Score for Score

Posts tagged "Olympic Talk"

The qualification round in Tokyo had some surprises for Team USA. By and large, they did not receive the scores they were expecting to receive. They were supposed to sail into the finals in first place, but instead they finished a full point behind the Russian Olympic Committee. I want to talk about why that happened.

But before I do that, let me be very clear: the US team earned the second highest team score in the world and qualified two athletes into every single event final. They should be incredibly proud of their performance and of all the hard work it took to get to this point. I am going to point out a lot of mistakes in this post; they are elite athletes and it’s fair to critique their performance. I still have the utmost respect for these gymnasts and all they do.

So. I’ve seen three main explanations for yesterday’s results:

  1. Team USA didn’t do their best gymnastics. They normally perform at a higher level than they performed at in qualifications.

  2. Team USA finally got the artistry and execution deductions that they deserve. The judges at US Olympic Trials turned a blind eye to some glaring issues, but international judges at the Olympics will do the Americans no such favors.

  3. Team USA was judged unfairly. The judges were biased and scored other athletes more leniently than they scored the US.

Let’s take these explanations one by one.

Team USA didn’t do their best gymnastics.

Several key routines went worse than they should have. Even the New York Times sports desk is tsk-tsk'ing over Simone Biles's performance. More importantly, Biles herself was clearly disappointed: she had wildly uncontrolled landings on three events.

We also saw some unexpected flops from Jordan Chiles. She had a fantastic domestic season, hitting 24 out of 24 routines. No other athlete on the US Olympic team can say the same. Yet she brushed her feet on the mat on bars and had two falls on beam. The US was hoping to count strong scores from her on those events, and they had to rely on McCallum’s typically lower-scoring routines instead.

Yet the more I looked into this, the more I began to feel that these big, obvious mistakes distracted from many smaller mistakes that really added up. Let’s use Suni Lee’s floor routine as an example. Her scores at Trials had me thinking that she had a shot at floor finals, and her routine in Tokyo yesterday didn’t look so different from her routines in St. Louis. She’s also an athlete with relatively few built-in deductions in her dance elements — good extension, good splits, good ring positions, etc. But then I took a closer look.

Landings are one place where deductions add up fast. At the end of a good pass, the gymnast should have her chest upright, almost like she’s sitting on a chair. Her movements should be under control and she should be able to stand right up. A low chest position, with the torso close to the knees, is up to 0.3 points off. A step greater than a meter is another 0.3. Four tumbling passes later, you’re looking at an execution score in the low 8s for an otherwise clean routine.

Lee had a much rougher landing on her last pass in Tokyo than she did at Olympic Trials. What’s more, she added in a fourth pass, a double layout, and landed that poorly as well. The points she gained in difficulty were easily offset by the execution deductions.

I’m not picking on Lee — there were issues like this across the board. The USA just wasn’t doing the precise, clean gymnastic that they’re capable of last night. The attention to detail just wasn't there.

Team USA finally got the artistry and execution deductions that they deserve.

This narrative states that the US was misled by overly generous domestic scores. International judges are less tolerant of the US chucking their jumps, falling out of their turns, and phoning in their choreography.

Let’s start with artistry deductions. I doubt these played a large role. Pure artistry deductions on floor exercise are at most 0.5 points, a deduction taken when “the exercise is not connected to the music itself.” In past quads, the full allowable artistry deduction has rarely been taken. And even if this deduction was taken in full measure for every single US athlete on floor, it wouldn’t explain the scoring gap on other events.

However, I think it’s very likely that other execution deductions played a role. We know from Kara Eaker’s experience that ring positions are not evaluated in the US the way they are internationally. Even after this discrepancy had disastrous consequences at 2019 World Championships, Eaker’s routine remained largely unchanged and her ring positions did not dramatically improve. I take this as strong evidence that the US national program has not made sufficient effort to align its understanding of execution evaluation with international standards.

Of the sixteen routines from athletes on the US team, twelve1 scored higher at Olympic trials than any routine that athlete has ever performed on that event internationally. For example, Suni Lee scored a massive 14.733 on beam at Trials on Night 1. It was a fantastic routine. But she has never, ever gotten higher than 14.150 in international competition, and that was at Jesolo. Now, gymnasts intentionally wait all quad to peak for Olympic Trials, so we might expect to see some unprecedented performances — but the extent of this phenomenon should certainly give the National Team Coordinator pause.

The qualifications experience should give next quad’s National Team Coordinator clear marching orders: listen to your brevet judges. Pay attention to the FIG’s scoring videos. Get your standards up.

Team USA was judged unfairly.

