Another stats person here, trained as a biostatistician, but these days my skills mostly have to do with smashing datasets together, and distilling them.
1 Like
The foundation of dressage is based on establishing a partnership between horse and rider.
A lasting partnership creates the greatest trust and understanding between horse and rider, for the benefit of both.
If a trainer/riderâs goal is only to be successful in âshow dressageâ with an eye on the competition calendar, thatâs their choice. The training will be done accordingly. By such- and- such month/year,
the horse must be at so-and-so level so we can qualify for, or attend, such-and-such show. Iâve seen this happen with a trainer friend who has deep international experience. The clientâs horse ( 5-7 years old) was bought in Europe, and when he arrived, he was moved up the levels quickly, and skipped one or two of them, seemingly just to test the brakes and steering. Although the horse could do the higher level movements in the show ring, his muscle development was still at the lower levels. He wasnât ready, and the scores reflected it. He never made it to the competition stratosphere they wanted, and sadly the horse is no longer competing. And we can recall the controversy years ago when a top rider publicly announced they were buying a made horse just to compete at a large international show outside of the U.S later in the same year . Riders who had been training for years to reach that show werenât happy.
When a trainer/riderâs goal is to develop a long lasting partnership with the work coming slow, steady, and correct, the show calendar isnât used as a training clock. Itâs completely inconsequential. May be theyâll go to a show someday, maybe not. Itâs the development and maturity of the partnership between horse and rider, that are the priorities, and the quality of the training will reflect that. Thatâs the traditional dressage concept.
Going back to Verdades and Udon, we can add Sabine Schut-Kery with Sanceo. Itâs truly remarkable when trainer/riders can do bothâŠhave that lasting partnership and earn long term success at top shows.
2 Likes
That makes perfect sense. But I guess I would still count that - they are buying young unproven horses and bringing them up within their program (assuming, of course, that their young horse riders and working students get the benefit of learning directly from Carl and/or Charlotte).
Yes, of course they do, but neither of them are riding them themselves. Young horse rider is its own job. Lucy is their head girl as well.
Meanwhile Iâm bowing out as once again the purists have determined which way is the âcorrectâ way to do dressage.
Dressage actually just means training, and all this âyou must have a partnership forged from years of forming a bondâ does not represent how horses are produced or competed today.
But the developing GP champion this year actually does ride her own, so there you go, everyone can calm down.
12 Likes
@atlatl and @quietannâŠperhaps you can opine on the ability of the scoring system to discriminate between rides.
I continue to torture the data. Since this thread started on the scoring in the FOC, here is the interval plot for the scores
https://equestrian-hub.com/public/show/184925/competition/dressage/2415459
Here is the interval plot for the 95% Confidence Interval for the scores in the class in the link above.
How well can we say that the scoring system has the power to resolve differences between quality of rides?
| Factor | Mean | StDev | 95% CI |
|---|---|---|---|
| 1 | 72.76 | 2.320 | (71.58, 73.95) |
| 2 | 71.794 | 0.583 | (70.607, 72.980) |
| 3 | 70.97 | 1.405 | (69.784, 72.156) |
| 4 | 70.882 | 0.560 | (69.696, 72.068) |
| 5 | 70.735 | 0.771 | (69.549, 71.921) |
| 6 | 70.647 | 1.502 | (69.461, 71.833) |
| 7 | 69.588 | 0.669 | (68.401, 70.774) |
| 8 | 67.411 | 2.045 | (66.225, 68.597) |
| 9 | 66.676 | 0.910 | (65.490, 67.862) |
| 10 | 66.293 | 1.439 | (65.107, 67.480) |
| 11 | 65.764 | 1.296 | (64.578, 66.950) |
| 12 | 62.764 | 0.903 | (61.578, 63.950) |
I did as well. Really encouraging to see how they worked as a pair.
4 Likes
These are your goalposts and guardrails. They are be no means universal.
In dressage, we celebrate the role of schoolmasters in teaching riders the movements and how things feel when done correctly.
Many people leverage the show ring to test their own nerves, and the horseâs nerves- their relaxation, submission, and obedience. They want a disinterested third partyâs feedback on what they see. They want to get that test back to study the comments against how it felt. after all, everyoneâs an Olympian at home 
12 Likes
And some people make their living showing and competing horses. This results in building a reputationâŠor a brandâŠwhich the higher it is, the more sponsors and more money and higher the price for the horses they sell.
2 Likes
Just got back from there yesterday. Iâve been out of the professional horse world for a few years but my previous employer wrangled me in to coming out there to help with the horses they brought.
I think yâall are reading a lot more into it than there actually is- some trainers have a good well of young stock to draw from and bring up (and are good at it), some trainers prefer to start with a more known entity, owners give and take rides on whims, some are super lucky to have/ be supported by owners that just want to enjoy watching a really nice go out there and compete.
