Analysis: Performance and Age

FasterSkierMarch 30, 20107

The Italian legend Maurilio De Zolt and  Norwegian great Hilde Pedersen are excellent examples of elite cross country ski racers who continued to excel at the international level into their 40’s.  While continuing to compete on the World Cup circuit at that age may not be common, cross country skiing is known for its athletes developing later than in other sports.  Top skiers often continue to be competitive through their late 30’s.

Hilde Pedersen winning the 2009 Marcialonga.

In light of these observations, it is natural to ask what precisely the relationship is between performance and age among elite ski racers.  To do this, we will focus on major international events (World Cup (WC), Olympic (OWG) and World Championship (WSC)) and restrict our attention to seasons where an athlete participated in a “significant proportion” of the available events.  In this case, we’ll be fairly lax and include any season by an athlete where they competed in at least five such events.  (Elite athletes at the extreme ends of the age spectrum might not be on the World Cup circuit full time, and we’d like to include these people.)

The graph below shows the results for distance events, broken down by gender.  Each point represents the average FIS points for a season by an athlete of the given age at the start of the season. (Ages are somewhat approximate, as I only have access to the year of birth.  This means the actual ages will vary by around +/- 6 months.)  Additionally, I’ve added a trend line to each panel.

Distance performance by age.

Several things immediately stand out.  First, there is the general downward trend over time.  Skiers improve over time!  Shocking!  Actually, part of what is going on here is selection bias.  If you’re over the age of 35 and still skiing WCs or OWG races, chances are you’ve proven yourself as an elite competitor.  So at the far right of these plots, we’re primarily seeing the effects of other, slower skiers retiring.  If we allowed these slower skiers to continue racing into their late 30’s it’s likely that both trend lines would start to slope up slightly by age 40.

More interesting are the differences between the panels representing men and women.  For men, we see steady improvement until around age 30 and then things begin to plateau.  For women, the improvement early (<24) seems slightly slower than for the men, and then suddenly drops down between the ages ~24-28.  From there the downward trend continues, but more slowly.

Additionally, there appears to be less of an attrition effect for women.  Notice how the pattern of points in the men’s panel steadily narrow like a funnel as we move right, while in the women’s panel even at age 35 there’s still a wide range of performance levels.  It seems that if you’re a man and still racing World Cups past the age of 35, it’s very likely that you are very, very fast, while if you are a woman, it may be possible to continue racing past the age of 35 without being quite such a dominant skier.  Why would this be the case?  I’m not sure.  Perhaps come commenters will enlighten me!

Let’s move on to sprints.  Using data on sprint racing is more challenging, as we have to contend with three different “results”: FIS points, qualification ranking and final ranking.  Personally, I’m not particularly enamored with FIS points as a way to compare sprint performances and so I’m going to stick with the final round ranking list for each event.  (So if you won the sprint race, you’d be assigned the value 1 and so on.  I may investigate alternate ranking procedures in future articles.)

Repeating the procedure from above, we get the following graph:

Sprint performance by age

Actually, I’ve changed things slightly.  I’ve included seasons with as few as four sprint races rather than five.  That increases the number of data points we have to work with, which is helpful since sprinting hasn’t been around nearly as long and so the data here are more limited.

First we should note that the five points in the lower right in the women’s panel are certainly having a disproportionately large effect on the trend line.  Without these points, the trend line would likely remain mostly flat through the late 30’s.  Also, I’m not sure how significant the slight upward trend in the men’s panel is.  Again, there isn’t a lot of data there, so I’d read that as “mostly flat” rather than “slightly up”.

Obviously, things look quite different here.  Gone is the fairly obvious improvement over time (allowing for the outsized impact of the five data points noted above in the women’s panel).  Overall, the trend is much flatter.  I wouldn’t read much into this, since we haven’t had a lot of seasons yet to evaluate the performance trends of sprinters over time.  But it wouldn’t surprise me if sprinting displayed less of a “steady improvement over time” trend that we saw with the distance races, since the conventional wisdom is that sprinting is a young person’s event.  But that’s just speculation…

One similarity, for the women, is the trend from ages 24-28.  As with the distance events, that seems to be where women make their largest improvements.

