shanshani

I agree that a single judge won't really be able to change competition outcomes except in the most marginal of cases. But I think it's helpful to address biased judging anyway, because it will increase the level of transparency and honesty in the sport. Speaking of judging bias, I finished my nationalistic bias in judging project. This took waayyy too much time. I used to have other hobbies. Results can be found here. Write-up here or at the other place, you know which. Tl;dr: so much biased judging everywhere from everyone. Edit: actually, I'll just post abbreviated results here for the lazy. Wall of shame - Bias highly probable to virtually certain (p < 0.01) Weiguang Chen (China) - PD: 10.1, p < 0.00001 Anna Kantor (Israel) - PD: 9.87, p < 0.00001 Peggy Graham (USA) - PD: 9.38, p < 0.00001 Lorrie Parker (USA) - PD: 5.12, p = 0.000024 Saodat Numanova (Uzbekistan) - PD: 5.76, p = 0.00010 Olga Kozhemiakina (Russia) - PD: 5.19, p = 0.00020 Tatiana Sharkina (Russia) - PD: 4.56, p = 0.000275 Jeff Lukasik (Canada) - PD: 9.56, p = 0.00082 Albert Zaydman (Israel) - PD: 4.83, p = 0.0043 Sakae Yamamoto (Japan) - PD: 4.47, p = 0.0044 Wall of probable shame - Bias quite likely (0.01 < p < 0.05) Masako Kubota (Japan) - PD: 3.67, p = 0.012 Daniel Delfa (Spain) - PD: 5.86, p = 0.012 Yuri Guskov (Kazakhstan) - PD: 4.36, p = 0.027 Janice Hunter (Canada) - PD: 3.63, p = 0.029 Igor Obraztsov (Russia) - PD: 3.04, p = 0.038 Nobuhiko Yoshioka (Japan) - PD: 4.74, p = 0.046 Wall of honor-or-at-least-no-shame-yet - No strong evidence of bias (p > 0.05) Na Young Ahn (South Korea) - PD: 2.48, p = 0.13 Wendy Enzmann (USA) - PD: 1.75, p = 0.17 Sung Hee Koh (South Korea) - PD: 1.5, p = 0.20 Agita Abele (Latvia) - PD: 1.4, p = 0.22

I don't know if the ISU would actually be interested in a serious and systematic investigation of nationalistic bias... Anyway, in non judge-bias news, in light of Yuzu's apparent fondness for 3A sequences at FaOI, I went ahead and did some calculations comparing his layouts if he replaced his 2T with a 3A sequence. Turns out the 3A sequence layout has higher BV by 2 points. So we might actually see him do 3A sequences in competition? He's better off attaching it to a 3Lz (or maybe 3F, if he wants to keep at least one jump of each type) instead of a 4T though, because +SEQ lowers the BV for the whole combination. 4Lz, 4Lo, 4S, 4T // 4S3T 3A1/2Lo3S 3Lz3A (+spins/steps/choreo) BV = 99.7 4Lz, 4Lo, 4S, 4T // 4S3T 3A1/2Lo3S 3A2T (+spins/steps/choreo) BV = 97.7 If he does this, his BV will actually be only one point short of the following Nathan layout, which I think is Nathan's highest BV realistic layout for the upcoming season (assuming Nathan doesn't get back his 4Lo), a BV gap of essentially nothing: 4Lz3T, 4Lz, 4S, 3A // 4T3T 4F 3Lz1/2Lo3S (+spins/steps/choreo) BV = 100.74

Still working on my judge bias assessment project. Literally every single judge (Parker, Chen, Kubota, Yamamoto, Delfa, Kozhemiakina) I've checked so far shows evidence of home country bias. The only difference is the extent of the bias (Chen > Parker > Delfa > Kozhemiakina > Yamamoto > Kubota) and how strong the evidence is (Chen > Parker > Kozhemiakina > Yamamoto > Kubota = Delfa).

I mean, I think the lesson is more like "be American, not Chinese" because Feng Huang cheated less than Lorrie Parker and still got suspended.

Haha, thanks. I'm in the middle of switching jobs atm, so I have a lot of time on my hands. Also, I think part of the reason I like figure skating is because it feeds my love of spreadsheets. It's super cool that I can figure stuff out about the sport with just some math. Now that I've thought about it more, I think I've got a pretty good method for sniffing out judging bias--way better than what the ISU uses at any rate (seriously, did they consult an actual statistician at any stage of the process of deciding how they were going to look for nationalistic bias? Not even an actual statistician--someone who did well in Stats 101 should be fine). Gonna run some more numbers, but from what I've been able to see by testing the method on Lorrie Parker's scoring history, it's both pretty powerful but also conservative in its assumptions, unlike some of the calculations I was doing earlier. (Though not the 0.11% calculation--that one is fine and employs the most conservative version of the methodology I'm testing. Seriously, I cannot overstate how obvious Lorrie Parker's nationalistic bias is.)

