Skating Statistics Project

[sk8indel]

Hey everyone, so for my IB Internal Assessment, I have to do a math project, and the guidelines lend themselves best to something statistically oriented, you know x^2 tests and all that jazz.

I want to do a statistical analysis with a female figure skater. Someone did one a little bit like what I'm going for with this one about Sasha Cohen a few years ago http://mkforum.net/forum/showthread....Just-the-Facts...

This is where you all come in. I'd obviously do that type of analysis including the success rate of every jump and sp vs. lp errors, etc, but I need an overarching question. Such as 'when xxxxxx makes one mistake, is it likely that she will make at least 2 more errors throughout the rest of the program'.

Basically, since all of my findings will ultimately be posted on these forums for your viewing pleasure, I wanted to know if there are any questions/skaters you all would specifically like statistical information about.

Thanks!

[literaryfreak]

Very cool! I have been meaning for a while to conduct a statistical study to find bias in judges' scores. Maybe I'll get around to it one day.

I'd love to see some statistical info about ice dancers Meryl Davis/Charlie White, but I'm not sure what exact questions/metrics, since there are fewer falls/errors in ice dance.

[kwanatic]

I want to do a statistical analysis with a female figure skater. Someone did one a little bit like what I'm going for with this one about Sasha Cohen a few years ago http://mkforum.net/forum/showthread....Just-the-Facts...

^^Hey, that was me! I did that! I have to say, I really wish I'd had a class in college where I could have used that. It was a lot of fun gathering all of the info and crunching the numbers...but it took a bit longer than I originally thought. But I really enjoyed it, plus it taught me a few things about Sasha as well! I said I was going to do Michelle's jumps stats too but I never got around to it.

I think a study of Yu-Na Kim or Mao Asada would be pretty cool. Maybe you could determine the success rate of Mao's 3a and how that correlates to her winning the competition? Especially since there was a lot of debate about whether the 3a is an advantage or not...or maybe something dealing with a lower base value and GOEs...or points in the bonus...or PCS vs base value + GOEs...

Under COP there's a LOT of math that can be used! I did Sasha's stats under the 6.0 system; with COP there are a lot more factors that go into the score so you have a wide range of stuff to choose from. The hard part will be narrowing down your question.

[sk8indel]

Maybe you could determine the success rate of Mao's 3a and how that correlates to her winning the competition? Especially since there was a lot of debate about whether the 3a is an advantage or not

First of all, major props to all your hard work. I remember when I first saw what you did with Sasha's stats, and I was like I WANT TO DO A SKATING PROJECT SOMEDAY, and low and behold that day has come, in highschool nonetheless. (I am uncharacteristically excited to do this math project)
I was debating this morning whether I wanted to do something with Mao's 3A or with the amount of pressure Mirai is under and her standing after the short impacting how she performs in the long. Because Mao's 3A success is a lot more of just raw, quantitative data, (and the fact that I sometimes get frustrated watching Mirai tank),I'm highly considering doing something with that. It also works out well that the 2004-2005 season was her first year of competing internationally, the first year COP was used, and the year that she started putting 3As into programs.

[kappa_1]

Add me to the chorus of posters who want to do some stats on figure skating!

Kwanatic's totally right- there are SO many things you could do with COP scoring. I was thinking something like correlation of GOEs with PCS factors, whether a skater's programs with falls over a season (or competition) get significantly lower PCS than when they don't fall, and also maybe the impact of a rule change on a skater, like the deletion of spiral sequences, change in rules of level gaining, etc.

I have some stats background, so feel free to PM me for help during your eventual project!

I got to chat with a stats professor when I was an undergrad who did some analyses on COP around the 2006 Olympics. Pretty cool stuff, and maybe you could get some ideas too:http://www.stat.yale.edu/~jay/OWG/

Yay for science and figure skating!

[os168]

What an interesting project, good luck to ya!