I don’t have a great ending for all this, so instead I’ll just invite you to discuss and comment away…and I’ll close with a trivia question: Can you name the (retired) skier with a FIS results record whose entire name consists of a single letter?

FasterSkier

Loading Facebook Comments ...

7 comments

  • Ian Nesbitt

    March 30, 2010 at 10:43 am

    This is awesome. Is it just me or do these graphs suggest that skiers compete in more events when they are younger? Also, who is portrayed in these graphs?

  • Dakota Blackhorse-von Jess

    March 30, 2010 at 1:00 pm

    Can you filter the dataset so that only athletes that are still competing at some reasonable later stage (say 30 or 32) are represented? It might be more telling to look at that average and remove the selection bias, although obviously I have no idea just what an effect, if any, that will have on your results and if they’ll still have statistical significance (ie just how many data points are you working with?). Does make me curious if the same trend exists though. That might eliminate your stars that burn bright and burn out early, which while interesting, may also be skewing your trend lines.

    On a different note – is just world cup / olympic racing a good limiting factor? Certainly it is important, but it would be interesting (if you’re digging deeper) to look at the fis points, per year (still based on age in the given year) over the course of an athlete’s career. You’d have a lot more data, but it might be telling to watch the average trend of careers for skiers that made it to the world cup level. Possible cases include junior stars who race world cups young and flame out (or succeed), or slow starters who improve a lot later in life, and when those improvements occurred, for example. Still limit the dataset to those skiers who raced world cups, olympics, or world championships, but in your dataset have each column be a year of age just like you have, and each row be an athlete so you have a complete picture of their athletic history in FIS points.

    Obviously it’s still only a rough look, but if it shows something really interesting, who knows. Might be worth it. Ignoring of course the impacts of the nuances of point awarding (ie racing at a high level in good FIS point field OPA cup races may give good FIS points, and then moving to the World Cup and racing okay for a few years may negatively impact their FIS points). But who knows how significant those impacts are?

    On a final note – a fun side project might be to break down the techniques and distances if you have access to that kind of information and look at those graphs separately and see if you notice anything cool. For instance, my personal favorite racer from when I first started watching World Cups was Jens Arne Svartadel. He was a kick ass sprinter. He was at the Vancouver Olympics… racing the 50k. While it’s common knowledge that athletes get better at longer distances as they get older, I wonder if there is a *significant* impact on the pure distance results (is there a general trend towards more success in the 50k as you get older, is there a drop off in success in the 10ks). Also, as I mentioned above – breaking it down by technique might be cool. Do folks get better at classic skiing as they get older and worse at skating? Or the other way around? (Although who knows… your “specialists” might skew your data)

    Just some thoughts – if you dig into it and find anything significant, it’d be cool to see!

  • Cloxxki

    March 30, 2010 at 2:35 pm

    Excellent article!

    Perhaps tradition comes into play for olde athletes. The men may need to accept that they’ll make more money in other jobs than being an athlete. The women may have spouses that already make a good living. This might explain that only the superpro’s among the men stick around. If so, it’s a shame, world cups could be even harder contested.
    A single lady racer at 36 years, attacking the circuit with 100% determination, that might be rarer than a guy the same age, in the same position. Women can also return from a baby break stronger than they were before. I’ve seen that in several other sports at least.

    If one would compare ski speed relative to the winner, 40 year olds like Hanevold in biathlon, barely lost any speed at all. He was never the fastest on the tour, but can still hold his own, and do what is asked from him as part of the 4 best biathletes, in the strongest biathlon nation.

    Another interesting thing would be to know how old the “geezers” were when they entered the WC circuit the first time, or better : how many seasons they actually contested.
    There’s a saying “what comes quick, goes quick. Some young racers just burn up before they have time to mature.

