Calculation of skating skills | Golden Skate

Calculation of skating skills

Elucidus

Match Penalty
Joined
Nov 19, 2017
First attempt to calculate mathematically such vague notion as skating skills - and compare different skaters based on that
https://www.sports.ru/tribuna/blogs/ternovblogfk/2518250.html
It's from the same guy who did statistical analysis of jump traits before:
https://www.goldenskate.com/forum/showthread.php?77001
https://www.goldenskate.com/forum/showthread.php?77740
IMO it's very interesting read as well - with some unexpected for many people results :biggrin: Again, I advise to use google translate to read it for nonrussian speakers. That article is dedicated to some Russian top juniors only. Author hinted in comments though that the next article can be about Russian senior ladies and maybe some Japanese ladies as well. But he wants to improve his current method of calculation first - for now it's a work n progress i.e. demo-version. Still I think it's very commendable achievement even in current state.
 

andromache

Record Breaker
Joined
Mar 23, 2014
So reading this in google translate and being extremely bad at math/statistics/etc. - I didn't get much actual information out of this. But I can still tell it is super impressive and interesting! Would love to see some more knowledgeable to read it and see what they think.

I would love to see this analysis done for the Russian and Japanese senior ladies to see how they rank using this methodology.
 

Orlov

Medalist
Joined
Jun 19, 2018
Interesting and useful. But the main problem is the weights of steps. That realized and the author himself.
 

Tavi...

Record Breaker
Joined
Feb 10, 2014
First attempt to calculate mathematically such vague notion as skating skills - and compare different skaters based on that
https://www.sports.ru/tribuna/blogs/ternovblogfk/2518250.html
It's from the same guy who did statistical analysis of jump traits before:
https://www.goldenskate.com/forum/showthread.php?77001
https://www.goldenskate.com/forum/showthread.php?77740
IMO it's very interesting read as well - with some unexpected for many people results :biggrin: Again, I advise to use google translate to read it for nonrussian speakers. That article is dedicated to some Russian top juniors only. Author hinted in comments though that the next article can be about Russian senior ladies and maybe some Japanese ladies as well. But he wants to improve his current method of calculation first - for now it's a work n progress i.e. demo-version. Still I think it's very commendable achievement even in current state.

I don’t think the definition of skating skills is vague, it’s just not something that’s easily measured, like the length of someone’s stride or the height of their jump. The author’s theory is that by measuring someone's speed / number of pushes, his analysis indirectly takes into account balance, flow, and one foot skating. However, he doesn’t consider depth of edges or turns or multi directional skating at all, and he confines his analysis of steps to “simple” steps, which for me is a major flaw even in analyzing speed and acceleration, because if a step is “harder,” it’s probably also harder to maintain constant speed and to gain speed out of it. Additionally, skaters are supposed to demonstrate “command of the [entire] skating vocabulary,” and varied use of power, speed, and acceleration, not just mastery of certain simple steps and the ability to skate fast and accelerate quickly. Finally, skating skills have an aesthetic component that simply isn’t considered here. While I appreciate the effort he’s making, I don’t know that it tells us very much about someone’s skating skills.

https://www.isu.org/figure-skating/...96-program-component-chart-id-sp-2019-20/file
 

gkelly

Record Breaker
Joined
Jul 26, 2003
It looks as though there are quite a few aspects of skating skills not taken into account:

How many difficult turns
How many clockwise vs. counterclockwise turns -- and also CW vs. CCW and forward vs. backward crossovers and other stroking
Those could be counted

Distance covered is a proxy for speed and power, assuming that everyone is skating for the exact same amount of time. But if you take into account stops/posing, prolonged spins that take time away from moving across the ice, and potential differences of almost 20 seconds in legal program length, as well as programs that use medium-fast skating throughout vs. those that vary skating speed from slow to medium to very fast, there could be skaters who cover the same amount of ice but have very different abilities in terms of maximum power and control to vary the speed, or vice versa.

Depth of edge is also a factor not taken into account, which if I understand the machine translation correctly the author is aware of. In theory this could be measured someday with more advanced technology and/or calculation methods.