9 Likes
I donât remember if the ânerd herdâ was on this board or another. That group published a paper on their statistical analysis of dressage scores. IIRC, the gist of the paper was that there was a lot of variability.
Anyway, given that there are multiple factors that go into each and every score, Iâm not sure that statistical analysis can be applied in a really meaningful way. Of course that all comes down to the actual question weâre trying to answer. Is the problem well-defined?
I think this because, in theory, each movement is scored based on multiple factors, the details of which we donât know. We just know the final score. These factors include not only the essence of the movement, e.g. was the volte in the right place and right size and was the horse properly aligned, but also the gaits, harmony, etc (the old collectives) and general impression whatever that means. Also, anything that may happen after the volte on the way to the next movement may be reflected in the score.
Until we can get in the judgesâ heads and gather information about each factor they thought about that contributed to the score, I think we canât really tell much more than different judges value different things. Also, I think we all know that judges do miss stuff sometimes and frankly, some are just plain bad judges.
All that said, years ago I won an All-Breeds end of year award and the difference between me and the 2nd place rider was in the second decimal. A chemist friend, you know who you are, mentioned there was no statistical difference between our scores so really it was a tie to which I replied that from a pure mathematical perspective, one number was in fact larger than the other, so I did win.
3 Likes
Thanks for your reply. Yes, I am familiar with the work of the COTH Nerd Herd. Their work was published in a peer-reviewed journal of the American Statistical Association.
WellâŠthat is your opinion. Perhaps it is the difference between a mathematician and an engineer practicioner who uses statistics.
Ever hear of Andrew Pole? He was a statistician for Target who predicted a teenager was pregnant before she told her parents. Forbes had some great articles before Target cut off all contact with the press.
Pole did this by data mining and determining what variables could be used to predict a behavior. Same data mining and statistics tools are available to study dressage.
This is a conclusion about a cause for âpoor judging.â Then again we can ask the question, "What constitutes âpoor judging?â
We havenât asked any questions about or explored the dressage judging system enough to say anything about what is source of âthe problem.â Eg., we have not done a measurement systems analysis of dressage judging to be able to quantify the underlying issues.
If Target can quantify the variables that define the behavior of pregnant women, figuring out the variables that go into dressage scoring should be easy.
Here is the New York Times article that broke this story
Bottom line is that there are established, industry standard tools that are available to study dressage judging. It is a matter of will and desire to study the questions.
2 Likes
Havenât read the articles but I suspect Target was paying closer attention.
Cheers!
2 Likes
Statistics is one piece of data modeling, mostly used in determining what variables are correlated and then how well a model fits the data/how well it can predict an outcome. Machine learning takes this to another level as it evolves the model and can create data where there is none.
In the case of the Target situation, you are seeing both things. The Target case is taught in a lot of graduate level modeling courses for Analytics. Predictive scoring is one of the most common analytical problems to solve; predicting a due date is altogether different because it doesnât actually know what any customerâs due date was until they filled out a baby registry. How you solve the second problem is why itâs often a case study.
Meanwhile the legalities of employers using analytics like this is one of the things I work with often.
3 Likes
I am glad there are other data nerds on here.
I agree that the legalities of data mining is a current topic du jour as computers and algorithms get more powerful. I donât want to derail the thread topic on the ethics of data miningâŠbut I agree it is a real issue.
The point is that you can use data to create a model. This sort of study is done for test validation in all manner of disciplines. Dressage just has chosen to do hand wringing about dressage judging instead of actually studying the problem.
1 Like
I believe that Poleâs work was attributed to have increased Targetâs earnings by many $$$millionsâŠso I am sure they âwere paying attention.â
1 Like
Do you think it is because TPTB are afraid of losing the âartisticâ aspect of the discipline? IOW, are they concerned it will be viewed more as a pure âsportâ than an âartâ? I know next to nothing about statistics and data modeling but is it truly possible to turn the evaluation of âartâ into a purely statistical exercise? For instance, can hard and fast rules be developed that dictate exactly how an art critic evaluates Michelangeloâs David? Would any evaluation that falls outside the ânormâ be considered invalid - and that critic subject to scorn and ridicule or even sanctions? But that may not be an apt comparison, because âDavidâ is a static object. Perhaps a more appropriate subject would be Swan Lake - and would the rules vary depending on who choreographed it, who the dancers were, the venue, the orchestra, etc.?
6 Likes
Iâm not sure âDressageâ (g) thinks there is a âproblemâ
6 Likes
You may be rightâŠor they may be taking the position that if they ignore any questions on the judging system, there there is no problem.
2 Likes
WellâŠdressage IS an Olympic SPORT. So I think that ship has sailed. The only place where art is brought in is in the musical freestylesâŠand that is in the one score for artistic impression.
Jorst may ride each day, but she also runs Kastel, which isnât exactly a small company. Sheâs very driven and works her ass off every day.
4 Likes
