Nate Silver-ing Skate America: ICE DANCE
I love data, as many of you know, so I thought it would be fun to geek out on the stats from this competition.
1.) This data shows, clearly, how important hitting the pattern dance is in being successful in the Short Dance. Over 54 percent of Cappellini and Lanotte's lead over the Shibutanis came from level 4 +GOE Finnstep sequences. It's worth noting that in other categories, including the non-pattern dance elements and in GOE, the gap is much smaller between the two couples. I've I was Marina, I would make sure they master those sequences, pronto.
2.) The clear outlier in this data is the fact that Cappellini/Lanotte BEAT Davis/White in the step sequence. Granted it's half a point, but it's still worth noting. Actually what's more significant is the basevalue. 1.5 more points than D/W. This is one major reason they're bordering on 70.
3.) The Shibtanis will need to work on those Finnstep sequences. What saved them in this competition was a clear edge in other elements. H/D beat them in Finnstep by three points. The Shibutanis beat them in everything else by 2.51 points. The gap in PCS was about a point.
4.) In another example of overcoming pattern dance blues. Carron/Jones were dead last in the pattern dance, yet managed to squeak in levels to everything else
5.) Overall a much higher scoring competition than in 2012. Last year 2 teams scored 60+. This year it's 3 teams. And everyone else (except for one) scored higher than the 4th place team from last year.
6.) Interesting to note that the Reeds are actually third as far as base value, but their grade of execution is so low, they ended up sixth. Goes to show while levels help a ton in improving your scores, execution still matters too.
Anyone else want to geek out with me? I'd love to hear any of your other data observations, computations, etc.
This level of analysis is ... AWESOME.
It is a SHAME that there isn't software that automatically does this. And fans and probably skaters see none of this in actuality because it is too time consuming to do. But it reveals a LOT. Data is data. The numbers are the numbers.
The one conclusion that jumps out huge to me once again. The judges have a fraction of the influence that the callers do. The most powerful people in determining the outcome of an ice dance event especially is the technical panel. Base value differences in footwork and the patterns determine way more than a +1 vs +3 by any given judge on just about any element.
THANK YOU for compiling. I encourage you to please continue to do so. Valuable analysis, not just favorites/opinion spouting.
Wicked Yankee Girl
The biggest takeaway here for me is that more and more it becomes obvious that Igor was, as everyone always said, the technical wizard of Team Canton. With him gone, D&W, V&M, and the Shibsibs have a much taller hill to climb. It's hard to remember now that the Shibs used to be known for technical excellence.
Meanwhile C&L and the Reeds have really prospered, technically, since they moved to Igor from elsewhere.
The anomaly is that D&W got all level 4 Finnstep levels at this point in the season. They appear to have benefited from their Salt Lake appearance.
and... World Peace!
It was pretty obvious last season, too.
Originally Posted by dorispulaski
Yes. I was thinking the same thing about Igor. If I were an up-and-coming dance team, I think I'd gravitate towards his camp.
Originally Posted by dorispulaski
All his dancers seem to rise up through the ranks with technically excellent programs.
Good on D/W for the Level 4... I love the opening straight line sequence, and I hope to see that at L4 by their next GP.
