the difference is the force that Work all Week places on the ground when he has his feet down is much higher than American Pharoah resulting in greater overall velocity.