Old Salem Farm Grand Prix Analysis (Spring 1. May 15, 2022)

 

The Grand Prix during the first week of the Old Salem Farm Spring series was a CSI3*, designed by Alan Wade.  It was held during the day on the new sand international arena at the OSF Venue. 

We saw a total of 47 faults in Round 1.  The most faulted obstacle was a Combination-Mid Oxer, (Vertical-Oxer-Vertical combination).  This jump was responsible for 17% of faults in Round 1.  

The next most faulted jumps were the A element of the same combination and a single other oxer (the jungle themed oxer with green rails immediately preceding this combination).  Each respectively caused 13% of faults.

In total, the Triple Combination was responsible for 38% of Round 1 Faults; moreover, if we also include faults from the Double Combination at 10AB, Total Combination Faults = 57% of Round 1 faults!  (This is significantly higher than the average for Combination Faults in a High Performance Class which we are seeing around 37%.)

All jumps except 2, an Other Related Oxer, were faulted at least once. 7 other jumps were each faulted only once.  All in, of the 16 course elements, half were responsible for 85% of the faults in Round 1!

The class was won by Jordan Coyle on Centriko Volo.  

JumpClear Member Statistics April

 

Overall, we saw significantly less activity from JumpClear members during April, as might be expected during what many use as a transition month between the winter and summer season.  

Only 1 in 4 members were active.  For those who did compete, Average Rounds per Month were lower than usual - 3.5 versus an average just under 5.0.  75% of horses competed in just one competition.  

Members who showed averaged a 38% Clear Round Average.  This is significantly below the baseline.  

Looking at Fault Source metrics, the most faulted course elements continue to come from the "other" Jump Detail with Other Vertical driving 28% of Faults and Other Oxer 17%.  The next most faulted were Combination-In Vertical (14%) and Liverpool Vertical (14%).  

Keeping in mind we are considering a fairly small number of faults, a few more detailed points of note:

• 70% faults were at Related jumps (versus 53% in the baseline)
• 72% of faults were at Vertical jumps (versus 49% in the baseline)
• for classes 1.40 and above, 47% of faults were at Other; 41% were at a combination element. For classes below 1.40, 42% of faults were at Other.  Just 8% were at a combination element.  (Note: this is an exaggeration of a trend previously discussed that Combinations comprise a higher percentage of faults as jump height increases).  

Performance Goals Part 2: Using Data to Make a Plan

JumpClear is a great tool for identifying your best opportunities to improve your results and setting measurable goals.

Yesterday we gave an example of using JumpClear data to identify an opportunity for improvement and set a goal.  But now what?  What do you do with all these numbers?  

(To be clear, JumpClear data isn't going to be a silver bullet.  The goal is to give a rider & trainer information to identify a challenge and be part of their overall tool kit for horse management and training). 

Fault Source Analysis encompasses 6 metrics.  We only used Jump Detail to set our goal but the data around the other metrics tells us a lot more about this particular challenge.

Using the same example - which full disclosure, is one of my horses - this is what happens when you zoom in on just Skinny jumps:

100% of faults are off the left lead

100% of faults are with a front leg

60% are at a single technicality

Hmmm...We now have a pretty consistent defined situation to problem solve and maybe come up with some mental cues. 

We also have a really concrete and bite-sized goal.  Recalling the "Model" course jump distribution, Skinny Jumps were calculated to be 4.7% of total jumps over 12 months.  If we use some other data around the average number of classes JumpClear members contest, that comes out to 24.8 Skinny Jumps.  If we assume they are evenly spread across leads, we're down to 12.4 Skinny Jumps to make a particular focus.  

Over a year of jumps, that feels like we've narrowed it down to a pretty achievable goal!

Using JumpClear to Set Performance Goals

JumpClear uses Fault Source Analysis to identify opportunities for horses & riders to improve their performance by showing the areas where they have a higher percentage of faults.  

Recently, we've been expanding this analysis through comparing an individual member's performance to average or baseline fault distribution.