By this point in this article, I hope it’s clear that this is the weakest of the three arguments. The US made many real mistakes. And scrolling through the results, I didn’t see a “smoking gun” - a case where a US athlete’s routine was outscored by clearly inferior routine. There are cases that are close — Ferrari and Biles come to mind — but none so egregious that I have no choice but to cry bias.

Then again, I watched four routines at once on the apparatus feed well after midnight, so I might have missed something. Judging across qualification subdivisions is notoriously inconsistent at every international competition, and it would be folly to take fair judging as an article of faith. If you’ve got a smoking gun, leave it in the comments below.

What’s Next

Qualifications are just qualifications. The US has placed two athletes in the all around final as well as every single event final. And in all likelihood, they will return home with plenty of hardware, including a team gold, and qualifications will be a distant memory.

But if they want that to happen, they’re going to have to step it up. Russia has an incredibly talented team and they deserved to come out on top yesterday. I cannot wait for what is sure to be the most exciting team final in a decade.

1Biles on BB & FX, Lee on VT, UB & BB, Chiles on VT, UB, BB &FX, and McCallum on VT, BB & FX

Tags: Olympic Talk

It's time to talk about US Olympic team selection.

Often by this point in the Olympic year, we have a pretty good idea of who might make the team. Some athletes are performing at a consistently higher level than others. Simone Biles, of course, is performing at the highest level ever seen. But the other team spots are somewhat murkier - and that's because we don't actually know what the selection committee is looking for.

Marta Karolyi's team selection process was the stuff of NBC legend. She picked athletes who would contribute to the team over athletes who would win individual medals. She loved a reliable beamer. She had a chip on her shoulder about US bars. She liked her gymnasts well-behaved. But most of all, she cared about consistency.

Tom Forster, however, has never picked an Olympic team before. On at least three occasions -- the 2018 World Championships, the 2019 World Championships, and the 2019 Junior World Championships -- he has shown a strong inclination to simply select the top all-arounders at the last selection camp/competition before the meet. Reading between the lines of his interviews, he seems to find a compelling fairness in this approach; it takes favoritism out of the decision entirely. What's more, with a four person team and four-up qualification in Tokyo, it's not hard to argue that every person on the team should be able to go up on every event.

So there's good reason to believe that Forster will pick the four-person Tokyo team based on the all-around standings at the end of the Olympic Trials. And with the top two all-around finisher guaranteed to get team spots according to the USAG selection procedures, fully half the team is guaranteed to be chosen in this manner.

This approach - the Forster strategy - has basically proven safe, if unstrategic. But it's only safe as long as two things are true. First, the US has to be guaranteed a comfortable margin of victory in the team final. They can't be in a situation where, in order to win, they need the extra tenths provided by, say, taking a VT/FX specialist and a UB/BB specialist instead of two all-arounders with no standout event. And second, gymnasts need to be performing consistently enough that their all-around performance at a single competition is a good proxy for how they'll perform at the big event. You don't want to pick a gymnast for her all-around performance at Trials if her performance in training and at other meets demonstrates that Trials was a fluke.

Now, I think that any US team with Simone Biles is probably guaranteed a comfortable margin of victory. But with several major meets come and gone -- Winter Cup, the American Classic, the US Classic, and the first day of the US National Championships -- only Simone Biles and Jordan Chiles have put together multiple truly solid performances. Everyone else has been surprisingly... splatty. We have certainly seen some amazing gymnastics. But most gymnasts with routines of the caliber that the US would like to display in Tokyo have yet to really hit those routines more than once.

Of course, many gymnasts are combing back from injury, and all of them are somewhat rusty after an extended COVID-induced break from competition. And everyone is planning to peak for Trials. But the inconsistency we've seen thus far throws the Forster strategy into doubt. We just won't have any way of knowing if gymnasts' all-around performance at Trials is a good predictor of how they'll look in Tokyo. Two days of competition is simply not enough time for a pattern to emerge.

Finally, I have to wonder if Forster's apparent intention of picking the team based on the all-around standings at Olympic Trials has perhaps caused more inconsistency in the lead up to Tokyo, simply because the pressure is off at the earlier competitions. Less pressure is generally a good thing. No one wants to return to the days of "the watchful eye of Marta Karolyi." And countless former US Olympians have described how the need to be in top form for weeks and weeks and weeks before the Olympics left their bodies and minds broken by the time the games began. But none of this is intrinsic to informing the athletes that their consistency across multiple competitions will factor into team selection. You don't need to pick the Olympic team based on all-around standings at a single competition in order to ensure that elite gymnasts are happy and healthy as they pursue their goals in the sport.

