Jump to content

Recommended Posts

58 minutes ago, shanshani said:


It's not just in theory. If you backtest the new rules using the actual GOEs the judges handed out (which I did here), you'll see that Yuzu loses around 4 points across the short and the free (entirely because of the loss of a jumping pass) even on messy competitions, whereas Nathan consistently loses double digits, even in his cleaner competitions. Yuzu would have won 2017 4CC under the new rules, not Nathan. If you run a hypothetical competition (for instance, here) with Yuzu having +4 GOE across the board and 97.5 PCS, while Nathan has +3 GOE across the board and 95 PCS, Yuzu wins in all cases except the one where he keeps something like his 2016-2017 layout while Nathan adds the 4Lo. Maybe +4 GOE across the board is optimistic for Yuzu, but so is +3 GOE and 95 PCS for Nathan, especially since he's landed the 4Lo exactly once and hasn't tried it again and still has trouble with his 4S and 3A.

Basically, in order to get a Nathan victory in a case where both of them perform similarly (ie. neither of them falls more than the other), you have to make some pretty generous assumptions about Nathan's progress and pretty ungenerous assumptions about Yuzu's. Note that so far, (on the 3 point scale) Nathan's personal best average FS GOE is 1.29 (Olys), whereas Yuzu's personal best average FS GOE (GPF 2015) is 2.64.
 


No, I was comparing the gap in BV between skaters that existed before and after applying the quad repetition rule. The gap always goes down once you apply the quad repetition rule. In both cases, the skaters will replace repeating a quad with repeating 3A or 3T (3T allows them to get rid of 2T). The replacement options are the same for both skaters, because they couldn't repeat any triples when they were repeating quads.

I highly encourage people to run numbers before drawing any conclusions on what will happen next season. I've made a template for this which you can find here--just input hypothetical layouts and GOEs and go to town, the spreadsheet does all the calculations for you (but you'll have to download it first). I feel like there's a lot of pessimism here that isn't really warranted if you do the math.
 

I don't think that testing the new rules based on past competition results would give an accurate result because we can't guarantee that the judges will evaluate the GOE the same way they did then (what happened in the past season will effect how judges perceive skaters a.k.a. reputation judging).

However, I did run the numbers on different layouts for all top men and ladies but unlike many people who did this I applied GOE based on historic reference. Basically, I looked at what the judges gave each element (for a specific skater) in the past and based on that tried to predict what they may give using the new system. For example, Yuzu's 3F was never given perfect GOE even though it hits the bullet points required, so expecting it to get +5 in the new system would be unrealistic. On the other hand Nathan and Shoma's quads received +3 GOE from some of the judges in the panel on many occasions and I can't for the life of me figure out what bullet points those jumps hit to deserve it. Thus, it is not beyond reason to expect them to get 3 or more in GOE for the same jumps in the new system. I don't really understand why many expect jumps that were getting above average GOE to suddenly start receiving less with the new rules, and I am expecting quads in particular to be clearing the required bullet points by default (unless there is an obvious mistake) just my opinion so ignore it if you don't agree

 

In addition, the younger men have an advantage over Yuzu in that they are on their way up while he is kind of stuck at the top. We can easily predict their PC and GOE to rise in the next couple of years while Yuzu has already maxed out on PC and his GOE is already very high. His score threshold for improvement is lower than theirs and the fact that he is, as of last year on top, doesn't mean it would be the same this year or the next. Those guys will get better and their score will reflect it because they have space to show their progress while Yuzu can have an artistic revolution on ice and we won't see much difference in his scores.

 

Disclaimer: I feel that I need to clarify that I do not think that Yuzu is a victim or that he is targeted by the new rules. I am just trying to look at the whole scene realistically and thinking about the best way to approach the new system (just for fun). I am also definitely not a hater and have no intention of being mean to or picking on Nathan and Shoma, I love them but they make for a perfect example that everyone would recognise. 

Link to comment
Share on other sites

Honestly I want the judging UI to display all the GOE bullets and have the judges actually press in the d*mn bullets instead of just pressing +1, +2, etc. None of the rules on face hurt Yuzu, or even of the older skaters should they stay. The issue is that no one is really giving time to assess quality, and there is no way to check if we really are assessing quality correctly and consistently by judges to a degree. =/

 

In so far as ceiling is concerned, well that's why Yuzu is going after 4A right? BV is kind of the only ceiling he hasn't come glaringly close to hitting yet. Nathan is actually IMO, almost at his ceiling already (this season, he might hit it)-his BV ceiling is probably hit, the GOE is almost done, and his PCS is just a few points away. Shoma is almost there. Boyang has more room, if the judges actually reward him is another issue. So the only one remaining is Zhou, who hasn't come close to hitting the GOE or PCS ceiling yet. =/ Only thing holding back PCs points from going off like rockets, IMO, is that the olys are over and the new quad is starting, so maybe points won't rise yet (but will rocket up 2020/21 season).

