An Intro to Horse Evaluation Part 2: How to Examine Speed Data
By Ryan Trost
It’s been quite a while since I wrote the article How to Think Like a Professional Risk Taker, which was inspired by the news in mid May that odds as we have come to know them would be phased out in the future. The purpose of that article was to give insight into how one can properly manage their money and potential profit in an environment where success is a function of risk versus reward. The simple fact is that in Zed Run, you need money to make more money. By definition the game requires financial input combined with skill and luck to drive a user towards profitability. In an effort to push you, the reader, closer to understanding your assets we recently published both a global Speed Statistics page as well as the same data for each individual horse. At first glance, I completely understand if this chart can be overwhelming. There’s a specific reason for that. We still don’t completely understand exactly what speed data means. There are many theories we here at KYH are developing surrounding the data we have gathered, but we think it is incredibly important for the community to have a baseline understanding of how to interpret the raw speed data before we all start making conclusions.
In Part 1 we discussed two defining attributes to the game, maximum ability and ability deviation. As a reminder, maximum ability is the horse's potential to win any given race against any given field. Ability deviation is the variation we see from horses best performance to worst performance. For as long as I have been a part of the community, the majority of evaluation for these two metrics has come from two sections: odds and placement distributions. However, the inconsistencies we see with horses outperforming their “odds”, the inability to accurately predict the ability to get 2nd, 3rd, etc leaves a big gap in our understanding. Further, lack of explanations for phenomena such as the Spook Chain (described here by our friends at Good Boy Racing) and the fact that no horse over 35 odds has ever won leaves data driven minds like ours to want more! As far as ability deviation goes, U shaped horses, A shaped horses; each leave the community questioning what it all means.
Enter: Race times. If we sit for a second and think logically, why would finishing times exist if they were not extremely important? Why do some of the top horses alive consistently find their way into the Breed Brief and the World Records? While it is certainly too early to make any definitive conclusions, we firmly believe there is a deeper connection between how fast a horse can run and its overall success as a racer. Now that the stage is set, allow me to guide you through your horses speed statistics using our girl: Vanilla Bean. For the sake of simplicity and ease of understanding, we will dive into just one distance: 1600m.
Right away, there is a lot of information to look at. Let’s start with the Mean (average). Even though VB is an incredible performer at 1600m, with lifetime average odds of 6.99, 22.71% WR and 0.848ETH in profit, it may surprise you that her average 1600m time is slower than average. How can that be? Doesn’t that define her as a “below average” horse? Not necessarily. We need to dive deeper.
Let’s look at Standard Deviation. Standard deviation describes the extent of how spread out the data for this distance is. As a general rule, 68% of VB’s race times will fall within 1 standard deviation from her mean. The SD of the average population at 1600m is +-1.51, while for VB it is +-2.45. While on the surface this may not look that important, when you “do math” (shoutout THBC), this puts 68% of all 1600m finishing times between 94.2-97.22 seconds. For VB 68% of her finish times are between 93.49-98.39. Further, if we look at 2 standard deviations, 95% of the average population’s 1600m times fall between 92.69-98.73 while 95% of VB’s races fall between 91.04-100.84.
I know it can be a lot to wrap your mind around, but what we are seeing here is that over any given race, Vanilla Bean has the ability to go either much faster or much slower than the average horse at 1600m. If you look at Vanilla Bean’s iconic U-shaped position distribution, you see this play out over thousands of races. While her maximum ability is quite easily seen by the fact she pulls lowest odds almost every single race she’s ever been in (barring some bouts with Rendezvous Peak), her inability to hold her class rating and constant need to finish last is almost certainly shown by the huge deviation in her racing speed. Across ALL distances, her deviation is on average 159% greater than the population's average deviation. Could this be the key to defining a U-shaped horse?
We list many other data points that can be used for comparison, and diving into standard deviation is currently of the utmost importance in our eyes. However, if you own a descending horse or even an A-shape, the way you look at this information and what you find will be quite different. Take one of the highest class point horses alive: Star Strider
Star Strider’s standard deviation is only 0.26 higher than the average. So the ability deviation is quite normal. However, it’s maximum ability is seen twofold. Once in VERY low lifetime average odds (9.19 in 1600m C1) and second in an average time of 95.08, 0.63 seconds faster than the average. Is this the key to finding a horse with the potential to rise through the ranks? Low ability deviation defined by a small standard deviation of finishing times combined with low odds and a faster than average mean finishing time?
There are different ways to interpret the information that has been provided, and of course there will be disagreements, theories and overall discussions as a result. My hope is that my analysis here serves as an example of how we should all be digesting this data. Becoming more data driven and knowledgeable in our approaches will become more and more important as better horses are born and competition gets smarter. As always, our goal is to help you lessen that gap. See you on the tracks! Look forward to part 3, where we compare horses head to head for even more data-driven analysis.