Ice scope data shows zero correlation between height & GOE

Edwin

СделаноВХрустальном!
Record Breaker
Joined
Jan 5, 2019
How does 'Axelcam' work?

Aren't all data extracted from the 2D image when the skater is 'in range' or is there a stereoscopic camera with Doppler radar we don't see?
Or is there the classical triangulation used by three cameras in sync and computer power after careful calibration by proof skaters?

I'd really like to see a stick figure animation of the jumps with accurate blade position on 'exit' and 'entry' of the ice surface to assess the linearity of the jump and the gain/loss of speed by cheating, UR or OR etc. And with the known mass of the skater, some calculation of the kinetic and potential energy involved.
 

Blades of Passion

Skating is Art, if you let it be
Record Breaker
Joined
Sep 14, 2008
Country
Russia
Amplitude on jumps is just one aspect of a tons of things that go into a performance, and so taking the time to score more objectively based on jump height/distance/landing speed is more accurate, sure... but it's drawing attention to such a trivial thing.

Amplitude on jumps is not trivial whatsoever. It's way more difficult to jump big and it was always supposed to be a significant part of considering technical merit, not to mention the visual appeal.

The judging and scoring system is bad right now, this aspect of jumping needs to carry much more weight. A height/distance calculator is a nice tool to have, although judges don't need it to do a good job. We can already see the relative size of jumps well enough and a good judge should already know how to use the differential within their scores (ie, something like a tiny jump being 0 GOE on average, medium size jump +2, huge jump +4, biggest jump ever +5).
 

CanadianSkaterGuy

Record Breaker
Joined
Jan 25, 2013
Amplitude on jumps is not trivial whatsoever. It's way more difficult to jump big and it was always supposed to be a significant part of considering technical merit, not to mention the visual appeal.

The judging and scoring system is bad right now, this aspect of jumping needs to carry much more weight. A height/distance calculator is a nice tool to have, although judges don't need it to do a good job. We can already see the relative size of jumps well enough and a good judge should already know how to use the differential within their scores (ie, something like a tiny jump being 0 GOE on average, medium size jump +2, huge jump +4, biggest jump ever +5).

I mean trivial compared to things like calling an element correctly, and assessing rotation. Jump amplitude is something that can be subjectively done for now, as the actual effect it has on a score is minimal. If a skater jumps 0.6 m high and 0.7m high the next time what does that mean... they get 0.004 extra points? And we should invest in the tech and audiences should have longer waiting time just to have that calculated and incorporated into the final score? We're getting overly meticulous - read: a little nuts - here....

It's fun for comparing skaters at a superficial level and getting "stats"... but there's so many factors that determine jump amplitude and what constitutes "very good height or distance'. Like I said, a hypothetical 3 foot tall skater jumping 3 feet high is more impressive than a 7 foot tall skater jumping 4 feet high, but the Ice Scope stats would just make it seem like the 7 foot skater is the better jumper.

It's one sample size too. I feel bad for anyone who didn't do their 3A, and got left off the list. :(

Can't wait for the new "Element-Matches-the-musical-structure Scope" which pinpoints exactly how "on" the music a jump was executed, and how many milliseconds off a skater is. And then we can create stats and debate about who has the most musicality on their jumps! :laugh:
 

elysium

On the Ice
Joined
Apr 11, 2014
If objective characteristics like the distance, height, even UR can be measured accurately in real time, it would definitely help the objectivity of the scores. Different judges view the same jump from slight different angles, even this simple fact can affect the perception of a jump. Although humans are good at judge length and speed, especially experienced judges, there are many well-known visual illusions like Müller-Lyer illusion and Jastrow illusion. So a portion of the score based on objective measurements can be really helpful. Then the judges can complement with judgement regarding the difficulty the entrance and the quality of the landing etc. I don't think the waiting time would be much longer either, measuring and passing the result through an algorithm to get a combined score can be done lightning fast these days.
 

Blades of Passion

Skating is Art, if you let it be
Record Breaker
Joined
Sep 14, 2008
Country
Russia
Jump amplitude is something that can be subjectively done for now, as the actual effect it has on a score is minimal. If a skater jumps 0.6 m high and 0.7m high the next time what does that mean... they get 0.004 extra points? And we should invest in the tech and audiences should have longer waiting time just to have that calculated and incorporated into the final score?

Why do you always sidestep everything that has been said and only give straw-man arguments? The effect on the score is not and should not be "minimal". If judges were properly assessing jumps and actually giving out GOE with a significant emphasis on their amplitude, then they would notice the clear visual difference between a 0.6m jump and a 0.7m jump, giving the former +3 GOE and the latter +4 GOE.
 