And then there is quality -- the security of the edges, cleanness of the turns, softness of knee action, what kinds of sounds the blades are or are not producing, etc. Those qualities don't come across well, or in some cases at all, on videotape but need to be perceived live. I don't know how those could be quantified more accurately than a knowledgeable expert rating them on a scale of 1 to 10 or some similar method.

At this point, we can say that this author has found ways to quantify a subset of all skills measured as "Skating Skills" and can offer comparisons of that subset only. Which is interesting information but doesn't mean that a skater who scores lower on this particular subset might not be demonstrate a higher overall skill level.
 

Elucidus

Match Penalty
Joined
Nov 19, 2017
At this point, we can say that this author has found ways to quantify a subset of all skills measured as "Skating Skills" and can offer comparisons of that subset only. Which is interesting information but doesn't mean that a skater who scores lower on this particular subset might not be demonstrate a higher overall skill level.

You're right formally. Still, considering that usually each SS criteria depends on others criterias and influence them too - I wouldn't be so sure about rightness of your statement. In other words - a skater who is bad in let's say two of SS criterias - usually is bad in all others too. And vice versa.
Also, whether this analysis is objective or not - it still shows Kostornaya's and Tarakanova's SS as top 2 - which is a widely known opinion formed by those who determined SS by convential means (judges, experts etc). Therefore matching results shows that this method must have some semblance of objectivity at least.
 

Baron Vladimir

Record Breaker
Joined
Dec 18, 2014
You're right formally. Still, considering that usually each SS criteria depends on others criterias and influence them too - I wouldn't be so sure about rightness of your statement. In other words - a skater who is bad in let's say two of SS criterias - usually is bad in all others too. And vice versa.
Also, whether this analysis is objective or not - it still shows Kostornaya's and Tarakanova's SS as top 2 - which is a widely known opinion formed by those who determined SS by convential means (judges, experts etc). Therefore matching results shows that this method must have some semblance of objectivity at least.

Hm, while i can see all the love for Tarakanova speed and flow, i dont recall she is scored better in her SS by the judges comparing to some of the others on the list. That is because all categories judges are looking for are not included there, which that guy admitted after all. However, i think its possible to measure in some way how much time skater spent on one foot, how much skated in both direction/or how many times changed direction and which difficult steps and turns were included in the program. According to just those criterias he used im pretty sure Kaori Sakamoto would be the highest scored lady, even from the judges point of view that wasnt always the case (or rarely is). Also, like some pointed out, sound of the blades and amount of 'snow' they produced could be another indicator of SS. Also, 'the way they are pushing' while gaining the speed based on the aestetic of body position (Tarakanova may be lacking in that for me as well, comparing to the others)... But after all, i appriciate his idea and what he was trying to do, thats for sure.
 

Alex65

Final Flight
Joined
Aug 11, 2018
Country
Russia
I like that at least someone is trying to find the grain of common sense based on specific numbers in this sanctuary of preferences and corruption called PCS.
 

readernick

Medalist
Joined
Dec 5, 2015
Obviously, this is interesting information and may show some correlation to reality ( I think most people would agree that Aliona Kostornaia has the best SS among Russian ladies ). But, not enough factors are analyzed, as stated above, and even if you want to just discuss speed and flow you need to consider the skaters' posture. Some skaters like Carolina Kostnor, Alena Kostornaia, and most Japanese skaters gain speed while maintaining good posture, others bend over while performing crossovers which makes it easier to gain speed and compensate for their lack of knee bend and power.

I find this analysis similar to the Russian blogger who ranked skaters based solely on jump height, it provides an interesting comparision but does not account for all factors in the quality of jumps. ( for instance jump length/distance, ease of the jump, transistions, landing speed in the latter example and posture, edge depth, variation of skating speed, multi-directional skating and so on in SS).