Link to comment
Share on other sites

There's also the 'serious' error PCS deduction. They capped maximum PCS for each component under the new rules, didn't they? But that will only affect the guys who already get an average of 9.5 for PCS. That would be Yuzu. So a not perfectly clean skate from Yuzu will be punished because even with a step out or a hand down, he should be getting well above 9 for a few components. That particular insertion benefits those who stay around 9-9.25 per component because they basically get no additional punishment for a serious error. 

 

@Neenah Have you posted your numbers anywhere? I'd love to see them even if you weren't thorough and haven't checked everything, to get a general idea.

 

Seriously though I don't think judges pay that much attention to bullets other than the obvious ones. The most obvious used to be the varied air position. That's probably why all the tano jumps got positive GOE. For all the other bullets, I think they just took the 6.0 way and applied it to GOE. And PCS too. With the new bullets, they can really do that because it's been made pretty blatantly subjective with the 'good' and 'very good' and all that. 

Link to comment
Share on other sites

On the other hand, though, if Yuzu gets his 4Lo and 4Lz back, and he lands the 4A as well, then everybody can go 4F themselves because he will have a BV advantage and with that, there's nothing anybody can do but pray he doesn't go clean. That is even if GOE and PCS all skyrocket and everybody starts getting boosted to high heaven so Yuzu doesn't get the quality advantage he obviously should.

Link to comment
Share on other sites

Bullet points actually became more streamlined and could potentially be easier to achieve, outside of whatever criteria they'll employ to measure 'very good height and distance', which might be used to selectively keep skaters under +3.

Link to comment
Share on other sites

1 minute ago, xeyra said:

Bullet points actually became more streamlined and could potentially be easier to achieve, outside of whatever criteria they'll employ to measure 'very good height and distance', which might be used to selectively keep skaters under +3.

I think they were made easier to evaluate and justify tbh. Just like the solo jump and steps. 

Link to comment
Share on other sites

4 minutes ago, xeyra said:

Bullet points actually became more streamlined and could potentially be easier to achieve, outside of whatever criteria they'll employ to measure 'very good height and distance', which might be used to selectively keep skaters under +3.

2 minutes ago, WinForPooh said:

I think they were made easier to evaluate and justify tbh. Just like the solo jump and steps. 

 

In that case, maybe ISU judges will actually shock the hell out of us all and actually judge properly??? :knc_yuzu1:

Link to comment
Share on other sites

Just now, OonsieHui said:

 

In that case, maybe ISU judges will actually shock the hell out of us all and actually judge properly??? :knc_yuzu1:

 

Nah. Too many personal and fed interests in judging to allow for that.

Link to comment
Share on other sites

2 minutes ago, OonsieHui said:

 

In that case, maybe ISU judges will actually shock the hell out of us all and actually judge properly??? :knc_yuzu1:

I didn't mean it nicely, I meant judges aren't following the more specific and better defined earlier bullets, so they just made it much more generic so they can mark how they feel it. 

Link to comment
Share on other sites

35 minutes ago, WinForPooh said:

There's also the 'serious' error PCS deduction. They capped maximum PCS for each component under the new rules, didn't they? But that will only affect the guys who already get an average of 9.5 for PCS. That would be Yuzu. So a not perfectly clean skate from Yuzu will be punished because even with a step out or a hand down, he should be getting well above 9 for a few components. That particular insertion benefits those who stay around 9-9.25 per component because they basically get no additional punishment for a serious error. 

 

@Neenah Have you posted your numbers anywhere? I'd love to see them even if you weren't thorough and haven't checked everything, to get a general idea.

 

Yes, that PC deduction is the stupidest rule they came up with. I hate rules that only apply to a subset of the skaters  :tumblr_inline_mzx90enaUF1r8msi5:

 

I haven't posted anything because I was just doing it for fun. I have a script to calculate the score (modified the one from the online calculator) and I tested different layouts whenever I had free time. Skatingscores provided the historic context of how the judges normally evaluated specific elements for each skater.

Link to comment
Share on other sites

2 minutes ago, Neenah said:

Yes, that PC deduction is the stupidest rule they came up with. I hate rules that only apply to a subset of the skaters  :tumblr_inline_mzx90enaUF1r8msi5:

 

I haven't posted anything because I was just doing it for fun. I have a script to calculate the score (modified the one from the online calculator) and I tested different layouts whenever I had free time. Skatingscores provided the historic context of how the judges normally evaluated specific elements for each skater.

