Speed Study Begins to Uncover Interesting Implications Between Long and Short Format

[RAyers]

I will be presenting this data in New Orleans. Suffice to say that when comparing a CCI* and CIC* run at the same venue at the same competition, guess which division has a FASTER XC average speed? ;)

Reed

[NeverTime]

Interesting!
I trust you to know how to design a good experiment, so how much did course lengths vary between the two and how did you account for that?

[Gnep]

teaser

[RAyers]

Interesting!
I trust you to know how to design a good experiment, so how much did course lengths vary between the two and how did you account for that?

Good question!

The data is "normalized" on time so distance has no bearing

Give we take data points at any time between every 1-7 seconds, we do not need to account for overall course length. In other words, we collect up to 100 speed data points for each rider on course and then we can compare those, what I call "instantaneous speeds" against other riders. Thus we measure a TRUE average speed. When I smooth the signals, the difference becomes even more obvious.

One thing to note is that the TRUE average speed on course is almost always LESS than the speed listed (e.g. 550 mpm). Almost not rider made that, however, many of them had instantaneous speeds up to 950 mpm.

We had 16 CIC and 10 CCI riders for measurement. We lost data from 4 rides due to a watch that went kablooie (Gnep ;) ).

[JER]

Right you are, Gnep. This is borderline cruel -- you click on the thread hoping for some answers and you're told you'll have to go to New Orleans, like it's the Amazing Race or something.

:lol:

[RAyers]

Right you are, Gnep. This is borderline cruel -- you click on the thread hoping for some answers and you're told you'll have to go to New Orleans, like it's the Amazing Race or something.

:lol:


Hey, man, if I am wearing a suit and presenting to folks at 10am on a Saturday, I want an damn audience. :D

Not to mention I am working the signals analysis at the moment, which is another way to remove the need to consider distance, rider, horse, etc. I will have the RMS signal done ASAP.


Reed

[JER]

Give we take data at any time between 1-7 seconds, we do not need to account for overall course length. In other words, we collect up to 100 speed data points for each rider on course and then we can compare those, what I call "instantaneous speeds" against other riders. Thus we measure a TRUE average speed.


Just for entertainment value -- what was the greatest change in speed between two consecutive data points (on the same rider)?

[JER]

Hey, man, if I am wearing a suit and presenting to folks at 10am on a Saturday, I want an damn audience.

Um, a minor but important correction from someone who used to live in NO:

10 am on a Saturday is bad enough but 10 am on a Saturday in New Orleans is a different animal entirely.

[Kairoshorses]

Hey, man, if I am wearing a suit and presenting to folks at 10am on a Saturday, I want an damn audience. :D

Not to mention I am working the signals analysis at the moment, which is another way to remove the need to consider distance, rider, horse, etc. I will have the RMS signal done ASAP.


Reed

How many TLAs will you be using?! (Three Letter Acronyms)....!

For those of us who will be unable to go because of job and family responsibilities....will there be post-conf. report? And perhaps a youtube video of you in a suit? ;)

[RAyers]

Just for entertainment value -- what was the greatest change in speed between two consecutive data points (on the same rider)?


I am not sure about specific riders, the AVERAGE spread for all riders was 330 mpm. Almost EVERY horse was at 300 mpm over a fence, regardless. That corresponds to conservation of energy (e.g. the forward momentum must be transferred to lifting the horse and thus the forward speed must drop).

I think we may be able to build a "energy usage" model of XC and may be able to quantify a "good round" (high speed, low speed variation, low energy expenditure) versus a "bad round" (high speed, high speed variation, high energy expendature).

[3dazey]

I'm a word geek, not a numbers geek, but even I am captivated by this information. Can't wait for the report from NOLA.

My wonderful barn owners/event organizers extraordinaire are going to the meeting, I will try to twist their arms so at least one of them will be there at 10 a.m.!!!

[RAyers]

How many TLAs will you be using?! (Three Letter Acronyms)....!

For those of us who will be unable to go because of job and family responsibilities....will there be post-conf. report? And perhaps a youtube video of you in a suit? ;)

I promise a full "report." I was just astounded as I waded through the data and the trends just kept getting clearer and clearer. Now be mindful that this is only one competition and that the difference in speed was as little as 26 mpm but as great as 100 mpm in some cases.

The XC was on the exact same course as it was both FEI divisions.

So, everyone, your USEA fees are producing some thing that I hope makes a difference. Thank you for your help!

Gnep, me and John Staples all donated watches and time to this work. Thank you to Kevin B. and John Sheetz for thier efforts to get us funded! Also thank you Desirea Herrera on the software development!

Reed

[flutie1]

"... many of them had instantaneous speeds up to 950 mpm."

Oy vey ......

[RAyers]

"... many of them had instantaneous speeds up to 950 mpm."

Oy vey ......


Don't you mean Oy gevalt! :D

And, yes, the CIC division had the HIGHEST maximum speeds.

And yes, thank you flutie1 for supporting our early efforts!!!!!

Reed

[RAyers]

And let's just say that the average instantaneous speed for the CCI* was almost 26 mpm slower (490mpm) than the instantaneous speed on the CIC* (517 mpm).

[RunForIt]

I'm a word geek, not a numbers geek, but even I am captivated by this information. Can't wait for the report from NOLA.

My wonderful barn owners/event organizers extraordinaire are going to the meeting, I will try to twist their arms so at least one of them will be there at 10 a.m.!!!

I vote for J.

GOOD LUCK TOMORROW!!!!!!

[JER]

I think we may be able to build a "energy usage" model of XC and may be able to quantify a "good round" (high speed, low speed variation, low energy expenditure) versus a "bad round" (high speed, high speed variation, high energy expendature).

I believe this would correspond to a 'good round' in motorsport.

[3dazey]

HaHa RFI, she's not the morning person!!! :lol::lol:

[Gnep]

Holy Mod1, you are wearing a suit, I hope that fact alone will impress everybody, did you lease it ?

[Evalee Hunter]

. . . . Thank you to Kevin B. and John Sheetz for thier efforts to get us funded! . . .
Reed

Ok, I know this is a tiny & unimportant point, but my curiosity is aroused. Is John Sheetz part of the family that has the Sheetz chain of gas stations/convenience stores here in the east? I know that Sheetz family has horses & kids involved in pony club.