Baron Vladimir

Record Breaker
Joined
Dec 18, 2014
Why do you always sidestep everything that has been said and only give straw-man arguments? The effect on the score is not and should not be "minimal". If judges were properly assessing jumps and actually giving out GOE with a significant emphasis on their amplitude, then they would notice the clear visual difference between a 0.6m jump and a 0.7m jump, giving the former +3 GOE and the latter +4 GOE.

Height and distance are not the only merrit of a jump in figure skating, we need to say. Its also important for the jump to be part of the program (transition in and out) and part of the music. Some jumps with too much height and distance may look too wild in the air/off axis and hard to control on the landing so they would not get good body position bullet and good landing bullet. Also, for the jump to be percieved as a good element efforthlessness is also important, and some may argue that eforthless through out is the most important GOE bullet regarding jumps (as it is for any other element in figure skating)... Now, its very common from the stats we saw that heighest jumps don't have the biggest distance and jumps with biggest distance are not the highest. Just rare skaters were able to combine those two (in pairs SP for example T/M were the only one to combine very big height and very big distance in their throw jump according to the stats ... they got +5 GOE across the board for that element tho)
 

Blades of Passion

Skating is Art, if you let it be
Record Breaker
Joined
Sep 14, 2008
Country
Russia
We know there are other aspects of jumps. Those things are also not currently being judged properly. Things like textbook takeoffs, delayed rotation, and finishing rotation entirely in the air should be given more credit than they are now. Amplitude in a jump should be the #1 consideration for giving +GOE though. The entire definition of a jump is launching yourself into the air. If aren't doing that to begin with, then you're not really jumping (although I'm fine with smaller jumps for specific choreographic purpose). Extremely difficult entrances are also undervalued right now. People do all kinds of busy work into and out of jumps these days, but there are very few truly outstanding entrances/exits - like for example a hydroblade into a jump, or doing a Triple Salchow directly out of a spin exit, or doing a sit spin directly from a jump landing.
 

hanyuufan5

❅*:・。.✨
Medalist
Joined
May 19, 2018
Ooh, this is a great idea!

Unfortunately, I don't think enough data exists, but in order for an analysis like this to be more correct, one would need an average of many Axels from many different competitions. Negative GOEs and possibly even falls would likely have to be included, possibly outliers eliminated, and the trend very well might be nonlinear or correspond to the average of the stats rather than the individual ones.

Also, it could be that higher, longer, faster jumps tend to come from shallower edges or less complicated entries or not be landed as cleanly. Likewise, the Axelcam can't measure edge depth, difficulty of entry and exit, etc. like a human judge can.

But I bet there would be more of a correlation with averaged Axelcam data.
 

Ladskater

~ Figure Skating Is My Passion ~
Record Breaker
Joined
Jul 28, 2003
Kurt Browning says "it's the landing that impresses him"
 

cohen-esque

Final Flight
Joined
Jan 27, 2014
For the Ladies event, you get a similarly strong association between GOE and PCS (r=0.63) for the solo 2A. But if you plot GOE vs PCS for all jumps in the Ladies event with at least neutral GOE, you get a notable but much weaker correlation (r=0.37).

Just noting the danger of trying to draw even broad conclusions based on only one type of jump. I suspect that if you included negative GOE, or only examined clean 3Lz jumps, you’d again get relatively different numbers. (And be pretty much nil in terms of actual explanatory power, sadly. That would require things that I’m not willing to do on spare time.)

Also, unlike the men and their 3A, when ice scope is data is available, distance had a moderately good correlation with GOE for clean 2A among the Top 20 women (r=0.47).
 

Miller

Final Flight
Joined
Dec 29, 2016
Also, it’s been a long time since I’ve taken physics but doesn’t someone have to sacrifice height for distance and vice versa? I’d think you’d need to take both data points into account. Someone jumping high but not far or far but not high may not be receiving scores as high as someone who is utilizing both height and distance.

I'm wondering if it might be possible to do a scatterplot of time in the air vs GOE. Also show the skater's starting number on there - that would really show if the later you go the higher GOE you get for the same element (other bullets notwithstanding of course). Some skaters may jump high + short, others long + shallow, but I would have thought it would be time in the air that would really end up impressing the judges.