So here's what I hope.

I hope that Jordan Chiles's three consistent all-around performances, each earning scores at or over 56.9, will earn her a spot on the team even if she falls once or twice at Trials.

I hope that if Sunisa Lee's sore Achilles hobbles her on vault and floor and knocks her down the all-around standings, she still makes the team for her standout bars and highly usable beam.

I hope that Jade Carey is left off the four-person team even if she places top four in the all-around at Trials.

And I hope that the fourth team spot doesn't go to whichever of Grace McCallum, Kayla DiCello, Leanne Wong, Morgan Hurd, MyKayla Skinner, Skye Blakely, Emma Malabuyo, etc. etc. etc. manages to eke out a tenth over the rest of the pack by the end of Trials if one of the others has looked consistently better in training. I hope there is more rhyme and reason than that.

But we'll have to wait and see.

Tags: Olympic Talk

The most recent version of British Gymnastics’ Tokyo 2021 selection policy lists a highly unusual selection criteria: potential to medal at the Paris 2024 Olympic Games. In other words, team selection won’t just be about pick the team with the best shot at winning medals in Tokyo. The federation also has an eye for the future: it wants to pick gymnasts who will go back to the Olympics and win more medals.

This is an absurd policy for a number of reasons. First, we all know there are no guarantees in gymnastics. Injuries happen, bodies break down, and people simply decide to move on from the sport. The idea that anyone can make a useful guess about what the medal potential of British WAG will be in 2024 is laughable. Second, this policy hilariously unsustainable: if every Olympics you pick the athletes who you think will do best at the next Olympics, you waste all the investments you’ve made in the past to give athletes much-needed international experience. And finally, it seems designed to be used as a post-hoc justification for leaving Becky Downie off the team.

But I want to put all that to the side for a minute and talk about the assumption that seems to lie beneath this rule: younger athletes have a better shot at making it back to a second Olympics.

It’s always seemed to me that gymnasts who are able to compete at the elite level past a certain age have developed a certain longevity that most elite gymnast never figure out. Gymnasts who are sixteen seem to be at much greater risk of getting injured or flaming out than gymnast who are still competing at 22. To compete at such a high level well into your twenties, an athlete must have figured out some training regimen that works sustainably. (In all likelihood, she must be biologically blessed as well.)

These observations are largely anecdotal. In 2016, I certainly thought that 41-year-old Oksana Chusovitina had a better shot at making it to Tokyo than, say, the perpetually injured 19-year-old Madison Kocian. The Wevers sisters both peaked well into their twenties, and German veterans like Kim Bui and Elizabeth Seitz have outlasted promising newcomers like Tabea Alt. And despite Simon Biles’s threats to retire at age 24 after Tokyo, it’s clear that she could come back for Paris if she wanted to.

So I turned to the data to better understand the trends. Can we truly expect more longevity from young athletes than from older ones?

First, let’s clear one thing up: a lot of gymnasts make it to multiple Olympics. Of the 1,454 gymnasts who have competed at an Olympics since 1928, 270 have come back for a second. That’s 18.5%. And this trend isn’t completely driven by the early days of gymnastics, when Olympic gymnastics was a sport for adult women that didn’t take as much of a toll on the body. Of the 846 gymnasts who have competed at an Olympics since 1976, 141 have come back for a second (16.67%).

Now, let’s look into returning Olympians by age. I calculated the average probability that an Olympic athlete of each age will return for another Olympics in the future. The results are in the plot below. The plot uses data for 1976 onwards, and I cut it age 32, after which point we’re literally just talking about Chuso.

Younger athletes certainly have a good shot at longevity: about 19% of 16-year-olds return for another Olympic Games. And at first, it does seem as though older athletes have less of a shot at making it back — only 10% of 19-year-olds return. However, the gymnasts who are still competing well into their twenties buck this trend. 22% of the 23-year-old Olympic gymnasts have returned for another games, and 25% of the 26-year-olds have as well. From this chart, we can see that it is not universally true that older athletes are less likely to hold on until the next Olympics.

I also wanted to understand if gymnasts who reach a high level at a very young age are less likely to have long careers. To examine this, I calculated the share of gymnasts who make it multiple Olympics based on the age at which the gymnast first competed at the Olympics.