Ah, okay. Next time you do it, I'd love it if you could send me a PM. I won't quote you anywhere and won't use it for anything, I'd just really like to know. No problem if you don't want to, though! 

Link to comment
Share on other sites

19 hours ago, shanshani said:

Oh, also, Olys scores if Judges Chen and Parker's scores are dropped and the rest averaged:

Free Skate
Nathan: 215.08 -> 214.15
Yuzu: 206.17 -> 206.65
Shoma: 202.73 -> 203.61
Javi: 197.66 -> 198.6
Boyang: 194.45 -> 193.79
Vincent: 192.16 -> 191.40

SP (Parker did not judge the SP)
Yuzu:
111.68 -> 111.52
Javi: 107.58 -> 107.96
Shoma: 104.17 -> 104.49
Boyang: 103.32 -> 102.09
Nathan: 82.27 -> 82.41
Vincent: 84.53 -> 84.32
 

Totals
Yuzu: 317.85 -> 318.17
Shoma: 306.9 -> 308.10
Javi: 305.24 -> 306.56

Nathan: 297.35 -> 296.56 (moves from 5th to 4th)
Boyang: 297.77 -> 295.88 (moves from 4th to 5th)
Vincent: 276.69 -> 275.72
 

As you can see, no podium changes, though Nathan and Boyang do swap places right beneath it. Interestingly, Chen doesn't seem to under-score Yuzu and Nathan the way she under-scores Shoma and Javi. Imo, the top 6 order should have been: Yuzu, Javi, Shoma, Boyang, Nathan, Vincent, but divergences between the actual order, the order minus Parker and Chen, and the ideal order have to do with other judging issues.

 

Nice to share! Did you calculate that yourself?

I am super glad that the podium didn't change after dropping those judges' scores. Imagine if it did, all sorts of conspiracy theories would spring up.

I'm just not the "what-ifs" type (until VAR, maybe).

 

Link to comment
Share on other sites

yeah the PCS ceiling for skates with mistakes are a bit ??!!??!!

They could be reasonable recommendations, and I can agree that it's weird that a program with falls would get a 10 for outstanding in, say, PE (even if skaters recover from a fall and make people forget it, at a certain point there has been a break in the spell, and imo that's something a judge should take into account).

But I don't like that they put an artificial limit for some components only for the highest marks. At this point, it would be more fair if ISU had said: for serious mistakes take away half a point from SS, one point from PE.

 

If a skater A with the best SS in the world and skater B with 'very good' SS make the same amount of mistakes, I'd expect the difference between their SS mark to stay the same as it would be if both were perfect, because it should reflect the actual difference in abilility in that particular component.

 

And anyway a mistake always affect our perception of a program, even subconsciously, so I'm not in favout of an authomatic deduction of PCS either, it would only become a double punishment.

In theory, if whatever mistake makes the performance derail or forces the skater to rush to catch up with music or if because of falls the skaters can't executes the exit they usually do after jumps, therefore reducing the skating transitions (example: Yuzu's COC FS), then that should be reflected in the marks judges give, without any need for any kind of ceiling.

 

So (and I can't believe I'm advocating for judges' freedom in this:facepalm:),  judges should be allowed to judge only following the darn guidelines for each component. Maybe it would make more sense to ask them to write down the reason why they give some sky-high PCS for VERY flawed skates. Sure, they could come up with BS, but at least it would take a little bit of effort on their part, more than pushing a button. AND they would be forced to re-read the handbook:P

 

Anyway, as they're only recommendations (one of ISU's most favourite words), judges will still have the freedom to enforce the ceiling or conveniently forget it depending on skaters (among still competing men it would basically affect only Yuzuru, Javi at Euros, but among Ladies and possibly even Pairs it could become more important. I was going to say ID too, but without V/M, now P/C would be the only ID pair that could be really affected by those new recommendations)

And it's not like their "final goe for SP solo jump not preceded by steps MUST be -3" has changed anything

the only thing that changed is that in the end the requirement got removed altogether:headdesk2:

 

 

 @shanshani, I haven't read the details of your work on the 20 judges bias but I confess I'm surprised you found japanese judges to be that biased, their reputation is somehow that 'they don't know how to help their skaters', tho of course it might depend on the judge.

I will go and see in which panels those judges were (I recognized one by name, so I must have checked that one in the past, the others I had no idea who they were LOL).

I remember the GOE table made by.........ok I don't remeber who:81: (where there were compiled the positive and negative deviation in GOE for their skaters vs the other, from each judge). It looked like Jpn judges were quite fair, compared to the others, but those data were only about top skaters iirc, so I wonder if those Jpn judges' bias is maybe stronger against skaters who are not direct competitors of Shoma and Yuzuru?O.o tho that would be mightly weird:tongueyuzu:

Or maybe they just weren't the same judges or the same competitions...