It is interesting data but certainly not an all encompassing review of SS.
 

colormyworld240

Medalist
Joined
Dec 9, 2017
I think this data is very interesting and it looks like they spent a lot of time on it. In the article, the author addresses the limitations in his approach. He also states what it is; looking at coefficients for acceleration as well as distance in ice coverage and speed across the ice. All with uncertainty values for margins of error which is more important than the values themselves. The ranking only uses the values, I believe, but are they all statistically different? Either way, he clearly states what his method looks at; not determining who has better skating skills, but who ranks higher in the components of skating skills that his method looks at. But I do believe this is the first study of its kind, so it's very novel and the results are interesting. With any study and method development, there are always limitations (and he is only measuring things that are measurable; things like multidirectional skating, deep edges, quality of turns, etc. are more complex and I'm not sure if anyone has made an attempt).

In terms of skating skills, I'd argue that Tarakanova is better than Kostornaia. She has comparable speed, but I think she has deeper edges that she really digs into the ice, leaving behind deep ice marks. Her turns also cover a lot of distance and she doesn't slow down to do them. When I watch them, I always thought she was faster than Kostornaia. Which is why I wonder if the two speeds are statistically different, even though Kostornaia has the higher value. If someone has time for the math :slink:
 

Tavi...

Record Breaker
Joined
Feb 10, 2014
I think this data is very interesting and it looks like they spent a lot of time on it. In the article, the author addresses the limitations in his approach. He also states what it is; looking at coefficients for acceleration as well as distance in ice coverage and speed across the ice. All with uncertainty values for margins of error which is more important than the values themselves. The ranking only uses the values, I believe, but are they all statistically different? Either way, he clearly states what his method looks at; not determining who has better skating skills, but who ranks higher in the components of skating skills that his method looks at. But I do believe this is the first study of its kind, so it's very novel and the results are interesting. With any study and method development, there are always limitations (and he is only measuring things that are measurable; things like multidirectional skating, deep edges, quality of turns, etc. are more complex and I'm not sure if anyone has made an attempt).

In terms of skating skills, I'd argue that Tarakanova is better than Kostornaia. She has comparable speed, but I think she has deeper edges that she really digs into the ice, leaving behind deep ice marks. Her turns also cover a lot of distance and she doesn't slow down to do them. When I watch them, I always thought she was faster than Kostornaia. Which is why I wonder if the two speeds are statistically different, even though Kostornaia has the higher value. If someone has time for the math :slink:

He did a similar analysis of jumps. I agree it’s interesting, but aside from what @gkelly mentioned above, he seems to have only looked at one SP and one FS for each of them, so I’m not sure how reliable it is, even for these skaters. And regardless of how clear he is about what is or is not being considered, I’ve seen an awful lot of people who seem to think it “proves” that of the skaters he analyzed, Kostornaia and Tarakanova have the best skating skills.
 

SkateSkates

Medalist
Joined
Feb 17, 2010
To me, it proves who has the best ice coverage, which in my opinion is related to one of the components of skating skills, but not all (as mentioned above by many posters).

I thought the maps of the skaters programs was the most interesting part of the article. It really does depend on the choreo of the program though - if there are a lot of stops and starts for the judges, it limits the amount of distance travelled which skews the overall results. It of course doesn’t take into account proper stroking technique, posture, multi directional skating, etc., but I think Kostornaia and Tarakanova being far ahead of the rest of the pack is correct. You can see it easily from the Russian domestic competition videos where they film from 1 angle.
 

colormyworld240

Medalist
Joined
Dec 9, 2017
He did a similar analysis of jumps. I agree it’s interesting, but aside from what @gkelly mentioned above, he seems to have only looked at one SP and one FS for each of them, so I’m not sure how reliable it is, even for these skaters. And regardless of how clear he is about what is or is not being considered, I’ve seen an awful lot of people who seem to think it “proves” that of the skaters he analyzed, Kostornaia and Tarakanova have the best skating skills.

Well, that's on them. But the author himself does not express this, and I think this study and the jump height study are good ones. As with any method development, not just in FS, it's not all encompassing (impossible to be), but specifically states limitations and the area of study.

But overall, of the Russian ladies discussed in his article, it's my personal belief that Kostornaia and Tarakanova do have the best skating skills, not from the study, but visually (and unquantifiable).
 
Top