Yeah, the decision not to investigate Parker is kind of absurd. She graded the USA men about 9 points above their actual scores on average, which in and of itself is a huge red flag, even ignoring her low-balling of Yuzu and Boyang. I actually did some math and got a 0.03% chance that that kind of difference would occur by random chance. Yeah, you read that right. 0.03%--or in other words, 3/10,000, way, wayyyy beyond the threshold of statistical significance. Edit: the first calculation was found by comparing Parker's score deviations for US men to score deviations in general for all judges in the Oly's men FS. Using a more conservative comparison, comparing only Parker scoring US men versus Parker scoring everyone else, I still got only a 0.06%~0.11% chance that her two average are that far apart due to random chance (note that on average, she scored non-US men 0.38 points below the other judges, and she scored US men 9.24 points above other judges--that doesn't even pass the sniff test). Basically however you look at it, Parker is super, super biased.

Brb gonna make myself feel like I wasn't wasting my time. ...Actually looking at it more closely, it's more suited for converting old scores than scoring (new or old) programs by hand. Plus I don't think it handles jump sequences, which my spreadsheet should. And the +REP calcs seem to be off (I think it's multiplying by 0.8 instead of 0.7). So my spreadsheet should also be more accurate?

I was bored so I made an excel document that will auto-score a program for you under the new SOV if you enter the program element abbreviations, GOE, and PCS. It works for both singles disciplines (though make sure you remember to change the discipline cell so that the PCS gets factored correctly). Doesn't work for pairs yet, but if there's interest I might work on adding it. If anyone wants to go and rescore old competitions (make sure to drop a jumping pass for mens) or wants to play the judge at upcoming ones, the sheet will do all the math for you, including looking up base values/dealing with obscure rules regarding how BVs are calculated for jump sequences/whatever, as long as you enter the elements correctly. It can be found here. Make sure to download it, as it isn't editable on google drive. All cells that are light blue are cells where you have to enter information. Red cells are cells that give you various values of interest based on what you input. White/grey cells are chiefly used as steps in the calculation and can be ignored. Further instructions for how to read/use the sheet can be found in the "How to use + Abbreviations" tab. As far as I know, the sheet can handle all regularly occurring element errors (eg. <, e, REP, SEQ) and other rule issues, but if anyone finds a situation which the sheet doesn't handle correctly, let me know. (I thought I saw a website that did something similar, but I couldn't find it again, so *shrug*.)

Hm, I didn't specifically look up Nathan's 3A success rates. Well, he did land them with positive GOE in his Olys FS and Worlds SP and FS, so it's possible he got better towards the end of the season. And he's almost certainly going to work on his 3A over the off-season, so I wouldn't count him out on repeating 3A, especially since he's done it before. But thinking about it more, there really isn't that much of a difference between 3Lz3T and 3A2T points-wise, so it might actually make more sense for him to go for the 3Lz and repeat 3T instead of 3A. Re: Yuzu's 4Lz--I agree that the chances of seeing it in the beginning of the season are low, but I do think he'll try to get it back by the end. I guess it depends on how the season is going though. The new rules should in theory help Yuzu substantially, so he might decide it's not worth the risk this season, depending on how his rivals are doing under the new rules regime. If he wants to break his FS and overall score record again though, he's going to need to upgrade from his 2016-17 layout at some point, unless the judges are extra generous with +5GOEs. Plus, if he doesn't upgrade, my calcs have a clean and somewhat strictly scored Yuzu (3.5 average GOE, 98 PCS) just a little too close than is comfortable to a clean and generously scored Nathan. Granted, I did score hypothetical Nathan fairly generously (95 PCS, 2.5 average GOE) and I also gave Nathan repeat 4F instead of repeat 4T (and decided that 3Lz3T was worth it over 3A2T for Nathan after all), but I think Yuzu would not want to be so dependent on the judges' whims. (You can find the calcs here under the "hypothetical layouts" tab. The more realistic layouts are towards the bottom.) Nathan for whatever reason seems more comfortable repeating 4F (he's done it regularly the past couple of seasons). So this is more based on past behavior than anything. Edit: Note that 95 PCS and 2.5 GOE is not that crazy based on what Nathan has been getting. While I do think it's more likely than not that his PCS will go down a little from his 2018 Worlds showing because he will no longer be getting the boost from following that complete mess of a final group, and at the end of the day I don't think judges will put Nathan above Shoma on PCS (and Nathan shouldn't go above Shoma on PCS if both are clean), but given score inflation and narrowing of PCS margins, it's not totally insane that he could get a 95. Nathan is also a pretty good performer with good musicality when he doesn't have to sacrifice his choreo in order to land quads, so it's also plausible that he'll genuinely improve significantly in PCS areas if he can figure out how to do both choreo and quads at the same time (and not just receive higher scores for whatever reputation or fed power or other reason). On the GOE side, while Nathan averaged (an imo generous) 2 in his GOEs (when refactored on the -5 to 5 system) in his Olys and WC free skates, if he had cleanly landed the couple of jumps he got negative GOE on in those skates with +1 average GOE on the old system, he would have averaged about 2.4 GOE on the new system. So it wouldn't be out of question for a truly clean Nathan (no negative GOEs) to average 2.5 GOE under the new system.

