mean ol’ dressage: post mortem edition

So! If you read yesterday’s post, you’ll know that Doozy and I rode two dressage tests at a schooling show over the weekend. Technically two different tests — Starter and Training 1 — but practically identical in pattern. 

It’s my impression that our efforts in each were reasonably equivalent, give or take. And while there’s no video of the Starter test, we do have past evidence of general consistency across multiple tests. 

All that to say, you’ll have to take my word for it that our execution between the two tests was relatively similar, but the final scores differed by more than 3%, a remarkably high margin all things considered. 

tests with more scoring opportunities (ie larger count of total movements) are less susceptible to having one blown movement wreck the whole score

As is my occasional habit, I decided to dig into the numbers from both of these tests. I want to understand exactly why the scores varied so much, especially if there are takeaways that can be integrated into my schooling sessions. 

Tho I’ll also admit right off the bat: part of my hypothesis is that the actual design and scoring structure between the two tests is responsible in part for the score variance, beyond just how we performed and executed each movement. 

Training Level Test 1
score before collectives: 59.47% ||| collectives alone: 59.29%

Because even tho the pattern for these two tests is nearly identical (slight variation in location of canter-trot transitions, trot-walk transition, and free walk — short instead of long diagonal), the scoring structure is a bit different. 

Most notably, eventing dressage tests did away with the suite of collective scores in 2022 — reducing the overall judge’s impression score down to a single metric: Harmony. Dressage tests, meanwhile, retain the original block of 5 ‘collective’ marks that give judges an opportunity to add nuance to their impression beyond just scoring the movements as presented. 

trotting: needs more of this more quickly after transitions

The Training 1 test also breaks out individual scores for working canter, so the canter receives 4 total scores compared to the Starter test’s 2. On the flip side, the Starter test scores the medium walk both before and after the free walk, for a total of 3 scored walk movements compared to Training’s 2. 

For Doozy, the differences in canter scoring opportunities didn’t have much effect in these tests — our overall canter averages across both tests varied by just 0.3%. The addition of that second scored medium walk, however — the walk right before transitioning back to trot — definitely does not work in our favor.

Starter Level Test
score before collectives: 57.31% ||| collectives alone: 50.00%

Doozy has a generally lovely free walk that so far most judges have been happy to mark as a highlight. But the tension increases as soon as I start picking the reins up, so our average walk scores for Starter are more than half a percentage point — 0.6% — lower than average walk scores for Training. 

The biggest variation in scores by far, however, came in the collectives. While we earned just one movement score of 5.0, the judge (imo) crucified us with a second 5.0 (with x2 coefficient, natch) for our single overall collective in the Starter test — which had an enormous impact on final score. Ouch. 

free walk: proof that the mare can learn to play the game

Charting both tests by looking at the running average as the tests progress kinda makes it more clear how those slight differences can have cumulative effects on final score. 

For example, in the below chart, both tests start out identically — a conservatively ‘satisfactory’ 6.0 first impression upon entry, then ho-hum continuing on thru the first trot and canter circles, then the wheels start coming off a bit as we shift into downward transitions to arrive at the walk work. 

progression of overall average score as test goes on

The free walk is a high point in both tests, but the bifurcation in running average happens at the point of that third walk score in the Starter test. Then we see relatively parallel downward trends as we work back toward the second canter (low points in both), before the final exit. 

Here is another difference in test pattern — Starter does that weird broken line from the corner letter back to X then to G, whereas Training does a traditional 10m half turn onto the center line at A. Considering all of our canters in this test happen in the corner after CL, it’s no wonder that Doozy is maybe thinking about canter instead of halt at this moment — whereas a true CL turn is another one of those ‘trick’ patterns that horses learn to recognize.

click for full size if you want to see comments

Anyway. Again, after the movements themselves, you can see in the chart above that the Training test running average perks back up again with the addition of collective marks, whereas the Starter test lands with a thud.

Naturally it’s any rider’s tendency to dismiss that sort of thing as just a harsh judge biased against tense red mares lol. Which, eh, ymmv. 

Realistically, tho, there are still some useful tidbits for me to chew on from this analysis. Namely: transitions. Dear lord. Up, down, all around. Our transitions are ultimately what’s getting us into trouble, and the design of the Starter test leaves virtually no place to hide from them.

final halt: everybody is happy when it’s over lol

It’s something I’ll admit to not focusing on a ton in our schooling sessions too, since at least in my experience doing a lot of transitions can significantly amplify tension and fizziness. So we probably won’t just start doing 8 million in a row, or something crazy like that. 

Instead, I might try to identify some baselines — like counting strides between a transition and when the gait feels “established.” How many strides does it take from when we pick up canter to when the canter feels balanced and organized? Right now that answer is… Well. According to these tests, more strides than can fit into a 20m circle LOL. 

And how many strides does it take from when I first ask for a canter-trot transition and when we are in fact trotting in a steady rhythm? I think focusing on shortening those distances, while incrementally increasing the number of transitions we do, could be a big help. 

Bc… Well… Obviously my hope is that we’re just about finished riding this test forever. But we’re not quite there yet. And I’d like our last few attempts at it to keep getting better lol.