[Badger]

Really interesting teaser. Looking forward to the full report. Hope a photo of the suited presenter appears in the Eventing Magazine or COTH.

[flutie1]

[QUOTE=RAyers;3687840]Don't you mean Oy gevalt! :D

What I really meant to say is Holy Shit!
And Badger, I'll take a camera in New Orleans to capture a lasting image of the suited one. It'll be a treasured moment in time!

[gholem]

Good question!

The data is "normalized" on time so distance has no bearing

Give we take data points at any time between every 1-7 seconds, we do not need to account for overall course length.

I think the point of the question was that if one course is 3 miles and another is 5, one would expect the average speed to decrease on the longer course simply because one has to travel farther. If you look at the average pace (measued in minutes/mile) in a race vs. its distance, the graph is essentially monotonically increasing for both humans and horses (except for distances that are so short, top speed isn't reached or only barely at the very end of the race.) If you want to measure it in meters per minute (mpm), which is more common for horses, then the graph would be monotonically decreasing.

Thus one _does_ in fact need to take into account course length. If you don't, you might be saying something as silly as "we've discovered that the average speed of half-marathons runners is higher than that of marathon runners!"

Editing to add: Your argument, in condensed form, is that you are measuring "instantaneous" speed (or close enough) so you don't have to worry about course length. However, it is clear that on a longer course the total average speed is expected to be lower. The naive expectation would be that all instantaneous speeds would be lowered by this amount. The reality is surely somewhat more complicated, but I would say it remains a factor one must consider until proved otherwise.

[gholem]

That corresponds to conservation of energy (e.g. the forward momentum must be transferred to lifting the horse and thus the forward speed must drop).


This argument is completely fallacious. Carried to its extreme, it would imply a horse (or person) not currently moving forward is unable to jump.

Every stride a horse is converting internal chemical potential energy into kinetic energy. When it jumps it can convert MORE potential energy into kinetic energy for three reasons. The first is that a horse is almost never running at top speed when jumping, so any muscles that could be used to accelerate could be used to jump. (To a degree, not all muscles are equally useful for all tasks, which brings up reason two.)
The second is that jumping engages more muscles than simply running - more muscles = more possibility for energy conversion. The third is that jumping tends to engage different muscle types than running - fast twitch as opposed to slow twitch - which generate more power over a brief period of time.

I used to run steeplechase (of the human variety) and you definitely "push off" much much harder the stride you're jumping vs. just running.

I'm not doubting your data, but I don't buy this explanation for the phenomenom you're finding.

[TB or not TB?]

Can someone record the presentation and upload it for those of us unable to attend?? It's sooooo interesting. :yes:

[RAyers]

You are making an argument from an uniformed position. I do not expect you to understand either the analysis techniques or the statistics. These are the same methods used in all manner of signal/motion analysis in human and animal biomechanical research. At the same time, note I said that the XC course lengths are the SAME. They were NOT changed in between the CIC and CCI.

On should NOT look at the absolute speed (max, min, or a single measurement taken at the end of the course) as the basis for comparison. It is the variance in the speeds on course that are important in the statistical comparison (ANOVA with Tukey's or Hsu's MCB). FYI, Tukey's is too conservative given that the samples sizes must be equal. Hsu's MCB shows that the speed differences are statistically different.

You can not simply say a person finished faster and thus had a higher speed, that fails to account for short cuts in path, errors in course measurement (e.g. at one MAJOR competition this year, the GPS data showed the course was off by 200 meters and this was independently verified by several sources including wheels), rider experience, etc. You must look at the running averages of speeds (variance) during the actual course in order to establish true differences. Put it this way, a horse can go slow and finish fast or start fast and finish slow, what is safer? And that is where we are going.

One can look at the speed plots and see that CIC speeds are highly variant (rough is the term used in signals analysis). The CCI speeds are relatively flatter. No AVERAGE DECREASE in speed is seen in ANY plot, thus leaving your assumptions incorrect. I can show this using root mean squared plots of the signals.

Yes, the speed may be decreased on XC for a long format versus a short format. Of course NOBODY EVER MEASURED the reality. Well, now we are getting real data that may show that while the FEI wants to reduces speeds, the long format actually took care of that. At the same time it may mean that course optimum times should be adjusted according to long and short format.

If you have a better idea for signal analysis, let me know. You can see examples of wht the polts look like on Garmin's Motion Base website. Look up "eventing" activities. At this time If you want to meet and discuss the data, I'd be happy to show you that you are mistaken in your assumptions.

Reed




I think the point of the question was that if one course is 3 miles and another is 5, one would expect the average speed to decrease on the longer course simply because one has to travel farther. If you look at the average pace (measued in minutes/mile) in a race vs. its distance, the graph is essentially monotonically increasing for both humans and horses (except for distances that are so short, top speed isn't reached or only barely at the very end of the race.) If you want to measure it in meters per minute (mpm), which is more common for horses, then the graph would be monotonically decreasing.

Thus one _does_ in fact need to take into account course length. If you don't, you might be saying something as silly as "we've discovered that the average speed of half-marathons runners is higher than that of marathon runners!"

Editing to add: Your argument, in condensed form, is that you are measuring "instantaneous" speed (or close enough) so you don't have to worry about course length. However, it is clear that on a longer course the total average speed is expected to be lower. The naive expectation would be that all instantaneous speeds would be lowered by this amount. The reality is surely somewhat more complicated, but I would say it remains a factor one must consider until proved otherwise.

[RAyers]

As for this, are you saying the horse can exert the SAME force on the ground when pushing from a stand still versus when moving? I suggest you go back to your basic fundamental physics/statics/dynamics classes and do a free body diagram. Your statement assumes that a horse in motion exerts the same force on the ground to push up that a horse must do from a stand still. Just like you proved in your personal experience. Thus your assumption that the speed must not drop is wrong.

First, not ALL internal energy is converted to kinetic energy. Much of it is converted to heat in the friction of muscles and tendons, as well as simple metabolic functions to maintain body homeostasis.