As with Shanshani's original scatterplots (thanks for all the hard work!), I would imagine you'll get a pretty much horizontal 'best fit' line, but it would be interesting to see. Also some sort of graph of GOE vs starting number. I would expect that to be a relatively shallow ascending line, but clearly not horizontal like the current scatterplots, and so show the lack of correlation very clearly.
 

samkrut@mail.ru

Medalist
Record Breaker
Joined
Mar 26, 2014
While distance and speed are more or less easy to measure, height is a different thing. I don't know how they do it. The easiest and the least correct way is the minimum distance between the blade and the ice. With such an approach those will be penalized whose legs are straight and who point the nose down. If the legs are not straight and the blades are parallel to the ice the height will be automatically bigger which makes little sense. Russian split jump will be the highest of them all then. More than 1 meter for those who can make a perfect split like Jason Brown. Although can one say that it is really the jump with the height over 1 meter?

The correct way to measure the height of the jump is to determine the elevation of the center of gravity. But how could one do that?
 

zounger

Medalist
Joined
Jan 18, 2017
The correct way to measure the height of the jump is to determine the elevation of the center of gravity. But how could one do that?

I was describing the simplified physics behind a reliable with good enough precision in the following post a couple of months ago.

Briefly, the calculations are based on the kinematic equations. More specifically on their simple form that we learn even in school physics. The results are quite good and close the real values.

So you consider the body to be a point mass (the center of mass of the skater). When the skater jumps his center of mass is doing a parabolic trajectory. This comes from the kinematics laws. In this trajectory the skater at some point is reaching his maximum height, let's call it H.

In kinematics, there is the law of the independent component of movement. So if the body is moving in the horizontal direction (the distance that is covered on the ice in his jump direction) and vertical the (the vertical move to reach his higher point of jump). You can consider the jump as a composition of this two movements.

Since this two movements can be considered independently we consider that the skater is just doing a straight vertical jump.

When the skater is in his higher point (H cm above the ice) he starts to fall. This fall we approximate it as free fall. The free fall equation is:

H = g x (t^2)/2 where is g is the gravity acceleration 980 cm/sec^2 , t is the time that is takes for the fall.

Because of the symmetry of the parabolic trajectory, we also consider that the time that it takes to skater to reach his higher point is the same as the time for the fall from the heighest piont to ice.

So we can calculate this time. You go by frame by frame and you check when the skater leaves the ice and when he is reaching the ice again, lets that time be T. Half of this time is for the fall (T/2). Now you use the formula above and you have

H = g x ((T/2)^2)/2 = g x (T^2)/8

and you find the desired height of the jump :biggrin:.

It's very simplistic approximation of the body movement but it works relatively good.


So if you have the technology to dermine when the blade leaves the ice and when touches back, you can calculate it with some nice precision.

One solution can be to introduce sensors to the blades (accelerator, velocity, gyroscope, weight, etc. ) which can give real-time measurements. You can solve the rotation problem as well with those sensors. Angle/position the blade leaving the ice against the angle/position on landing. The problem is you need to standarize the blade(and boots) on boots so that everyone has the same sensors on them plus the reliability that no one will "hack" the measurements. Technology wise I think the problem is trivial with the current means.
 

RobinA

On the Ice
Joined
Nov 4, 2010
I, for one, hope it never comes to this in figure skating. Measuring jumps? Ugh. Maybe we need to develop some new skating disciplines for the fans who want a numbers game and leave the current discipline to those of us who want to see beautiful skating. We have speed skating. Now maybe skate jumping judged by computers.

It seems to me that if you want numbers, almost every sport out there relies on numbers. Figure skating is for those of us who want something else. Please let us have our little corner, small as it is, of the sports world. And please, no "It's a sport" arguments. There is nothing in the definition of "sport" that says it has to be judged using numbers.
 

samkrut@mail.ru

Medalist
Record Breaker
Joined
Mar 26, 2014
I was describing the simplified physics behind a reliable with good enough precision in the following post a couple of months ago.

In theory time=height makes sense. In practice with 24 frames per second which means that 1 frame is equal to 0.04 sec the precision for a 50 cm jump will be about 1 cm. Hence, one could distinguish between 50 and 52 but not between 50 and 51.

The main problem is that the trajectory is not ideal and those who land with straighter legs will be somewhat penalized: their center of gravity will be higher on the impact.

If the sensors were introduce, I think, the key advantage would be the precise calculation of revolutions in the air, the second one in terms of meaningfullness would be take-off edge quality. Current ice scope is just a toy. The judges can see with their eyes if the jumps are big or small. And to get the necessary number of revolutions one can just rotate faster.
 

NaVi

Medalist
Joined
Oct 30, 2014
I'm a little baffled why airtime wasn't measured as that seems like it would be the most balanced measurement between people of different heights... but I could be totally wrong about that. It would probably wouldn't require any special equipment to measure.