Here, we see a similar trend. Gymnasts who first go to the Olympics before age 17 have a good shot at making it back — over 15% of them return for at least one more Olympics. These are the athletes who can attend multiple Olympics at age 20 or younger. After that, the likelihood of returning falls off; only 9% of those who first make it to the Olympics at 19 or 20 come back for more. But, yet again, we see this trend reverse — gymnasts who’s first Olympics occurs at age 21, 22, or 23 are more likely to make it back. These are the athletes who peaked late in their career: Jessica Lopez, Marta Pihan-Kulesza, Diane Dos Santos, Lyudmila Grebenkova, and more. After age 23 this trend drops of — mostly because there are just so few gymnasts who make it to their first Olympics when the are older than 23. But even so, we can see again that older athletes can have more longevity than younger athletes.

In this context, British Gymnastics’s selection criteria simply doesn’t make sense. In general, we can expect young athletes to have more years left in elite gymnastics than older athletes. But true longevity often doesn’t emerge until athletes reach their early twenties. When we think about which of the gymnasts in Tokyo will make it back for Paris, there are many adult women who are at least as likely to make it back as there are fresh-faced teenagers.

In gymnastics, as in so many other things, age is just a number. And if there’s one thing I’ve learned as a data scientist, one number never tells the whole story.

Tags: Olympic Talk

Gymnastics fans love to debate the strategic merit of adding upgrades to a routine. How fierce is the competition? How consistent is the skill? When is it worth the risk? And the most tantalizing scenario is one in which a gymnast who played it safe during qualifications decides to go all-out for the finals, risking a fall for the extra few tenths in difficulty that could end up making all the difference. How common is this? I turned to the data to find out.

There’s an important reason to talk about this now: Baku. The 2020 Baku World Cup ended in a state of limbo: between qualifications and finals, the Azerbaijani government banned all sporting events in response to COVID-19, so finals never took place. A few weeks later, the FIG announced that scores from the qualification round of the 2020 Baku World Cup would be treated as the final results. The prize money, the medals, the glory, and — most importantly — the Olympic qualification points have been awarded on the basis of qualification scores. Gymnasts like Vanessa Ferrari and Emma Spence are sort of screwed by this decision.

A lot of questions can’t be answered fairly during a global pandemic, and “What to do about the aborted Baku World Cup?” is probably one of them. But the decision to treat qualification results as final struck many as particularly unfair because of gymnasts’ tendency to upgrade for finals. Gymnasts picked their qualification routines under the assumption that those routines would not determine their final placement, so the qualifications results don’t reflect a fair ranking of each gymnast putting her best foot forward. Given the chance to upgrade for finals, we might have a different set of athletes going to Tokyo 2021.

This context lends extra weight to the question: how often do gymnasts really upgrade for finals? I pulled data from the last five event world cups, and compared the difficulty scores that gymnasts got during qualifications and finals to determine which routines were upgraded. Of course, d-scores don’t perfectly measure intentional, strategic changes in routine composition: sometimes a d-score goes up because a gymnast got her double turn around a little further, or got her hips a little closer to a laid-out position. In other words, we don’t know if a higher d-score represents strategy or better execution. But it’s the best information we have.

Here are the main results:

In these five world cups, 22.5% of the routines performed during women’s event finals were more difficult than that gymnast’s qualifying routine. That suggests a lot of upgrades — or much higher levels of performance that cause way more skills to be credited. The average routine only increased in difficulty by 0.003 points between qualifications and finals, but 14.3% of routines increased in difficulty by more than 0.2 -- which can make a big difference.

I also ran the the same numbers for the men’s competition. For men, 30% of the routines in event finals were more difficult. The average MAG routine increased in difficulty by 0.02 points, but a full 22.5% of routines increased in difficulty by more than 0.2. I’m not surprised that it’s way more common for men to upgrade: there’s more of a culture of “chuck it and see what happens” in MAG.

The share of downgraded routines for both genders is quite surprising. You never hear talk of gymnasts intentionally playing it safe during event finals, so it’s likely the case that most of these “downgrades" occur due to poor execution: gymnasts don’t get credit for all their attempted turns, connections, twists, or body positions. To understand this a little better, I broke down the results by event.

First, both upgrades and downgrades are way less common on vault — you can always count on Maria Paseka to whip out something crazy for finals, but she’s the exception rather than the rule. Upgrades are also less common on bars. To my mind, there’s fewer unintentional upgrades on bars and vault — there are no turns, fewer ambiguous body positions, no questionable connections, etc. This suggests that a lot of the increase in d-scores is just a result of better execution rather than strategic decision-making.

