Monday, March 18

Following Up on the Current State of Skating

I am going to continue where I left off with my previous post and just add a few comments and observations based on the feedback I have received.

The problem seems to remain that everyone is fired up about the result of the mens competition at the World Championships, but no one seems to have any ideas on what to do to fix the apparent problem.

Dave Lease of The Skating Lesson videocast seems to have this belief in his head that I want to add more math to the already complicated point system. In a way, I guess. But I don't know what anyone else wants to do about it. I honestly think that reverting the system entirely (again) is just going to confuse people even more.

Maybe we should just count up the number of triples completed and then decide the winner from that. In the case of ties, it could come down to the preference of the viewer to decide which program they liked better.

Seriously. It sounds like some people think that is the answer. I'm not one of them.


My major proposals first included having a set point value for a fall-- regardless of the severity. Like the example I used in my previous article, this means a fully-rotated quadruple toe loop that results in a fall earns 2.58 points. No questions asked. No battle over whether the jump should have been a -1, -2, or -3. A fall is a fall. As I discussed previously, in the current system, a fall on a fully-rotated quadruple toe loop is worth 7.3 points at a -3 GOE. There is also a 1-point deduction automatically incurred for the fall, meaning the skater has essentially earned 6.3 points.

3.72 points difference between these two methods. People are acting like this is completely going to take someone out of contention for taking a risk. I beg to differ. 3.72 points can be made up by scoring  .50 higher than another skater on each of the components in the free skate (as each of the averaged components scores for the men are multiplied by 2), and it leaves over a point to spare.

You take a risk but you have a really strong program versus someone just throwing the big jumps out there? In a perfect world of PCS scoring, this isn't as make or break as everyone seems to think, but it is enough that I believe it shows the failure of the element.

I received a question about how this would apply to combination jumps. I am actually in favor of changing a few things when it pertains to the elements we see that have two or three jumps. Dave Lease's head might spin at this point for more math.

While reviewing Chan's free skate protocol, one of the things that stuck out to me the most was the fact that he was able to still score 9.70 points on his triple flip-half loop-triple Salchow sequence, which by newer ISU rules, is considered to be a triple-single-triple combination. Chan had a bad fall-out of the second jump, and received -2 across the board aside from one judge who deemed it to be -1 and one judge that should be fired for giving it a 0.

If I remember correctly, an earlier edition of IJS (probably before 2006) had it to where the grade of execution on combination jumps was applied to all three jumps.

I'll explain. Say a skater does a 4T+3T combination, and it has a +1 GOE as marked by the judges. Under current rules, only 1 point is added to this combination, because the GOE of the higher-base value element (here, the 4T) uses a -3.0, -2.0, -1.0, 0, 1.0, 2.0, 3.0 decrement/increment. 17.2 points base value, plus the 1.0, and the skater has earned 18.2 points.

What I would like to see is the +1 that has been deemed for the entire jump also applied to the 3T, which uses -2.1, -1.4, -0.7, 0, 0.7, 1.4, 2.1. The skater would earn an additional 0.7 points here, and the combination is now worth 18.9 points. Succeed, and you are really rewarded type of deal.

However, the same also applies to the opposite situation. Let's use the previously mentioned Chan combination: 3F+1Lo+3S. Since this combination was done after the half-way point of the program, Patrick started with a base value of 11.00 points (5.83 + 0.55 + 4.62). The nearly -2 GOE would be applied across the board for these jumps-- here, we will just say it was actually an even -2 final result. The flip loses 1.4 points, the loop loses 0.2 points, and the Salchow loses 1.4 points. He now receives 8.00 points for the combination rather than 9.70.

On the opposite end of the spectrum, Chan's quad toe-triple toe combination in the beginning would earn around 2.0 points more than it actually received, because of the near +3 GOE he received for the judges. He's actually earning slightly more points here after the GOE for each jump scenario is applied, but I prefer this system better.

Now.. what do you do with combination jumps where the second or third jump is a fall? Take Carolina Kostner's short program 3T+3T combination, where she indeed fell on the second jump. She was able to receive 6.50 points on the element. A 1.00-point deduction for the fall essentially drops it down to a 5.50.

I would do the following: the entire jump series garners a -2 GOE from the judges. The second jump automatically gets 1.03 points for the fall-- nothing is subtracted. That value is solid. The first jump still has the -2 applied, which would be -1.40 points. She now earns 2.7 + 1.03 points for the jump, or 3.73 points overall.

