Quantifying titles

[millie]

May her big jib draw!

Hey by’e, You are as close to being a Newfie as you can get...long may her big jib draw....

 

[chopinskate]

That's an interesting point. Is it better to win a championship and then successfully defend it the next year? Or is it better to come back and regain the championship after some time has passed? I think maybe the latter is more difficult and more rare. (?)

And comebacks, too.

It is why I think we need that ELO rating.

 

[Old Cat Lady]

If I wanted to evaluate a career statistically, I wouldn't value titles at all. If the best in the world are all competing against each other at the Ondrej Nepela Trophy, there is no reason to abstractly value this competition more than then Olympics unless skaters are timing themselves for a peak later in the season.

So if I really wanted to evaluate skaters "statistically" I would probably design a multiplayer variation on the Elo rating system by approximating each event as a pairwise competition between each competitor. There are also variations which include "momentum" into the rating system that would probably be relevant for skating. Like I said, there is no statistical reason we should value an Olympic win more than a Nepela Memorial Trophy win if all the same competitors are there - except if a skater is intending to peak at the Olympics, in which case Elo deals with that by updating the scores based on current relative scores (meaning if you are under-scoring because you haven’t peaked yet, you’ll get a much bigger point boost from winning than someone who has dominated all season).

They talk about the ELO on the tennis forums a lot. Produces a few interesting results. Don't think it would work in figure skating - I think a lot of skaters do try to peak for worlds and Olympics. And Olympics does have the most competitive field since it is prioritized highest by the athletes. Same with worlds having the second most competitive field. Also, I don't think it would work as well since there's a much smaller set of data - skaters generally have shorter careers and only do around 6 international competitions a year as opposed to 20-30.

That being said, I don't claim to fully understand the system so who knows

 

[Izabela]

They talk about the ELO on the tennis forums a lot. Produces a few interesting results. Don't think it would work in figure skating - I think a lot of skaters do try to peak for worlds and Olympics. And Olympics does have the most competitive field since it is prioritized highest by the athletes. Same with worlds having the second most competitive field. Also, I don't think it would work as well since there's a much smaller set of data - skaters generally have shorter careers and only do around 6 international competitions a year as opposed to 20-30.

That being said, I don't claim to fully understand the system so who knows

The ELO rating system though could be modified to include the importance of Olympics and Worlds in relation to other events. IIRC the unofficial ELO rating for tennis also includes a K factor to reflect the importance of the 4 Opens compared to others, as well as who they get to play against to win the title. It both captures the importance of events and the depth of the field. It doesn’t mean it’s entirely objective and people won’t question it (i.e. Federer fans crying because Djokovic is #1). I’m pretty sure the ELO system can be modified to calculate comebacks and at least provide higher points not only for those who win back to back but also to those who won it (particularly Worlds) after years of not competing (agreeing on Mathman and chopinskate here that the latter is more impressive feat.)

 

[Baron Vladimir]

I think ISU is done with any scoring system based on the comparation between the skaters.