Yeah I'm very uncertain about whether Nathan will choose to repeat 4F or 4T. 4Lz seems unlikely. I only went w/ 4T because of what I've heard about his practice, but I don't follow Nathan that closely so I could be wrong. 4F makes more sense from the perspective of raw point potential. Even though Boyang hasn't repeated 4Lz before, I think it's the easiest way to eek out more BV and potentially GOEs for him and keep him competitive. Right now he lags Nathan in BV, Shoma and Yuzu in PCS (and starting to lag Nathan...), and Yuzu in GOE, without being better than them at anything in particular (except his 4Lz!), making it hard for him to win even when he skates well. Therefore, he needs to compete on all fronts, and repeating 4Lz is a way to do it without taking the focus away from improving skating skills IMO. I think there's a good chance Yuzu will break the SP record as long as he skates clean and executes with his usual degree of quality. It really depends on how GOEs are judged. If the judges happily use the whole range of GOEs, then it should actually be easier, not harder, for him to beat his old WR under the new system, because the total points from high quality elements goes up instead of down despite the BV reductions (eg. 4Lo +5GOE is 15.75 points, whereas 4Lo +3GOE used to be 15 points). If they're stingy about 4s and 5s then it'll be harder.

Interesting. I glanced at the figure skating tieba, but my Chinese sucks pretty hard so if I don't put in a lot of effort to translate, I only get about half of what's being said. But I also got the sense that people were assuming that this decision was the result of politicking in response to various calls being made against China. Hopefully whatever's wrong will be fixed in the next couple of seasons. As for layout guessing, if Yuzu regains 4Lz (I'm going to be optimistic and say this is more likely than not): 4Lz, 4Lo, 4S, 4T // 4T3T 3A1/2Lo3S 3A2T No reason Yuzu wouldn't backload his combos, since that's what he does already. It's slightly riskier without the solo jump at the end, but if he botches an early combo he can just do 3T or 1/2Lo3S instead of 2T without losing too much BV (2T is only 1.3 pts). 4T3T could be 4S3T, but I think 4T3T is more likely. If he skates this layout clean, it should be possible for him to break his WR. If no 4Lz: 4Lo 4S 4T 3F // 4T3T 3A1/2Lo3S 3A2T 3F over 3Lz because he's put in too much work on his flip to give it up for a measly 0.6 BV. Probably no world records with this, but should be perfectly competitive with the changes to SOV. Nathan: I'm not optimistic about Nathan re-adding 4Lo next season, because I think between Yale and his iffy 4S and 3A he has too much other stuff to work on, and I suspect 4Lo is actually the hardest non-axel quad. So: 4F3T 4Lz 4S 4T // 4T1/2Lo3S 3A2T 3A Nathan's layout is a lot harder to guess than Yuzu's. He might repeat 4F instead of 4T, but apparently he's been working on his 4T, which to me suggests he's trying to turn it into a high GOE element and also tack on a harder combo. I think he's going to repeat 3A instead of going for 3A and 3Lz as his triples, as he has repeated 3A in the past. With the new rules buffing GOE skaters and nerfing BV skaters, he can't afford to give up too many points. If he does go for 3Lz, it would be in order to repeat 3T instead of 2T (which I think works out to slightly higher BV but less potential for GOE points. Backloaded 3A2T is 10.23 and 3Lz3T is 11.11, but the GOE bonus is calculated off 8 and 5.9 respectively. Granted, the GOE has to be pretty high for 3A2T to > 3Lz3T. But keep in mind that 3A and quads tend to get graded more leniently.) Boyang: Not foreseeing major layout upgrades because Orser is probably going to make him work on skating skills/other aspects of his performance first. BV isn't quite as important as in the past now anyway. Only upgrade of note might be to repeat 4Lz? 4Lz3T 4Lz 4S 3F // 4T2T 3A1/2Lo3S 3A Dropping 3Lz might make his 4Lz more stable versus dropping 3F. Boyang gets relatively high GOE on his 4Lz, and so it might be worth repeating over 4T, especially since the amount of GOE points available for 4Lz is more than for 4T. Leaving solo 3A at the end allows him to earn back BV if he botches an earlier combo. Not sure about other skaters. My impression is that Shoma has not had a good success rate on his 4Lo, nor has Mikhail had a good success rate on 4Lz. The new rules punish falls a lot more harshly, so I think those two are either going to have to improve significantly on those two jumps or drop them.