For men, we see the most upgrades on pommel horse and high bar. I don’t follow MAG so closely, but this result tallies with my impression of where and when we hear gymnasts talk about planned upgrades. The level of consistency on still rings is astonishing. Could that event get any more boring? (*ducks*)

In summary, we see a lot of cases where gymnasts perform a higher difficulty routine during event finals. If the gymnasts that qualified to finals in Baku would have upgrade their routine more than a fifth of the time, then it’s absolutely absurd to treat qualifications results as finals. However, we can’t be sure if all the upgrades in the data are due to strategy or simply better execution. There’s only one way to know for sure which gymnasts truly plan to upgrade for finals: ask them.

Tags: Fun with Score Data, Olympic Talk

In past quads, the US Olympic Team has featured a first-year senior. The story is a compelling one: a fresh face rises through the ranks, peaking at the perfect time to make her dreams come true. Laurie Hernandez, Kyla Ross, the archetype herself Dominque Moceanu… the list goes on.

With under a year until the next Olympic Games, the question has been coming up more and more: which first year senior will on the team in Tokyo?

If all the seniors are healthy, I’m guessing none of them. While we might see Kayla DiCello and possibly Olivia Greaves as an alternate, the numbers suggest that the four team members and two individual athletes will all be gymnasts who are already seniors.

I’m not the first to make this prediction, but I wanted to bring in some numbers to back it up. So lets look back at some scores.

The year before the Olympics, the juniors who ultimately made the team were always scoring competitively with the seniors above them. In 2015, a year out from Rio, Kyla Ross scored 117.650 over two days of competition at the junior national championships. Among the seniors, only Jordyn Wieber bested that score. (To be fair, nationals that year was a bit of a splat-fest.)

It was less obvious that Laurie Hernandez would make it to the top. She scored a 117.5 at nationals in 2015. That score would have given her sixth place amongst the seniors, behind Simone Biles, Maggie Nichols, Aly Raisman, Bailie Key, and Gabby Douglas. On the one hand, Hernandez got this score even with a performance on the first night that was almost two points of short of her potential. On the other hand, there’s a strong possibility that she would not have made the 2016 team if either Nichols or Key had been in anywhere close to top form.

So lets look at the same numbers for the juniors trying to make it to Tokyo. Of the ten juniors on the national team this year, seven will be age-eligible for Tokyo: Kayla DiCello, Olivia Greaves, Ciena Alipio, Anya Pilgrim, Sophia Butler, Lilly Lippeat, and eMjae Frazier.

Here’s how five of these juniors would have placed compared to the seniors using their two-day all-around scores. (I didn't realize Lippeat and Frazier were age-eligible when I did this analysis; thanks to Cordelia Price forpointing it out.)

GymnastScoreSenior Placement
Kayla DiCello112.7004
Olivia Greaves110.8506T
Ciena Alipio107.30015
Anya Pilgrim107.05015
Sophia Butler105.35016

Only DiCello is at the level we’d expect to see going into an Olympic year. However, many senior gymnasts ranked below her had unexpectedly poor showings, and McCusker withdrew entirely, rendering this a somewhat skewed picture of the senior field. What’s more, DiCello’s performance was pretty close to the best she can do without upgrading. “4th place amongst the seniors” is really the absolute best-case scenario for her.

Of course, all-around performance isn’t the only relevant factor. Given the individual spot(s) available in 2020, a standout event could really help a gymnast make it to Tokyo.

To look at this possibility, I identified the event on which each gymnast would have placed highest amongst the seniors over two days of competition at nationals. For vault, I’m looking at the scores for each gymnast’s first vault over both days of competition, since that’s what’s relevant for a team competition. These scores are below:

GymnastEventScoreSenior Placement
Olivia GreavesUB28.9503
Kayla DiCelloVT29.5003T
Ciena AlipioBB27.8008
Sophia ButlerVT29.1508
Anya PilgrimVT29.1008T

DiCello and Greaves look a little more competitive here. DiCello’s DTY outscoring Skinner’s and tied with Chiles's. Greaves beat out Biles on bars. But with a four-person team, “top 3 on an event” isn’t necessarily enough unless you’re top 3 on other events as well. DiCello also would have come in 4th on floor, which helps, but still doesn't quite make her a standout on any event, especially because her DTY alone wouldn't help her qualify to the vault final.

Given these scores, and the precedent set by Hernandez and Ross, it seems unlikely to me that any of the current juniors will be selected to represent the US in Tokyo unless multiple seniors are injured.

Now, is this a sign that the US developmental program is failing? Or that the turmoil at USAG has ruined an entire generation? Absolutely not. We shouldn’t expect talent to be equally distributed across all birth years, and in some ways the US seniors field has never been deeper.

Of course, I’d love to see a junior prove me wrong!

Tags: Olympic Talk