I, for one, hope it never comes to this in figure skating. Measuring jumps? Ugh. Maybe we need to develop some new skating disciplines for the fans who want a numbers game and leave the current discipline to those of us who want to see beautiful skating. We have speed skating. Now maybe skate jumping judged by computers.

It seems to me that if you want numbers, almost every sport out there relies on numbers. Figure skating is for those of us who want something else. Please let us have our little corner, small as it is, of the sports world. And please, no "It's a sport" arguments. There is nothing in the definition of "sport" that says it has to be judged using numbers.

I previously wrote a post on a new format using jump metrics.

Air and Art - A Competitive Format For Extending Careers
https://www.goldenskate.com/forum/s...rt-A-Competitive-Format-For-Extending-Careers

But in an age where ladies are doing quads, I think it would be a good idea to make it easier for those with big triple jumps to compete.
 

Shanshani

On the Ice
Joined
Mar 21, 2018
For the Ladies event, you get a similarly strong association between GOE and PCS (r=0.63) for the solo 2A. But if you plot GOE vs PCS for all jumps in the Ladies event with at least neutral GOE, you get a notable but much weaker correlation (r=0.37).

Just noting the danger of trying to draw even broad conclusions based on only one type of jump. I suspect that if you included negative GOE, or only examined clean 3Lz jumps, you’d again get relatively different numbers. (And be pretty much nil in terms of actual explanatory power, sadly. That would require things that I’m not willing to do on spare time.)

Also, unlike the men and their 3A, when ice scope is data is available, distance had a moderately good correlation with GOE for clean 2A among the Top 20 women (r=0.47).

Do you have a link to this data? I would love to see it. If indeed there is a stronger correlation for 2As, that may lend some credence to the theory that in judges minds, all 3As and quads check the height and distance box merely for being 3As and quads. I would disagree with that (clearly, Yuzuru's 3A is outstanding, and I would also like to give a mention to Boyang's 4Lz, and I think both should be rewarded for that. And in my opinion, even Samarin, whose skating is often validly criticized, deserves extra credit for the size of his quads.), but it would be an interesting addition to our knowledge of how judges behave. I agree that in general more data would be helpful.

I'm a little baffled why airtime wasn't measured as that seems like it would be the most balanced measurement between people of different heights... but I could be totally wrong about that. It would probably wouldn't require any special equipment to measure.

I'm not convinced that height isn't a perfect proxy for air-time. In an ideal physical system, those two quantities are perfectly correlated, and while another poster did bring up an interesting point regarding air-resistance for rotational motion, I'm not convinced it's a significant factor in vertical motion.
 

hippomoomin

Record Breaker
Joined
Oct 30, 2012
Height, distance and landing speed are only three factors out of many, so I am not surprised there is no correlation. It ignored air position, difficult entry, landing position, tano/rippon... Karen Chen's jumps for example have great distance and very good height but she just cannot fully rotate.

If there is a correlation (suppose it is statistically significant) between psc and goe, it absolutely does not mean psc "predicts" or "determines" goe, because pcs is given after goe is awarded. The opposite can be argued: high goe indicates judges are willing to give high psc.

If there is anything I want to be scored by a computer, it is the number of rotations in air. If someone can complete the required rotation with a small distance, height or landing speed, that is still a fully rotated jump. It bugs me pre-rotations and under rotations are not judged consistently.
 

Miller

Final Flight
Joined
Dec 29, 2016
For the Ladies event, you get a similarly strong association between GOE and PCS (r=0.63) for the solo 2A. But if you plot GOE vs PCS for all jumps in the Ladies event with at least neutral GOE, you get a notable but much weaker correlation (r=0.37).

Also, unlike the men and their 3A, when ice scope is data is available, distance had a moderately good correlation with GOE for clean 2A among the Top 20 women (r=0.47).

If indeed there is a stronger correlation for 2As, that may lend some credence to the theory that in judges minds, all 3As and quads check the height and distance box merely for being 3As and quads. I would disagree with that (clearly, Yuzuru's 3A is outstanding, and I would also like to give a mention to Boyang's 4Lz, and I think both should be rewarded for that. And in my opinion, even Samarin, whose skating is often validly criticized, deserves extra credit for the size of his quads.), but it would be an interesting addition to our knowledge of how judges behave. I agree that in general more data would be helpful.

My guesses on this is that you often get men in the early groups doing fabulous 3As because of their physical abilities whereas for women there's a much more gradual progression - you don't often get women with one outstanding jump, but not others. Also you've got the business of some of the top women getting PCS scores that a lot of people would say don't bear much relation to their actual performance, so that could break the correlation between PCS and GOEs - not too sure about the 2As though and why that should correlate with PCS more, perhaps it's one of those things that needs more data and competitions.
 
Top