So far, nothing I have done has been that drastic of a difference compared to what we see right now. However, there is enough of a difference that I believe we would start to see more accurate results.

Still skeptical? Think about this. A +1 3Lz+3T combination is only going to get an additional 0.70 points-- we will say it was applied to the Lutz; the toe loop stays neutral in a way because it doesn't have a GOE added. Now if a skater does a 3T by itself with a +1 GOE, the 0.70 remains. Don't you think +1.40 for the combination would make more sense? It is, after all, one of the highest-risk elements for most skaters.


Aside from playing on words for my blog title, I have never really understand how flutzes and lips could still earn so many points when they are, in fact, truly Lutzes or flips. Skaters of the past like Elena Liashenko were doing four triple Lutzes in their programs if you use the severe outside take-off edge as the only indicator. How is that fair that the Lutz came so easy to her that she could just keep doing it, but someone who finds the loop the easiest is limited to doing it twice? A third attempt receives absolutely no points.

We live in a generation of skating where hardly any skaters have both a true Lutz and flip. All this talk about rewarding skaters that have a 'full set of triples' can almost be thrown out the window after the improper edge calls, save about five or maybe ten of them. Seriously.

What really drew my attention to the issue is how Brian Joubert, who has been competing on the senior level for over ten seasons now, showed up to the European Championships and did a 'triple Lutz' from the exact same set-up that he's done a triple flip for so many years. And what do you know? Because he decided to list the planned element as a 3Lz, it got called as one.

This is going to create quite a stir real quick in my opinion, and the only way to fix it would be to call the jump solely on the basis of the take-off edge. It's either an outside edge or an inside edge. You want less math and less nit-picks? Get rid of the 'e' calls that we see in protocols. Can only do a Lutz? Then you're limited to two Lutzes in the free skate. Try a triple flip and it's on the outside edge, you get no points, just like if you do any other triple more than two times.

This will not only make the skaters have quite a serious reason to fix their technique at a younger age, and it also doesn't put those at an advantage who may prefer the flip or Lutz over the other triples and can just do the same of one up to four times, while only losing minimal points for the two with the incorrect edge.


Let's face it, they are. This is where everyone will have the most differing opinions. Former US Champion Ryan Bradley was exclaiming on Twitter that he believed Brian Joubert should have won the World Championship the other night. Yes, really. I had him outside the *top TEN* in the free skate because of the lack of strong choreography, and as always with Joubert-- no transitions in sight.

But that just shows the potential for a major difference of opinion. What might be a great interpretation to music to one person might be a total snooze to someone else. What one person might see as a great interpretation because of audience involvement and excitement might be seen by another as a way to mask the true lack of anything really going, which I find often to be the case.

Speaking of major difference of opinion, I can tell you that Michal Brezina skated right through his short program without a hint of choreography or transitions. He could have been skating to any music. Six of the nine judges thought he was at the 8.00-level or above for interpretation. Seven of them had his transition mark at 7.50 or higher, and the other two were just behind. In a fair world, Brezina's transition mark belonged in the 3.00 or 4.00 range. Re-watch the program. It's really that bad.

Instead of worrying about how Chan was still able to score in the 8.00's when he actually did listen to the music, people should be more concerned about scores like that of Brezina which are clearly way off the mark-- fan or not.


One needs to look no further than Chan's grades of execution on his double Axel at the end of his free skate to see how the judges seemed to try to work some more points back into his program. The element was a 0 for me, but he got some +2's and even a +3. I guarantee you that if a skater in the first or second group did a double Axel with ten times the quality of Chan's particular jump the other night, they would be getting a 0 or maybe +1.. at best.

Again, instead of going all crazy because you believe Patrick Chan's skating skills dropped from a 9.00 to a 6.00 or something in the course of four and a half minutes (which I promise you, they did not), you should be looking at these instances as another huge problem.

Until next time..


Ken Scott said...

I like your suggestions Tony. Personally I think a fall on a quad (or a triple for that matter) should be worth about the same as a successful triple of the same type. So a fall on a quad toe would be worth around 4.1 I think if you make the penalty too steep, athletes might be more inclined to not even try the jump and just go for something safer. So the deduction would not be quite as low as your suggestion, but still better than the 6.3 that it is worth now.

The components are always going to be the major problem but it's the nature of the sport where 50% of the mark is so subjective. I think having the panels split up as you suggested is a good idea and will make some improvement to the accuracy of components scores but there is always going to be a way for the judges to "cheat" the system so to speak because of the subjectivity of the sport.