Second, you make my point with your last statement. Of course you had to push harder. Newton's second law comes into play and again conservation of momentum. Thus your kinetic energy changes by a factor of 4! GO back to basic physics. You HAD to push harder even as your speed dropped simply to cover Newton's First and Second Laws.

Think about this, a 1 gram bullet moving at 1,000 meters a minute has less momentum but MORE kinetic energy than a 1 kg mass moving at 1 meter a minute. The 1kg has more momentum but LESS kinetic energy. Which one is easier to change direction?

Reed

This argument is completely fallacious. Carried to its extreme, it would imply a horse (or person) not currently moving forward is unable to jump.

Every stride a horse is converting internal chemical potential energy into kinetic energy. When it jumps it can convert MORE potential energy into chemical energy for three reasons. The first is that a horse is almost never running at top speed when jumping, so any muscles that could be used to accelerate could be used to jump. (To a degree, not all muscles are equally useful for all tasks, which brings up reason two.)
The second is that jumping engages more muscles than simply running - more muscles = more possibility for energy conversion. The third is that jumping tends to engage different muscle types than running - fast twitch as opposed to slow twitch - which generate more power over a brief period of time.

I used to run steeplechase (of the human variety) and you definitely "push off" much much harder the stride you're jumping vs. just running.

I'm not doubting your data, but I don't buy this explanation for the phenomenom you're finding.

[Gnep]

I guess if a astronaut would not understand that they would get smashed like a over ripe orange.

But I think the most important part we are going to see in Reeds presentation with a suite, that the overall curve is more even, fewer spikes, from low to very high, the rides are more even.
That means, that far less stress on the horse in the LF, even on the same courses.
Horse and rider are far more even, relaxed and are able to come closer to maintain an even pace.

I bet a CASE of BEER on that, and I don't know the data. Reed is a mean and does not give anything away, him and his rented cloth thing, called suite.

By the way, Reed you broke one of those watches, it did not record the data, shame on you and John, I am deeply disapointed in the two of you, how could you.

[Kementari]

I feel like I get smarter just opening this thread... :lol:

Seriously, though, as someone who is nowhere near New Orleans ;), I eagerly await the full report! :yes:

[RAyers]

I didn't break the watch. Talk to John. ;) Come to think of it, I think that was the watch that was on a rider who fell at NAYRC. Actually, that is another example of our quality control. By plotting all data we can compare measurements between watches to be sure that there is no discrepancy.

gholem, here is a VERY easy way to think of what I am saying, the ABSOLUTE moment you or a horse leaves the ground, you/they are in ballistic flight. Netwon's First and Second Laws are in TOTAL control over you. The forward velocity is always the cosine of the total velocity (vertical velocity is the sin). Thus, it HAS to be less than the forward speed at take off. For example, if the optimum angle at take off is 45 degrees, the forward velocity is decreased by 0.7. However, most horses (as well as you during steeplechase) take off much closer (e.g. 60 degrees or more) so the speed is 50% less than the original forward speed. Of course I am going off the top of my head here.

Here are the kinematic equations for a jumping horse (Moghaddam and Khosravi, European Journal of Scientific Research, 2008)

The horizontal velocity component is determined by:
V cosθ 0 V X =RCG F /T (2)
Whereas the initial CG velocity could be determined:
V0= V X /cosθ = R /( .cosθ ) CG F T (3)
The maximum height of CG was determined by the forward velocity.


Reed

[RunForIt]

HaHa RFI, she's not the morning person!!! :lol::lol:

hhmmmph!!! I DO remember one EARLY March 2nd morning when dear J came bearing a birthday cake in the middle of Roger Haller's jump judge meeting...she has it right at the best times. :D

I would LOVE to be in NOLA ANYTIME...the horsies do love their hay though. AGAIN, GOOD LUCK this weekend!!!!!! AND, lots of fun! :D :cool:

[Gnep]

oh well now you blame John and thats what happens if you give those kids expensive and fragile equipment, they crash, did you sit that kid down and have a serious talk with IT. Its parents should sponsor the program with at least 6 watches.

You know that proofes that those underage laws are absolute stupid. If It had known how many beers that watch had cost, IT would not have broken it.
You can not get better proof for the stupidity of a law.
This next generation of eventers is going to be dreadfull.

Cavemen arise, we need change.

[LexInVA]

GNEP and Reed wagering beer. This could get REALLY ugly. :lol:

[JER]

Hey, man, if I am wearing a suit and presenting to folks at 10am on a Saturday, I want an damn audience. :D


Um, Reed, I just looked at the USEA schedule and it appears that your talk is scheduled for 9:30am. :D

(And this is one of those mornings when you'll want to have enough time to find both of your socks.)

[LexInVA]

10 would definitely be a better time to present. :lol:

[RAyers]

GNEP and Reed wagering beer. This could get REALLY ugly. :lol:

What do you mean wagering? It is always, Reed brings beer for Gnep. Of course I bring my cheetah piss (Gnep's words) so I get it all for myself. ;) Gnep drinks Coors! I work 4 blocks from Coors and Grover Coors is a professor in my department. I'm not drinking that crap. ;)

At least in NO I can get a few Abitas!

Reed

[sofiethewonderhorse]

Good question!

The data is "normalized" on time so distance has no bearing

Give we take data points at any time between every 1-7 seconds, we do not need to account for overall course length. In other words, we collect up to 100 speed data points for each rider on course and then we can compare those, what I call "instantaneous speeds" against other riders. Thus we measure a TRUE average speed. When I smooth the signals, the difference becomes even more obvious


This is very interesting! Sorry we can't make NO this year

Would you still be willing to educate me? More of a math lesson on my part!

'The data is "normalized" on time' 'so distance has no bearing'.

Truly, just educate me:

Here I am on my horse riding on 'Prelim Course A :

Prelim Course A is the same distance as Course B: however Course A is flat, back and forth zig zag course: every fence is visible, there is no terrain to think about..back & forth...this course (conceivably takes less time)

Prelim Course B is an 'up & down terrain'...can't see half of the fences, must ride up, 'jump a fence', ride around a corner, ride into a hay barn, out of a hay barn etc. (this course takes more time)

However, the (published) distances of the 2 courses are the same (true) AND as I understand the rules (Janet, fetch me up here) the speed per meter is the same ratio, right?