Free Dance Analysis
Davis/White: 7.21 (STw4)
Cappellini/Lanotte: 7.07 (STw4)
Shibutanis: 6.93 (STw4)
Reeds: 6.57 (STw4)
Carron/Jones: 6.57 (STw4)
Hubbell/Donohue: 5.71 (STw3)
Zlobina/Sitnikov: 5.57 (Stw4)
Tobias/Stagniunas: 4.93 (STw3)
Davis/White: 22.50 (CuLi4/Sil4+RoLi3/RoLi4)
Cappellini/Lanotte: 20.93 (SiLi4+RoLi4/RoLi4/CuLi4)
Shibutanis: 19.71 (SiLi4/CuLi4+RoLi4/RoL4)
Hubbell/Donohue: 19.57 (RoLi4/CuLi4/SiLi4/CuLi4)
Tobais/Stagniunas: 18.71 (RoLi4/CuLi4/CuLi4+RoLi4)
Carron/Jones: 18.61 (SiLi4/RoLi4/CuLI4/StaLi4)
Zlobina/Sitnikov: 18.43 (SlLi4/RRoLi4/CuLi4)
Reeds: 17.94 (CuLi4/RoLi4/SiLi4/RoLi4)
Davis/White 18.14 (CiSt3/DiSt3)
Shibutanis 14.86 (DiSt3/CiSt3)
Cappellini/Lanotte 14.07 (CiSt3/DiSt2)
Hubbell/Donohue 13.50 (CiSt3/DiSt2)
Reeds: 11.78 (DiSt2/CiSt3)
Tobias/Stagniunas: 11.42 (DiSt2/CiSt2)
Zlobina/Sitnikov: 10.86 (CiSt2/DiSt2)
Carron/Jones: 10.43 (DiSt2/CiSt2)
Davis/White 6.07 (CoSp4)
Shibutanis 5.57 (Sp4)
Hubbell/Donohue 5.43 (CoSp4)
Reeds 5.29 (Sp4)
Carron/Jones 5.29 (CoSp4)
Tobias/Stagniunas 5.21 (Sp4)
Cappellini/Lanotte 4.93 (CoSp3)
Zlobina/Sitnikov: 3.07 (CoSp2)
1.) Well Davis/White just about swept the floor in everything. 14+ in GOE is pretty incredible. There wasn't a single +1 anywhere (EDIT: Okay, there was one.....! However, there is still point getting potential here since there were three level 3 elements. They could get level 4 on that lift. It's tougher to do on step sequences, but can be done.
2.) The Shibutanis' key to success is a throwback to their breakout 2010-2011 season. Get the levels. They had the highest levels out of everyone. Man of you have recall that they have struggled with levels the last two seasons.
2a.) In the mini-U.S. Nationals battle Shibutanis edged H/D in every technical element, including base value, GOE and scoring on each individual element. H/D error on Twizzles were particularly costly. Now, H/D did get higher PCS, which is not surprising given the relative lack of speed/crispness we saw from the Shibs compared to past FD performances. That said, in the end the Shibs won on TES and by a good clip. However the Shibutanis should not rest easy. The nearly tied +GOE between the two teams indicate that the judges believe Hubbell and Donohue, with a few exceptions, are at a similar level of execution at this competition. I would not be surprise if the winners of the two remaining spots, just as the gold medal between the two top teams, came down to who hit the levels.
3.) So the data shows why levels matter in the case of Cappellini/Lanotte. They got incredible +GOE in this dance. However their lower levels put their TES as the same as the Shibutanis. The gap in GOE and PCS kept them easily in second place.
4.) Levels also help the Reeds win the battle for 5th place. They again ranked 3rd, this time behind the Shibutanis and Davis/White. But again their execution clearly needs work given their low +GOE scores.
Well, D/W were excllent, but I think that they were a bit overscored: the spin, for example: there was a clear mistake there, at the beginning of the element, or the twizzles, which were synchronized but didn't have a lot of speed... Did these two elements REALLY deserve +3 in your opinion?
In the spin, they did receive one +1 (it was their only one in the whole bunch, so my original statement wasn't quite correct; will edit...). 2 judges gave +3. Statistically speaking those two judges would be the outlier. The other six judges gave it a +2.
Originally Posted by FSGMT
Now with twizzles, they received a few more +3 (four), however more judges opted to give it a +2.
Believe it or not, however D/W are capable of scoring higher in the spin: Here at SkAm, D/W got 6.07 for the spin. They received a 6.36 for it at Worlds 2013, with six out of nine judges giving it a +3
In twizzles: D/W got 7.21 here, which matched what they did at Worlds 2013. At 4CC last year, they got a 7.36, with 6 out of 9 judges giving it a +3.
Point is, as high as their overall score is, in those two particular elements, the judges don't consider it perfection and probably took account the factors you mentioned.
I think he's more of a baseball guy.
Originally Posted by dardar1126
Yes, I like Nate's step-back approach. He does not get too hung up on the horse race (speaking in particular of his political/election coverage) and focuses strictly on what the data actually states. He's not always right (he predicted the Seattle Seahawks would be at the Super Bowl last year....sad that was the ONE Prediction he didn't get right), but he very methodological in his strategy so he could be at least right most of the time.
I guess my goal is to step back and try to get down to brass tax regarding these competitions. The exercise has been more fun for me....I think i might try to do some other data analysis regarding other topics in FS, but it depends on time!
I liked Nate's methodology. Since he's left the NY Times, he is going to focus more on baseball over at ESPN (at least until the next big election in 2014, I guess).
Originally Posted by Mrs. P