 

 

Link to comment
Share on other sites

1 hour ago, wingman said:

 

Nice to share! Did you calculate that yourself?

I am super glad that the podium didn't change after dropping those judges' scores. Imagine if it did, all sorts of conspiracy theories would spring up.

I'm just not the "what-ifs" type (until VAR, maybe).

 

Yup, I have a spreadsheet of all the judges' scores that I used to recalculate.
 

 

47 minutes ago, LadyLou said:

@shanshani, I haven't read the details of your work on the 20 judges bias but I confess I'm surprised you found japanese judges to be that biased, their reputation is somehow that 'they don't know how to help their skaters', tho of course it might depend on the judge.

I will go and see in which panels those judges were (I recognized one by name, so I must have checked that one in the past, the others I had no idea who they were LOL).

I remember the GOE table made by.........ok I don't remeber who:81: (where there were compiled the positive and negative deviation in GOE for their skaters vs the other, from each judge). It looked like Jpn judges were quite fair, compared to the others, but those data were only about top skaters iirc, so I wonder if those Jpn judges' bias is maybe stronger against skaters who are not direct competitors of Shoma and Yuzuru?O.o tho that would be mightly weird:tongueyuzu:

Or maybe they just weren't the same judges or the same competitions...

 

 

Japanese judges were less biased on average than judges from the rest of the major feds, but they still showed significant evidence of bias, though maybe not to the extent that it's obvious just from eyeballing protocols. I looked at all scores given by the judge at all major ISU competitions, not just for top skaters. It helps to look at lower ranked skaters' scores as well--judges actually inflate those as much as high ranked skaters.

10 hours ago, Neenah said:

Yes, that PC deduction is the stupidest rule they came up with. I hate rules that only apply to a subset of the skaters  :tumblr_inline_mzx90enaUF1r8msi5:

 

I haven't posted anything because I was just doing it for fun. I have a script to calculate the score (modified the one from the online calculator) and I tested different layouts whenever I had free time. Skatingscores provided the historic context of how the judges normally evaluated specific elements for each skater. 

I would be curious to see this as well. Basically all the simulations I've run put Nathan at a significant disadvantage. Even the one where I gave them both the highest GOEs they've received for each element, refactored, (or in Nathan's case, since I gave him layouts with elements he hasn't done yet, the highest GOE for elements similar to the elements in their hypothetical layout), Yuzu and Nathan were about even assuming both upgraded. This way of modeling shrinks their GOE gap to around 0.5 (note that in the 2018 Olys Free skate, where Nathan was sort of clean and Yuzu was not, Yuzu's average GOE was still close a point higher on a -5 to 5 scale). Nathan occasionally gets high GOEs, but averages much lower than Yuzu--by giving Nathan the max GOEs he's ever earned on each element, I'm basically erasing Yuzu's major advantage of consistently high GOEs. I also gave Nathan 95 PCS despite him never having been near that and Yuzu 97.5 despite that being relatively low for a clean skate from Yuzu.

It's basically only when I start shrinking the GOE/PCS gap to 0.5/2.5 (and these are fairly unrealistic numbers, at least for Nathan) that I start getting Nathan victories for cases where both upgrade or both stay the same. Otherwise, the only cases where Nathan wins in clean skates (and these would be exceptionally clean for Nathan) are where Yuzu doesn't upgrade from 2016-2017 and Nathan adds 4Lo. Basically:

(Note that for these simulations, I had Yuzu doing 3A2T instead of the 3A sequence we've been speculating about. 3A sequence would turn ties into Yuzu wins)

Average GOE/PCS margin 0.5 GOE/2.5 PCS 0.5 GOE/5 PCS 1 GOE/2.5 PCS 1 GOE/5 PCS
Yuzu has 4Lz, Nathan has 4Lo Nathan Victory Yuzu Victory Yuzu Yuzu
Yuzu has 4Lz,
Nathan doesn't have 4Lo
Yuzu Yuzu Yuzu Yuzu
Yuzu doesn't have 4Lz, Nathan has 4Lo Nathan Nathan Nathan Tie
Yuzu doesn't have 4Lz, Nathan doesn't have 4Lo Nathan Tie Yuzu Yuzu


Historically, Yuzu and Nathan's GOE and PCS margins have been muuuch closer to the 1 GOE/5PCS end of the table, or beyond it. (Note that their personal best average GOEs are more than 1.3 points apart on the -3 to 3 scale, and even during the Olys FS when Yuzu made two errors, Yuzu's -3/3 scale average GOE was still 0.5 higher than Nathan's.) I also think that Yuzu is more likely to add back 4Lz than Nathan is to add back 4Lo. So I don't see any reason to be pessimistic.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...