We use this approach as the basis for setting unique performance goals for horses: the logic is essentially finding fault metrics where the horse significantly differs from average performance and calculating how it could be performing if its faults were more like the average.

Here's how it works (this is a real example):

                                                               🠉

We start with a horses faults for the past 12 months.

                                                                                        🠉

We compare that to the Average Fault Detail distribution.  This is where we can identify Target Areas where a horse's faults differ significantly from the average.  Here, we highlight Skinny jumps which are more than 5x average!

                                                                                                           🠉

Then we multiple the Average by the horse's Actual Total Faults for the past 12 months to get the expected number of faults for the jump detail in our Target Areas.

                                                                                  🠉

Finally we calculate the difference between the Expected and Actual Faults.  This tells us the change in faults possible if the horse improved its Target Areas to the Average.

Here, this horse could decrease its Total Faults by 15% if she improved this one fault area!

Comparing Faults Types Between Different Horses

In the last post, we analyzed the baseline distribution of the Jump Detail fault metric and compared it to a hypothetical modeled course.  Our conclusion was on average, faults aligned well with the frequency different types of jumps occurred.

However, there are significant differences between individual horses.  These different fault distributions reflect the behavioral asymmetries of horses & riders and are the basis of JumpClear's Fault Source Analysis methodology.  

                                                                                            🠉

Standard Deviation measures the variance within a range of numbers. Here, it is used to capture the difference between the fault patterns of the individual JumpClear member horses.  We see the SDs are all large compared to the absolute values for the Baseline values.

Likewise, when we calculate a Fault Range of + or - one Standard Deviation, there is huge variance in possible outcomes.  There are 7 metrics alone where horses could vary from no faults (a negative value is calculated) to up to 10%! 

Going forward, JumpClear will use this approach of comparing a horse's fault distribution to the Baseline and Standard Deviation Range as part of identifying each horse's unique opportunities for improvement.   

Are Some Jumps Faulted More Frequently? Analyzing Fault Distribution for Showjumping

 Are some jumps faulted more than others?  Is it because they're "harder" or just used more frequently?  We looked at the data to find out.

The Approach:

We designed a model course to arrive at an approximate distribution of jump types.  The Course was based on an average of 10 different jump configurations to capture different Combination Types, and use of elements like Liverpools, Skinny Jumps, Planks, Walls and Water.  

We calculated the percent each element comprised within the model course and compared that to Baseline Data for JumpClear members for their Fault Distribution for Average Round 1 Faults over a 12 month period.  

Findings

Overall, the Model Course aligns very well with Average Faults.  This data suggests that faults - viewed as an average across multiple horses and rounds - occur roughly in proportion to the frequency of the type of course element.  Otherwise said, no particular jump is a bogey causing a hugely disproportionate amount of faults.

We know, however, that faults vary significantly between individual horses.  We'll look more closely at this in the next post.

Showjumping Course Analysis Combination Elements

As the JumpClear database grows, it's exciting to deliver more insights based on total member data.  Here we take a deeper look at Combination Faults...

Starting with the basics, Combinations - across all elements - made up 34% of Round 1 Faults for JumpClear members over the past 12 months.  They were a much smaller percent of faults in the jumpoff - just 15%.  

There numbers are interesting because they suggest that combination faults come down to simple math.  Combinations are faulted roughly in proportion to the percent they make up of total jumps on the course*; they are not - in fact - horrible bogey obstacles that stand in the way of every horse & rider's clear round.

(If you want to check that math: the average first round has between 12 - 14 numbered jumps with somewhere around two doubles and a triple or two doubles, so 19 to 21 obstacles of which combinations comprise 6 or 7.  The average jumpoff has about 7 - 9 numbered jumps with one double, so about 9 to 11 obstacles of which combination elements make up two).  

However, we see that combination faults differ quite significantly by division.

As a general rule, the lower heights have fewer combination faults.  This is largely - but not entirely - driven by the impact of the "mid" element and likely reflective of the limited use of triple combination in the 1.30 and 1.20 national classes.

By a slight margin, the most rails come from the Combination In Element.  Across all elements, faults are split nearly evenly between oxers and verticals.