>>The data is normalized on 'time'<< ergo, 'distance' has no bearing..(right?)

However, in these two scenarios, distance would nor speed would have been the question: in these two scenarios, the question would have been the effort presented by the terrain. :)

Humbly... Kathryn

Give we take data points at any time between every 1-7 seconds, we do not need to account for overall course length. In other words, we collect up to 100 speed data points for each rider on course and then we can compare those, what I call "instantaneous speeds" against other riders. Thus we measure a TRUE average speed. When I smooth the signals, the difference becomes even more obvious.

One thing to note is that the TRUE average speed on course is almost always LESS than the speed listed (e.g. 550 mpm). Almost not rider made that, however, many of them had instantaneous speeds up to 950 mpm.

We had 16 CIC and 10 CCI riders for measurement. We lost data from 4 rides due to a watch that went kablooie (Gnep ;) ).[/QUOTE]

[Gnep]

it does not change your speed above ground.
the next interesting question would be to evaluate, altitude changes, twists, turns, combinations and how far the jumps are actually spaced on differant courses, same level and than compare the spikes in the curve.
Bu that would need some real serious data colection, 20 watches or so and some real serious data evaluation, money and man hours.

It is a real task, because we ride, coach and ride and we colect the data. I did it ones at show, rode my women and than colected data. Whyle I rode I had some help, but they have not all day.
So what Reed and John have done this year, to get any data, worth talking about is just outstanding and than evaluating that mass of data. Colection is easy compared to the hundreds of ride data and than evaluate them. Remember its done on spare time, rental suite on top of it

[RAyers]

sofiethewonderhorse, the data can not be used to compare 2 different courses as you present for the very reason Gnep states. In the case I am presenting here, the ONLY difference was the format. When I say "normalized" I mean the numbers must be made to provide a statistically testable set (e.g. looking at residuals and making any transformations etc.). In our case, using signal analysis techniques, this is done using a regressive moving average based on time.

The most simple way to analyze speeds without having to worry about the path a rider took, or how the course was measured and all that, is to look at the "instantaneous speed" on course. In so doing we eliminate numerous variables and human errors that can not be accounted for using other methods (start and finish time) because they no longer come into play.

For example, if I look at each rider's speed over every 10 meters, it is very unlikely that the path they took will affect the speed (Distance = rate*time) and so we can compare their speeds directly.

Here is a mind exercise to give you an example of what we are doing:If you want to test this, have two kids race for 20 meters in any direction so each takes a different route. How do you determine who was fast? Thus why we look at speed every few seconds. What if you measure 20 meters but was off by 2 meters? Can you say who was fast? Shorten and lengthen the distances.

This is why races are on race tracks and marathons are closely monitored and marked. In our case, many times you can cut or add a large distances by cutting or taking long routes without changing your OVERALL AVERAGE speed so you may still end up at the finish line at the same time but you may have run MUCH faster or gone slower on course. How do you measure that? Good GP jumper riders know this and use it to their advantage.

At the same time, the large variation in speeds may have a large physiologic impact on EIPH. We can also look at sections in courses or even single complexes. This data will also feed into kinematic calculations for fence design such as done at TRL for frangible pin as this now becomes REAL data and not just assumptions.

And yes we have altitude data as well to do what Gnep suggests. Looking at the time among all of the volunteers previously listed, it is somewhere in the 500-700 donated hours of data collection, software development and data analysis for just the 3 horse trials John and I will report this year.

Reed

[UNCeventer]

I am so glad you did a study on this and will be presenting it! I wish I could be there!!!!

I knew it was not just chance that the number of rotational falls, etc increased when the short format was introduced. I look forward to seeing what you have to say!

[mcm7780]

And let's just say that the average instantaneous speed for the CCI* was almost 26 mpm slower (490mpm) than the instantaneous speed on the CIC* (517 mpm).

27mpm seems like a large difference. At what level is it statistically significant?

I can follow most of what you guys have been saying, but I do get a little lost with some of it! Would be very interesting to know more about what you found once you guys are able to "release" the info to the masses!

[RAyers]

27mpm seems like a large difference. At what level is it statistically significant?

I can follow most of what you guys have been saying, but I do get a little lost with some of it! Would be very interesting to know more about what you found once you guys are able to "release" the info to the masses!

It is not significant using Tukey's HSD but it is when using Hsu's MCB which allows for different sized groups. Tukey's does not. Both were done at 0.05. For a normal 2-sample test, the Z is -2.3497. So it is so close. If one were to look at the simple time over the entire XC course((finish time-start time)/(distance)), there is no difference as we don't know the real distance each horse actually travelled.

The standard error for each division was 8.5 mpm for the CCI and 6.4 for the CIC.

Actually, all of this is helping me get ready for NO and we will try to present this at a vet conference as well.

Reed

[Badger]

I now truly get all the rocket scientist jokes. And, clearly, I am NOT a rocket scientist but I'm glad Reed is.

Sounds like the Cliffs Notes is that the short format stresses horses more that the long format because short format horses reach higher speeds and experience more changes and adjustments and greater variety of speeds.

Is that accurate?

Any idea how the stresses of short format CIC compare to those of a regular horse trial at the same level?

[ss3777]

Wow, very exciting...............can not wait for the cliff notes. Many thanks Men!!!

[PetesMom]

This is so interesting - I can't wait to see the results. Were there any differences between the rider/horse experience between those that ran the CCI compared to those that ran the CIC?

[RAyers]

I now truly get all the rocket scientist jokes. And, clearly, I am NOT a rocket scientist but I'm glad Reed is.

Sounds like the Cliffs Notes is that the short format stresses horses more that the long format because short format horses reach higher speeds and experience more changes and adjustments and greater variety of speeds.

Is that accurate?

Any idea how the stresses of short format CIC compare to those of a regular horse trial at the same level?

I can not say "stressful" because we did not measure stress. That is why we need to tie this data with heart rate monitoring - which we are developing. Some of the vets I have spoken with do feel there may be a correlation between speed and stress though. All we can say, because that is what the data says, is that speeds are higher and more variable in the short format versus the long format. Bensmom and tuppysmom are helping put together the heart rate monitoring.