If the ISU is so worried about how they look to the casual observer of the sport. Maybe they should change the breakdown to a 70-30 or 60-40 type of thing to give the components scores a bit less of a factor on the overall outcome.

aims said...

I agree with Ken Scott's idea of lowering the factors for the PCS. Mathematical adjustments to the TES side of scoring, while logical, might be too complicated to implement at this stage and will require too much adjustments to the system, while changing the factors to lower the worth of the PCS wouldn't be too drastic and can be easily implemented. It would prevent overscoring of performances that are marred with technical errors but are saved by the skaters reputation judgings in the secong mark.
I'd go further to have differnt factors for different segments of the PCSs as well, like in ice dance. IMO, PE and IN which would differ according to the actual performances on that day should be given higher factors than CH,SS and TR which wouldn't change so drastically given the already more or less established skills of the skater.

Anonymous said...

"The components are always going to be the major problem but it's the nature of the sport where 50% of the mark is so subjective".

Ken, I have a different opinion reg PCS representing 50% of skaters marks.
TES and PCS are like apples and oranges (I believe it's the way they are intended to be, too). They are measured in 2 completely different ways, using 2 very different scales.

It makes little sense to compare apples to oranges, but we can compare the %s (or ratios) of both fruit sorts as ingredients of the final dish aka fruit salad aka TSS.

There's 7 tech elements in mens SP, and 7 judges (2 scores, one highest and one lowest, are always excluded from the calculations of the element mean score). 7 x 7 = 49 tech marks per skater.

There's 5 program components, which gives us: 5 x 7 = 35 component marks per skater.

49/35=1,4 which is the ratio of tech/ component for mens SP (equals 58% of the total number of marks per skater).

Let's do the same for mens free skate marks.

13 tech elements x 7 judges = 91 tech marks

5 components x 7 judges x 2 (because they use 2.00 factor in mens free)=70

91/70=1.3, or 56.5 % of the total number of marks in fs

Next, we can look at the individual recipes / salads / TSScores (I'll use 2013 WC men sp marks).

1. Chan: TES/ PCS=1.15
2. Ten: 1.25
7. Fernandez: 0.99
8. Aaron 1.26
16. Majorov 1.16
17.Verner 0.88

There's no identifiable pattern in the individual ratios, even though the fruit salad portions (TSS) are getting smaller and smaller with each placement.

The same applies for the fs scores.

Based on the above, it's hard to tell if the TSScores are made up of 50% TES and 50% PCS (especially if we take into account the different scales for TES and PCS and unknown GOE and PCS 'weights').

Anonymous said...

To anon 10:34 AM March 20
The PCS marks are factored so that they will ideally represent approximately 50% of the TSS scores. They are factored differently for mens, ladies and pairs, and also differently between short and long programs as well to balance them out with the different numbers of technical elements required. Take a look at the 'factors' that you'll find on the protocols.
Obviously the actual balance in each score will differ with each skaters in accordance with how they skate their actual programs.

Tony said...

The last Anonymous has it exactly right. There shouldn't be an identifiable ratio between skaters down the line, because you have skaters in the lower ranks who won't earn the points technically but have absolutely gorgeous, well-choreographed programs. You also have skaters that score massive points technically but have the bare bones of a program otherwise.

In the end, 100 points is the maximum score for the men as far as the PCS goes in the free skate. That means the skater received straight 10.00's for all five components after judges scores are dropped and all of that.

We saw, as an example, both Takahiko Kozuka and Patrick Chan come extremely close to the 100-point mark technically at the 2011 World Championships. They both skated great programs (Chan with two quads, Kozuka with a quad and two triple Axels).

I think the way the PCS is factored right now is as good as it gets.

Anonymous said...

10:45 AM, March 21,

--"They are factored differently for mens, ladies and pairs, and also differently between short and long programs as well to balance them out with the different numbers of technical elements required. Take a look at the 'factors' that you'll find on the protocols."

Yes, I'm aware of different factors used in different disciplines, that is why I mentioned in my previous post that my findings were for mens wc2013 sp and fs.

--"Obviously the actual balance in each score will differ with each skaters in accordance with how they skate their actual programs."

Yeah, obviously.

Tony, I take it you see numbers 100 and 100, and for you that means there's a balance. Because 100 is a 100 is a 100. I tried to explain in another thread why I think differently.