US Nationals - What are your thoughts?

[hockeyfan228]

I still think that barring a breakout performance, which I think Flatt is capable of, the judges will reward the more experienced senior skaters with the PCS to bring them over the newbies. I also think that Hughes will pull it out of her toenails over Flatt, who may succumb to a bit of nerves, given the expectations on her.

 

[R.D.]

OK, now we're getting somewhere! I ran these numbers and the results of 2006 U.S. Nationals through my calculator. Here is the formula for the "least squares regression line." That is, this formula predicts a skater's score at nationals from her scores in the Grand Prix or Junior Grand Prix events.

SCORE AT NATIONALS = 43.35 + .82 x AVERAGE GP SCORES.

The "coefficient of correlation" was r = .81, with r squared = .66. That is, 66% of the variation in scores among all skaters at nationals was correlated with GP scores. 34% is due to "other random factors."

Applying this formula to the skaters who did the grand prix this year and will also be at nationals, here is my prediction for 2007 nationals:

Meissner 186.07
Hughes 165.72
Czisny 164.43
Liang 160.51
Oster 154.65
Williams-Stewart 154.44
Kahle 150.13
Bulanhagui 147.38
Zukowski 146.48
Hyatt 144.63
Canarozzo 144.44

Applying the same formula to Rachael Flatt's score at Sectionals:

Flatt 173.95

So there you go. You can take that to the bank!

Wow, you really are the math man. But Kimmie's still far ahead of everyone else! :yes:

 

[Joesitz (RIP)]

Mathman = You're beginning to make sense with the calculation of several scores rather than just picking out the top score as the guideline for predicting. I'm much impressed with this system, and look forward to seeing other results where data is available.

What kind of data would you use for Worlds. Not easy.

Joe

 

[sk8inbladez2010]

How about for men and pairs? How do your calculations add up for Nationals?

 

[Mathman]

OK, I did the calculations over again, correcting one mistake on the data for Emily Hughes and also including some more data. For instance, now Bebe edges ahead of Alissa because of Bebe's good marks at Nebelhorn.

For a rule of thumb, the approximate formula works out to be:

To predict scores at U.S. Nationals (ladies), take 80% of the scores in previous events that season and add 40 points.

Meissner 179.25
Hughes 163.74
Liang 159.74
Czisny 158.11
Oster 152.53
Williams-Stewart 151.32
Kahle 148.28
Bulanhagui 145.18
Hyatt 142.39
Canarozzo 142.20
Taylor 135.99

 

[Mathman]

So, does this really work?

Here are the the scores predicted by the formula for last year's U.S. Nationals, followed by the actual score.

Cohen: Forumla predicts 185.23; Actual score 199.18
Meissner 167.66, 171.04
Hughes 144.68, 165.72
Taylor 152.16, 152.54
Liang 150.27, 151.23
Zukowski 143.30, 151.23
Czisny 169.50, 149.51
Corwin 125.60, 137.64
Williams-Stewart 129.61, 129.44

Average amount that the prediction is off: 7.02 points (how was I to know that she would throw in an extra triple Lutz with a 10% second half bonus and +0.42 GOE? )

 

[emma]

Mathman...those calculations are really cool; thanks for doing them and thanks for sharing them!

 

[SeaniBu]

Very nice MM, I am rather bummed not to see Firth's namehwell:

 

[Mathman]

Firth, predicted score = 160.42

 

[SeaniBu]

Firth, predicted score = 160.42

NICE!!!!!! Thanks MM!

 

[Mysticchic]

later tonight I will compile a list of Mens and Maybe dance and pairs. Then Mathman can dazzle use with more cool math perdictions.

 

[Joesitz (RIP)]

later tonight I will compile a list of Mens and Maybe dance and pairs. Then Mathman can dazzle use with more cool math perdictions.

I not only want a dazzle, I want a razzle dazzle.

Joe

 

[Mathman]

Later tonight I will compile a list of Mens and Maybe dance and pairs. Then Mathman can dazzle use with more cool math predictions.

Hey Mysticchic, you can dazzle your own self! :agree: Do you have a calculator with a statistics register (or a "linear regression" submenu)? Take the results for a bunch of skaters from last year. For each skater, enter the average of the GP results for x and the score at nationals for y. Press "r," "a," and "b."

"r" tells you the amount of correlation between the two sets of numbers. If r = 1.00, that means there is 100% correlation between GP scores and nationals scores. In other words, our predictions for this set of data will be exactly right every time.

If r = 0, this means there is no relation whatever between GP scores and scores at nationals, and we can't make any predictions at all.

In the examples that I did, it turned out to be about r = .80 (80% correlation). This is just a way of quantifying what common sense tells us anyway -- that overall, the best skaters tend to get the best marks at both the GP events and at nationals, but still, many surprises happen.

The numbers a and b are used to predict this year's nationals scores from this year's Grand Prix scores. To predict a skater's score at U.S. Nationals, go like this:

Nationals = b times Grand Prix average, plus a.

 

[Tonichelle]

time to break out the bedazzler! We'll have more fun than a costume designer for an ice dancer!

 

[Sylvia]

later tonight I will compile a list of Mens and Maybe dance and pairs.

BTW, I've already compiled IJS scores (international, regionals/sectionals) from this season for senior men, pairs & dance -- I should be able to post a link by later tonight and then Mathman can "crunch away" with his calculator. I'm not a Mathperson at all, so I'll defer to him.

Dumb question: How are the junior international scores factored in for the senior level skaters?

 

[Mysticchic]

thanks Syliva and Mathman SAY WHAT???....I am so stuip when it comes to math. General math not a problem....and since my son left for college I don't have anything but my trusty computer calculator.
And I have the same question about the Jr GP vs the Sr GP points.
You have the gift of Math not me.

 

[jsteam4501s]

For the Ladies, I believe Kimmie will win and I think Alissa will take the silver not too far behind Kimmie in total points. Bronze? Probably Emily. I just haven't seen any improvement in Katy - but maybe Christine Z could challenge for the bronze.

 

[R.D.]

maybe Christine Z could challenge for the bronze.

you're kidding right??

 

[Joesitz (RIP)]

BTW, I've already compiled IJS scores (international, regionals/sectionals) from this season for senior men, pairs & dance -- I should be able to post a link by later tonight and then Mathman can "crunch away" with his calculator. I'm not a Mathperson at all, so I'll defer to him.

Looking forward to that compilation, Sylvia. I am the most reluctant to accept scores as a determinator for competition results. I do think they give a guide to who the top 6 will be (give or take a mistake) but not to pick THE Winner.

Joe

 

[Sylvia]

Looking forward to that compilation, Sylvia. I am the most reluctant to accept scores as a determinator for competition results. I do think they give a guide to who the top 6 will be (give or take a mistake) but not to pick THE Winner.

My purpose in compiling these scores is NOT to predict what will happen at Nationals, but merely to list the scores from this season in international and U.S. qualifying (regional/sectional) competitions. Sorry for the delay -- the score lists will eventually be posted at http://unseenskaters.blogspot.com/ (right now, there's a compilation of Novice-Senior skaters' Nationals program music up at this link).