As for comparing to a horse trial, we need more volunteers and watches. John Staples was the only one there (and he had students in the horse trial) so he was not able to put watches on those riders. I am working on data from NAYRC and COHP HT that may start to answer the horse trial question. However, they are on different days and thus have different weather conditions.

petesmom, I am not sure what you are asking? The data is blinded to me so I do not know specific rider/horse combinations. All I literally see is "Ride 1," "Ride 2," and so forth.

Reed

[LexInVA]

Just thinking about all of this reminds me why I only did research proposals in college and never any real research. The variables are mind-boggling, especially when dealing with live animals and different conditions.

[PetesMom]

I was wondering if perhaps more experienced rider/horse teams tackled the CCI and this is why they have a steadier, less variable speed ride.

[RAyers]

Just thinking about all of this reminds me why I only did research proposals in college and never any real research. The variables are mind-boggling, especially when dealing with live animals and different conditions.

Ah, it gets to the point like buying a car, you pick a few options you really want and everything else gets tossed in the "blackbox."

petesmom, good question but I will leave that to somebody to research. That variable will be measured using competition and USEA records. Realize one thing though, why would more experienced riders want to do the CCI given that there is no long format at the 2-star and above and a CCI* is not necessarily needed to run a CCI**?

[Hony]

Rayers: Thanks for all your hard work on this. What an interesting study. I too am looking foward to the cliff notes!

[vineyridge]

OT, but--

One of the sad losses from Katrina was the very good Dixie Beer. The Wiki article about it says that during the worst times, the building was vandalized and the equipment stolen.

Since Dixie Beer was the perfect accompaniment to Red Beans and Rice, the real New Orleans is still suffering from the hurricane.

[LexInVA]

Dixie Beer is still around though. They just don't make it in LA anymore. It's brewed in Wisconsin and you can get it without any problems in NOLA. They are in the process of getting a new brewery setup in NOLA.

[NeverTime]

Reed,
I was asking for the reason gholem brought up, about how riders going faster over a shorter course than they would over a longer course, and how you'd accommodate for that. I asked b/c the omnibus advertised the CIC course would be 2800 meters and the CCI course would be about 4200. Sounds like they tossed that out the window and had everyone run the exact same course?

((Off the topic of this thread, but that would be either an exceptionally long CIC course or an exceptionally short CCI course, and out of character for what riders in each division would expect to ride anywhere else.))

[Kairoshorses]

27mpm seems like a large difference. At what level is it statistically significant?


I've taken one stats class (graduate Ed Psych), but I read recently something by Deirdre McCloskey which indicated that we've been mis-interpreting "statistical significance" for a long time now. It's in her book...which darned if I can recall the title of now. So maybe it's all sound and fury, signifying nothing.....;)

[Kairoshorses]

I was wondering if perhaps more experienced rider/horse teams tackled the CCI and this is why they have a steadier, less variable speed ride.

I was thinking the same thing. Dang it, I've been thinking about this all weekend. I need more turkey!

[purplnurpl]

Here are the kinematic equations for a jumping horse (Moghaddam and Khosravi, European Journal of Scientific Research, 2008)

The horizontal velocity component is determined by:
V cosθ 0 V X =RCG F /T (2)
Whereas the initial CG velocity could be determined:
V0= V X /cosθ = R /( .cosθ ) CG F T (3)
The maximum height of CG was determined by the forward velocity.


Reed

Seriously guys, get a room.
LMAO!

[RAyers]

Nevertime, good question because upon further review of the data, based on the GPS, the CIC was 4500 meters and the CCI was 3700 meters.

Here is a representative example of a CCI* ride:
http://trail.motionbased.com/trail/activity/7076784

If you go to the list at the far left and click on speed you can see the basic signal plot I use to do the speed calculations. As I download the raw distance and time data I can calculate any istance covered. Based on my look at the second "high speed" section (cross country), the course was 3683 meters. If I add in a few time points on both sides - where the horse is moving faster than 200 mpm, the course is closer to 4000 meters.

Here is a representative example of a CIC*
http://trail.motionbased.com/trail/activity/7077154

NOTE! The distances in Motion Base are in MILES! do the conversion if you need. In this case the rider started the time as they entered the start box. This ride was 4500 meters.

Reed


Reed,
I was asking for the reason gholem brought up, about how riders going faster over a shorter course than they would over a longer course, and how you'd accommodate for that. I asked b/c the omnibus advertised the CIC course would be 2800 meters and the CCI course would be about 4200. Sounds like they tossed that out the window and had everyone run the exact same course?

((Off the topic of this thread, but that would be either an exceptionally long CIC course or an exceptionally short CCI course, and out of character for what riders in each division would expect to ride anywhere else.))

[flutie1]

Reed - This is all fascinating, and I only wish I understood a tenth of it! I'm hoping that you plan to dumb down your presentation at NOLA a bit for the general population who will no doubt be suffering from the excesses that the city has to offer after sundown!

flutie

[mcm7780]

I've taken one stats class (graduate Ed Psych), but I read recently something by Deirdre McCloskey which indicated that we've been mis-interpreting "statistical significance" for a long time now.

From what I've read, she said that while something may not be "statistically significant", it may still be important. I know that at least in the human sciences (taken 4 stats and experimental research courses, undergrad and grad), statistical significance is still considered an important part of the results to show that the difference probably isn't due to chance. We can still talk about how the change is important but it's hard to get anywhere without showing that it wasn't due to chance. But I wonder if her view (from an economics standpoint) differs from that of, say, a psycholgist or biologist, because of the different variables studied. Does that make sense? (I hated stats. :lol:)

[RAyers]

Reed - This is all fascinating, and I only wish I understood a tenth of it! I'm hoping that you plan to dumb down your presentation at NOLA a bit for the general population who will no doubt be suffering from the excesses that the city has to offer after sundown!

flutie

I promise it is a lot more clear when you have the plots and images in front of you.

As for statistical significance, it is a measure that when comparing numbers, the difference between them is not just due to chance. It does not tell you why the numbers are different.

[gholem]

You are making an argument from an uniformed position. I do not expect you to understand either the analysis techniques or the statistics.