Not sure if this helps, but here're stats example compiled on ladies figure skating from NYTimes back in 2010

http://www.nytimes.com/interactive/2...ade.html#tab=4

Problem with these stats obviously does not show how they are effected by things like

- Whether it has a difficult transitions proceed and after it to affect the success rate.
- Whether it is in the 2nd half of the program which deemed more difficult
- How are the other elements impact that major element etc.
- Whether the figure skater were injured at the time, or burnt out later in the season etc.
- Leniency of the judges for under rotation, edge calls etc.
- How national competitions affect results.
- How are the pressures of competing on world championship level affect the success and failure

[geod2]

I'm sure these links have been posted before but in case you have trouble finding them here they are.....maybe something helpful in there.


http://journals.humankinetics.com/js...skatingjudging


http://blogs.wsj.com/dailyfix/2010/0...-still-biased/


http://www.dartmouth.edu/~ericz/transparency.pdf

It seems as though whatever question you explore would have to have enough data such that you could draw statistically valid conclusions.
As you know, small sample size leads to bad conclusions in every aspect of life (politics especially!!!).

One area I've always wondered about is how CoP scores for skaters who compete at the international level compare to those same skaters' scores in their home country. We often read comments about how the "international judges don't like (fill in the blank)" or words to that effect.
There would be a number of things you would have to normalize for (is that the right term?). For example, doing different programs (different content) would automatically cause different scores between the home country and the international judges. One way around that would be to look just at individual elements home country vs. international judges. It might be different if you look at PCS ....Is PCS automatically higher if program content is more difficult? IMO they should be independent of each other.
Anyway, the home country vs. international comparison of PCS scores might be an area ripe for analysis.
If you have to look at just one skater, then the sample size issue obviously becomes a bigger problem.

Cool project....good luck with it.

BTW the NYTimes link was great. Never saw that before. I wish they could expand it to include all jumps and some example spins too.


. .

[gkelly]

It might be different if you look at PCS ....Is PCS automatically higher if program content is more difficult? IMO they should be independent of each other.

The PCS aren't automatically higher if the program content is more difficult, but the program content probably does have an effect on the PCS -- in practice I don't think they are completely independent. That might be something to analyze statistically.

Of course, if by "program content" you also include content between the elements, i.e., the transitions and the multidirectional and one-foot skating, then those components should be higher with higher content, unless the quality suffers as a result.

Not a question for this thread specifically, but I am curious how we might think that program content should influence PCS, and how to quantify that.

For example, take any complex program from the past 20 years, cleanly skated, with lots of content between the elements and 6-8 successful triple jumps. What kind of PCS do you think it deserves?

What would you think about the exact same program skated with the exact same scores, but with only double jumps? Or with at least two quads?

Should the PCS remain completely independent of the jump content? Or is the performance less impressive with content that should be easy for a skater at that level? Or more impressive with jump content that challenges even the very best jumpers?

[kwanatic]

IMO, the PCS should be independent from the jump content and (let the constructors of the COP system tell it) that is what's done...however, I beg to differ.

On too many occasions I review the protocols from a competition and find myself scratching my head when it comes to figuring out where some of these scores come from. In some cases it's a matter of a skater not being rewarded for what they do correctly; in other cases it's a skater receiving high scores for doing basically nothing. It's rarely addressed by commentators but it's a huge flaw in the system. The PCS should reflect the program put on the ice...but they usually don't.

[demetriosj]

IMO, the PCS should be independent from the jump content and (let the constructors of the COP system tell it) that is what's done...however, I beg to differ.

On too many occasions I review the protocols from a competition and find myself scratching my head when it comes to figuring out where some of these scores come from. In some cases it's a matter of a skater not being rewarded for what they do correctly; in other cases it's a skater receiving high scores for doing basically nothing. It's rarely addressed by commentators but it's a huge flaw in the system. The PCS should reflect the program put on the ice...but they usually don't.

If you are correct, then what is the point of COP??!!

COP flaws sound just like the complaints about 6.0?

[geod2]

If you are correct, then what is the point of COP??!!

COP flaws sound just like the complaints about 6.0?

That point has been discussed on other threads, I think.

sk8indel still needs comments/suggestions for their project.

I'm sorry if I sent this thread OT.

.

[aliceanne]

Math makes my head spin. Too bad you have decided to select a female. Patrick Chan's scoring has caused the most controversy.

[l'etoile]

Maybe statistics for TES/PCS ratio of top 10 ladies skaters and male skaters (in regards of ISU world ranking) would be interesting and there should be something to tell there.