Speculation in other places is that it's retaliation/pressure over (what China sees as) unfair judging in the Winter Olympics across various events. I think the main complaints are short track, aerials, and maybe not the judging itself but the investigation of two Chinese figure skating judges for judging bias (one of whom totally deserves it, the other of whom I think is being unfairly singled out versus equally bad or worse judges coughUSA men's judge Parkercough). It's just speculation, but apparently China has pulled out of hosting all winter sporting events for the upcoming season. *Shrug*

Actually, Yuzu's clean skates didn't go down very much at all, even after deleting a jumping pass--2017 Worlds FS only went down 4 points, 2015 GPF FS went down 2 points (and Yuzu's score actually went up slightly across the whole competition, because clean+high quality SPs tended to go up in score). This is because the gain to GOE points outweighs the loss of BV of the quads and partially offsets the loss of a jumping pass. If Yuzu can skate something very similar to his original planned layout for 2017-2018 minus the lowest scoring jumping pass (so this would be a layout of 4Lz, 4Lo, 4S, 4T, and backloaded 4S+3T (or 4T-3T, there isn't a big difference), 3A+2T, 3A+1Lo+3S and get around GOE +4 on all his elements and 98 PCS, he would score about 230, according to my calculations. Edit: I went ahead and added a page to the spreadsheet calculating scores for hypothetical layouts, so you can see how I got to that number. I also calculated hypothetical scores for a Nathan Chen-like skater, for comparison purposes and to see what someone like Nathan would have to do in order to break Yuzuru's WR. I gave him a fairly maxed out layout that I'm fairly certain Nathan would need a season or two of development to pull off well, +3 GOE across the board, and 92 PCS (a smidgen higher than he got at Worlds). Overall, it's pretty generous. Even with those stats, he only matches Yuzu's 2017 Worlds FS record under the old system. To be honest, I feel like they made it a little too difficult to compete on BV--that one point GOE advantage across the program I gave 'Yuzu' completely obliterates 'Nathan's' BV advantage.

Figure I might as well post this here too, since it's probably of special interest to you guys and I put way too much work into if for me not to try to show everyone. I went back and rescored some old competitions under the new scale of values + backloading rules + one less jumping pass in men's (didn't include the quad repetition rule because it would have messed up the calculations too much). Here's the link for men's and here's the link for ladies, and you can find the write up of how I got the results and a bunch of other gory details on goldenskate (are we allowed to link gs?) or on reddit. The long and short of it is that under the new rules, Yuzu benefits a lot.* For instance, Yuzu would have won 2017 4CC by almost 4 points. Moreover, Yuzu's 2018 Olys skate beats Javi's personal best by 2 pts, Shoma's by 3, and Nathan Chen's pb by 7 points. There's almost a 20 point gap between Yuzu's PB and everyone else's. Even when Yuzu doesn't skate clean, his scores hold up much better than almost all of the rest of the field's (notable exception being Javi, who also gets high GOEs and sees a similar effect on his scores as Yuzu)--note that everyone's scores fall because of losing the 8th jumping pass, but Yuzu's falls less than everyone else. Nathan loses like 10 points on average, whereas Yuzu loses 4 (at least in the competitions I rescored). In one case, (GPF2015), his score actually goes up, which is kind of absurd if you think about it. Losing a triple + quad BV getting nerfed, yet he winds up with a higher score off of GOEs. *At least if you assume judges will award GOEs in proportion to how they award them now. It's a pretty big *if*, to be fair.

The reason we might see more 3Lo combos is because of the reduction in number of jumping passes. Before, a lot of men jumped 3Lo as a solo jump, preventing them from jumping it in combo (presumably you have higher priorities for what to repeat than 3Lo). With one less jumping pass, a lot of men are going to be dropping their 3Lo solo jump, freeing it up to be used in combination. Most men jump 2T as one of their combinations in the free skate because of the Zayak rule, but 2T has very low value in comparison to 3Lo. Therefore, if you can replace your 2T with 3Lo, your base value goes up by a decent chunk (more than what it would go up if you replace a 3A with a 4S for instance). Some people speculate that men have a harder time with loop combos due to their hip structure though, so whether it will actually happen is not certain.