It is true that I wasn't in uniform at the time when I formed my position. You should, however, expect me to understand both the analysis techniques and the statistics.


At the same time, note I said that the XC course lengths are the SAME. They were NOT changed in between the CIC and CCI.

On should NOT look at the absolute speed (max, min, or a single measurement taken at the end of the course) as the basis for comparison. It is the variance in the speeds on course that are important in the statistical comparison.


Now, we have established in a later post that the course lengths were, in fact, different. I never suggested we only needed to look at the absolute speed. I suggested that the instantaneous speeds needed to be modified if you are comparing them between two different groups that ran over different length courses. I made this suggestion based on the fact that average speeds over longer distances decrease.

It is possible, and would prove to be interesting, if the speeds approaching jumps are consistent across varying length courses, with the main differences in speed coming in the flat galloping sections.



No AVERAGE DECREASE in speed is seen in ANY plot, thus leaving your assumptions incorrect.



And let's just say that the average instantaneous speed for the CCI* was almost 26 mpm slower (490mpm) than the instantaneous speed on the CIC* (517 mpm


I'm not sure what you meant in your first quoted statement, since you clearly found a difference. Now, the question is whether the average instantaneous speed on the CCI* was slower BECAUSE it was longer or due to some other factors.

In other words - you have shown that there is a statistical speed difference between the two groups - that's great. Why? Because of something related to the long/short format, or simply because the courses were different lengths?

I don't know the answer, but it is something that needs to be addressed.

I do think a very simple way of slowing down cross country runs would be to simply make the courses much longer (without increasing the number of jumps).

[gholem]

Think about this, a 1 gram bullet moving at 1,000 meters a minute has less momentum but MORE kinetic energy than a 1 kg mass moving at 1 meter a minute. The 1kg has more momentum but LESS kinetic energy. Which one is easier to change direction?

Reed

A minor point - but they have the same momentum. And, yes, for a given force, the acceleration it causes a small mass will be bigger than for a large mass. F = ma. I don't see your point here.


As for this, are you saying the horse can exert the SAME force on the ground when pushing from a stand still versus when moving?


I was not suggesting this. What I was saying is that one cannot simply invoke conservation of momentum as you described it to account for a decrease in the horses horizontal velocity when it jumps.

My example's purpose was to point out the flaw in your argument (basically the missing term in your "conservation of momentum" - generation). Since a non-moving person or horse is perfectly capable of jumping and they have no momentum, we cannot simply say their horizontal velocity (and momentum) decreased to account for their increase in vertical velocity (and momentum).

There was "generation" of momentum inside the control volume which occured because chemical potential energy was converted into mechincal kinetic energy by the muscles.

Now when a horse is running normally, the forces acting might look something like this (mostly horizontal)
^
|
------->

I'm arguing that when it (or a person) jumps, while running they can make the force vector look something like this:

^
|
|
|
|
|
------->

Which would preserve the same horizontal speed and momentum they've had all along in addition to accelerating them vertically. This is a greater total force and would correspond to a greater perceived exertion...

[gholem]

Here are the kinematic equations for a jumping horse (Moghaddam and Khosravi, European Journal of Scientific Research, 2008)

The horizontal velocity component is determined by:
V cosθ 0 V X =RCG F /T (2)
Whereas the initial CG velocity could be determined:
V0= V X /cosθ = R /( .cosθ ) CG F T (3)
The maximum height of CG was determined by the forward velocity.


Reed

For everyone that might be interested here is a direct link to the cited article:
http://www.eurojournals.com/ejsr_22_1_12Morteza.pdf

To sum up the paper, they film a horse going over a jum p from the side and use this to determine the position of its center of gravity as function of time. From this they use the equations above (which are just the basic equations of projectile motion as seen in high school physics) to determine what the velocity of the horse must have been the MOMENT AFTER it left the ground. It must be the moment after it leaves the ground because before that there are still forces acting on it besides gravity and the assumption of these equations is that gravity is the only force.

This could actually be an interesting technique to use on cross country and steeplechase jumps to determine what exactly is happening with .05sec of resolution. Maybe after I turn in my disseration in a couple months I'll have time to do this, it seems interesting. Sort of carry on the school legacy since we were the first ones to use photographs to analyze horse gaits.


The forward velocity is always the cosine of the total velocity (vertical velocity is the sin). Thus, it HAS to be less than the forward speed at take off


Here is where there is disagreement.

Lets say that V1 is the velocity of the horse in the penultimate stride, V1_y is relatively small compared to V1_x. Now the horse touches the ground in its last stride and jumps. Its new velocity V2, now has a large V2_y component due to more "momentum generation" in its last stride - it exerted more force in the vertical direction while still exerting the same horizontal force it has been all along. So V2_x is still the same as V1_x (or maybe it is smaller or even bigger, I can't say without measurements, my only point was that your application of conservation of momentum without considering the generation term was incorrect.) So the magnitude of V2 is indeed greater than V1. It isn't really any different than if the horse decided to accelerate and start running faster as oppoesd to jumping.

[Snapdragon]

I'm trying very hard to follow all of this, but all I can think of is those dreaded "If a train leaves point A at 70 mph . . . " problems :) and my brain is instantly scrambled. So glad I decided to go into writing/editing even though I have a science degree!

As you hash this out, I'm eager to see explanations and results for the layperson! Thanks for all your effort!

[RAyers]

You should, however, expect me to understand both the analysis techniques and the statistics.

Do you? You have no idea what the data looks like and you have not even come close to showing you have any familiarity with the statistics or signal analysis techniques.


Now, we have established in a later post that the course lengths were, in fact, different.

However the slower speeds were on the SHORTER course. So how is it that the average speed decreases? Look the the basic plots do you see a trend? If you apply RMS to the plot you see relatively flat or INCREASING speed. Hence, to me, you are commenting from a UNINFORMED position.

Again, if you want to actually help, then give me your email and I will happily send you data to analyze.

It is possible, and would prove to be interesting, if the speeds approaching jumps are consistent across varying length courses, with the main differences in speed coming in the flat galloping sections.

What specifically do you suggest? It is a little hard to have a flat course. I am not sure where you are, but few courses have flat galloping sections including Galway. Go to the web site I provided and look at the elevation change.

I'm not sure what you meant in your first quoted statement, since you clearly found a difference. Now, the question is whether the average instantaneous speed on the CCI* was slower BECAUSE it was longer or due to some other factors.

In other words - you have shown that there is a statistical speed difference between the two groups - that's great. Why? Because of something related to the long/short format, or simply because the courses were different lengths?

If I was on your committee and you made an observation like that when I already stated the CCI* was the SHORTER course, I would be highly suspect of your ability to pay attention to your research.

Reed

[gholem]

However the slower speeds were on the SHORTER course. So how is it that the average speed decreases?


Sorry, I'm not an eventer, the conversion from CCI/CIC to long/short format was not clear to me. Now, if I have this correct, CCI is the long format with a shorter cross country phase and slower speeds.

Anyway the explanation is not hard to come by - the exertion of the previous three phases causes them to slow down by more than the amount one would expect them to be faster if they hadn't done the three previous phases.


Look the the basic plots do you see a trend? If you apply RMS to the plot you see relatively flat or INCREASING speed.


Here, I'm guessing you mean during the course of a specific run, which is interesting, but not entirely relavent to the point I've been trying to make.


Again, if you want to actually help, then give me your email and I will happily send you data to analyze.


Sure, I'll send you PM when I'm done with this reply.



What specifically do you suggest? It is a little hard to have a flat course. I am not sure where you are, but few courses have flat galloping sections including Galway. Go to the web site I provided and look at the elevation change.


I meant flat as in no jumping efforts. But this just brings up another variable that must be controlled for between the different groups.

[Gnep]

gholem guessing is not good enough, the cci is the Long Format with a far longer Cross Country.
The rather interesting point for me as an eventer, how many LF made the time compared with the CIC, despited having a smoother curve and less average speed.

For me that would be the point of interest.
Knowing Galway, they probably left just a few jumps out, but ran the CIC course still all the way to the second water, just a couple of loops more for the CCI. Basicly the same course, same terrain.

To the question if the more experioned people were running the CCI, I doubt it, I would say that they run the CIC for qualification purpose.
Anyody who comes from the LF times knows that D compared to an equal HT X-C, is far more relaxed, even that it is far more demanding. naturaly the added distance always helped to relax, even going the same speeds ( officialy ).


Gholem, welcome, join us, do you have an engineering degree I can use, statics maybe or any other structural design engineering ?

[Badger]

As for comparing to a horse trial, we need more volunteers and watches.


How much are the watches and how many do you need?

[CookiePony]

Reed - This is all fascinating, and I only wish I understood a tenth of it! I'm hoping that you plan to dumb down your presentation at NOLA a bit for the general population who will no doubt be suffering from the excesses that the city has to offer after sundown!

flutie

And for those of us who will be reading at home. I am having a little trouble following, but probably because I just went out for margaritas with an old friend who is in town for the holiday. Otherwise these equations and things are second nature for me. :D

[subk]

Otherwise these equations and things are second nature for me. :D
Yeah me too. In fact that "V cos? 0 V X =RCG F /T (2)" stuff is exactly what's going through my mind about 2 strides out.

[Gnep]

Badger I got some of the watches on e-bay for around 180, 10 to 20 more would be nice and natural including the people who help.


that Math stuff, only when I am stoned

[RAyers]

It is not just watches we need. We need folks willing to work the competitions, as well as folks to help with data entry conversion. We have tried to automate it as much as possible but there is still a lot of number crunching.

Gnep, yes, gholem is an engineer.

Reed

[RunForIt]

which competitions are you targeting? That info would help folks know if they can help you out gathering data.

[gholem]

gholem guessing is not good enough, the cci is the Long Format with a far longer Cross Country.

Gholem, welcome, join us, do you have an engineering degree I can use, statics maybe or any other structural design engineering ?

Ok, this is getting confusing. In Nevertime's post he says that the CIC (short format) was advertised as 2800m and the CCI (long format) as 4200m. Which would agree with Gnep's assertion that the cci has a "far longer cross country". However, Reed's measurement suggest that the CIC was in fact 4500m and the CCI was 3700m. It seems pretty incredible to say a course is going to be 2800m and then have it turn out to be 4500m.

As far as the CCI having more people make the time than the CIC, the first thing I would like know is how the optimum time is determined? Could it be that the CIC's time was simply harder to make? That is what the data suggests to me, since they were going faster on average over a longer course...

I am an engineer, but my specialty is more fluids and computational methods. I'd be happy to talk about whatever you're thinking about, I may have some ideas/thoughts.

[Gnep]

well first you have to get you figures right, 4000 for the CCI and 3800 and something for the CiC.
the Omnibus was a missprint, a 2 instaed of a 3.

Numbers, well that is Reeds department, I am mechanics.

Fluids is great, everything is anyhow fluids, Reeds department.

What else ?

[NeverTime]

I am NOT getting in the middle of Scientist Smackdown 2008 :D but I am very curious about what you said --
a CCI course *should* be significantly longer than a CIC or horse trials course. Even if the omnibus wasn't right/was a misprint, it is bizarre to me that it was the other way around in real life. Could there be something wrong with the data, if that's what it is indicating, or did Rebecca actually make CIC riders do a longer XC course than the three-day riders?!

[RAyers]

gholem and Nevertime,

XC courses are measured using a wheel and walked by the course designer. They can make any path they want and the riders must guess what that path is when they measure it. In many cases you will see discrepancies up to 200-300 meters between folks walking courses (hence why I think wheeling a course for optimum time is a waste). The wheel can be off, the CD may make take a long route while all of the riders make a cut (thus shortening the course) without anybody knowing.

Thus this is why we look at instantaneous speed. It precludes this type of discussion and the need for general assumptions.

Optimum time is based on a chosen speed. That depends on the course designer and technical delegate etc. There are suggested ranges of speeds in the rules. I believe in this case, it is between 520-550 mpm.

But remember, the data we have does not depend on the optimum time nor the length. If you want to look at time penalties per division, some one else can do that. That will tell you if it was harder to make the time on the CIC vs. the CCI.

As for the CIC being longer than the CCI, look at the two rides I presented in this thread. They are representative of all of the data I have. The CICs are consistently longer in distance so was the course mis-wheeled? Was the wheel out of calibration? Did the CD make certain assumptions that the riders didn't or vice versa? Who knows.

This is why we HAVE to get away form this idea of looking at the entire course to make comparisons on speeds. Too many assumptions must be made and thus the result is as bad as having no data.

Also remember, this is the first time this is being done. The data shows that so far the "advertised" length of the course never matches the reality. CDs are human and can not be perfect. It can be they had no idea what they were doing (not that they were ignorant but that they were adding distance error upon error) when they were measuring the course.

Here is another problem, folks are relying on a single person to be perfect in their measurement of course length and so you all automatically assume they are perfect without considering the reality of the data. I am NOT saying the GPS is perfect but the data is consistent ride after ride. The statistics begin to show that there are other issues to be considered.

Reed



Ok, this is getting confusing. In Nevertime's post he says that the CIC (short format) was advertised as 2800m and the CCI (long format) as 4200m. Which would agree with Gnep's assertion that the cci has a "far longer cross country". However, Reed's measurement suggest that the CIC was in fact 4500m and the CCI was 3700m. It seems pretty incredible to say a course is going to be 2800m and then have it turn out to be 4500m.

As far as the CCI having more people make the time than the CIC, the first thing I would like know is how the optimum time is determined? Could it be that the CIC's time was simply harder to make? That is what the data suggests to me, since they were going faster on average over a longer course...

I am an engineer, but my specialty is more fluids and computational methods. I'd be happy to talk about whatever you're thinking about, I may have some ideas/thoughts.

[Gnep]

I think there is a little bit of a mix up in the three letter words. Even with a knew course designer, there are 3 fixed position at Galway, D-Box and the 2 waters ( I doubt that they changed it so much that those fixed locations have changed ).
Basicly the X-C loops around behind the racetrack. The HTs N to P go to the first water and than circle back, I-A - FEI go to the second water and than circle back.
For the CIC and CCI 1 star that makes it between 3800 and 4200 meters, depending how far the loop at the second water is.
Gholem Optimum time is measured by the Speed of each level, example 520 mpm for the 1stars ( CIC and CCI ) and the length of the course, for the 1 star around 7 minutes.

Question for Reed, are the data from the fall or the spring show, I am getig a little confused here, because the fall had the CCI LF and the CCI SF, they ran over the same course and OT was 7:02
Nice to see that they had such a strong field in the LF, good suport from the riders

[RAyers]

Gnep, it was the Fall competition.

Reed

[Kairoshorses]

It is not just watches we need. We need folks willing to work the competitions, as well as folks to help with data entry conversion. We have tried to automate it as much as possible but there is still a lot of number crunching.
Reed

Reed, I'm in Area V. I'd be willing to help out. I can travel to some outside of the area, too, if they're not TOO far out.

[kb]

I will be in N'awlins and I will be more than happy to take pictures of Reed in a suit for all of you to see......of course I didn't know he owned a suit so I am curious to see where that is coming from :lol:

I asked Reed to let me preview the presentation and make sure it makes sense to the non-mathematically inclined....surely my advanced degrees in education are good for something in this deal (and someone needs to proof the grammar).

And remember, some of us hear this stuff from Reed almost everyday....and it still doesn't make sense. Sometimes the scariest question I can ask him..."what are you doing?"

[gholem]

Gholem Optimum time is measured by the Speed of each level, example 520 mpm for the 1stars ( CIC and CCI ) and the length of the course, for the 1 star around 7 minutes.


So, it would seem to me, that if the optimum time is determined based upon an assumed length of the course (which Reed has pointed out, is not very well known and apparently quite different from what the course designers thought) divided by a known pace (in terms of distance/time), there could be quite a bit of difference in the actual paces to which the optimum times correspond.

Ex: CCI Course length thought to be 4000m / 520 mpm =~ 7:40, but the actual course length was 3700m so the pace really was 3700 / 7.7 = 480mpm

The CIC was thought to be 3800m / 520 mpm = 7:18 but the actual course length was 4500m so the pace was really 4500 / 7.3 = 616mpm!

In which case it isn't at all surprising the CCI had much less dificulty making the time than the CIC did.

[Gnep]

If you go by the result sheets of the fall show, than the courses had a OT of 7:02 at 520mpm, that lets you calculate the from the Designer meassured length of the course.

Naturally the Designer does very often walk a very differant line than the riders like to use when they ride. Mostly the designers lines are far tighter, which means shorter distance the differance can be 200 to 300 meter or more.

For me as a rider that was the most interesting part in using a GPS Watch this year, the official couse length, than what I measured and than what I rode.

[RAyers]

The courses were the same e.g. it was a CCI* with and without steeplechase per the omibus. As Gnep noted both courses had the same optimum time. I do not know where the 2800 meter course length comes from.

My statement that there was a CIC is incorrect and I will admit it comes from my mind where a CCI without steeplechase is a CIC.

Reed

[Gnep]

Reed you are looking at the Prelim Data, 2800 is the Prelim course.

[TLA]

The length of the course should not be a mystery, although it will almost always differ from that published in the Omnibus, given the lead time needed for publishing. It is the responsibility of the CD to ensure the correct measuring of the course. If I find that competitors are differing greatly from my measurements, I will measure again. I use the track that I think the average horse at that level will take, as in Show Jumping.
From the perspective of a CD, putting the speed data with the actual jumps (or jump shape ) would be very useful information.

[NeverTime]

The courses were the same e.g. it was a CCI* with and without steeplechase per the omibus. As Gnep noted both courses had the same optimum time. I do not know where the 2800 meter course length comes from.

My statement that there was a CIC is incorrect and I will admit it comes from my mind where a CCI without steeplechase is a CIC.

Reed

A ha! Now I understand why the distance/course was the same! Now it feels more apples-to-apples and it does become quite intriguing, why one group was faster than the other. :yes:

[canterlope]

It is the responsibility of the CD to ensure the correct measuring of the course.No. While the CD is responsible for the layout measurement of the cross country course (EV175.2.a), it is the PGJ, in consultation with the TD, who is responsible for ensuring the correct measuring of the course (EV171.2.a).