    Not to be put into numbers, but what is the consensus : do racers get better technique over 20 years on the WC, to the point that they make up lost speed with it, or is a 20 y/o worldcupper already supposed to be the absolute pinnacle of form?

  • 2PACmosDEF

    March 30, 2010 at 5:10 pm

    I don’t really buy this article. The best distance skiers in the world -with the exception of the late-developing Italian athletes- are all really young, like 24 or younger (Northug, Cologna, Hellner, Manificat etc). You are looking at way too much data in you article. You can’t look at every body on World Cup, that’s kind of stupid, look at the top competitors, and you’ll find that skiers start going downhill at the age of 25.

  • JimGalanes

    March 30, 2010 at 5:47 pm

    Joran-Thanks for your effort on this. I know a study like this is a lot of work. It is good to see some critical thought on performance evalaution.

    I am not sure the data as presented can tell us much more than skiers enter the world cup, some improve to a high level, some don’t and eventually wash out and some retire or quit. All resulting in a downward shift in your trendline becuase only the faster skiers keep skiing as they get older.

    I think it would be valuable if the data is in a form where you could do some basic comparisons. For example group all skiers who finish in the top 10, 20, and 30 in the world cup at the end of this season. Then chart their history over all the years they have been competing on the world cup. This would be telling data in that it would show us at what age, and points level an athlete entered the World Cup level of skiing and how they progressed over years. It would also demonstrate if anyone started on the world cup with very high points ever developed to the level of top 10-20 finishes.

    Good Job

  • JoranElias

    March 31, 2010 at 7:33 am

    Thanks for all the excellent comments! I’m on vacation overseas at the moment so I don’t really have time to respond fully. However, I will say that I agree that what Ive done here is the very simplest (and possibly least informative) approach to the question of performance vs age. I’ve been trying to keep these articles very simple so as to appeal to a large audience and avoid getting too technical.

    I am able to look at the things you’ve all suggested, and plan to in the weeks/months to come. So I hope you can overlook the oversimplified natureof these first few articles (maybe one morein a week or two)…I’ll get more sophisticated down the road.

  • JoranElias

    April 6, 2010 at 3:33 pm

    I’m back in the States and have access to my computer again…I don’t know if anyone is still checking these comments, but I thought I’d provide some more detailed responses in case you are.

    @IanNesbitt

    I think you may be confusing what the y axis here represents. There’s nothing in these graphs that represents the number of races per athlete. The increased variation toward the left, at younger ages, reflects the fact that the results by younger, less experienced racers are more varied: some of them are really fast, others are really slow. By the time people are over 30 and racing on the WC, they’re probably proven athletes to some degree. So the only people left competing at that age are the really fast people.

    @Cloxxki and @JimGalanes

    Looking at performance within athlete tiers is definitely on my to-do list. It’s been bouncing around in my head for a while, but there are some tricky aspects to getting the right subset of data, such that I’m not really satisfied that I have a good strategy for it yet.

    Performance over time by individual athlete is easy to do, but hard to put in a single graph with enough individuals to be meaningful. But having done some of that, I can tell you that (unsurprisingly) there’s a lot of variation in the “development pattern” of WC athletes, but some systemic biases, too. You rarely see an individual athlete get gradually slower over time, simply because they’re likely to retire or be pulled from the WC circuit before that really happens. So you see lots of improvement and plateauing.

    @2PACmosDEF

    I wasn’t trying to say that the best athletes are 30+ (at least I don’t think I said that). Rather, I started from the idea that xc skiing has a reputation (along with other endurance sports, see Lance Armstrong) of seeing higher performances from relatively older athletes. So my claim is really that it is _possible_ for 30+ yro’s to excel on the WC, not that they’re automatically the best.

    I agree that many of the top men at the moment (but not all, eg Vincent Vittoz) are under 25. But I don’t think this has necessarily always been the case. Daehlie’s back injury that more or less ended his career happened at age 30, and it’s hard to say that he’d been declining for the previous 5 years, or that he wouldn’t have continued to be successful for another five.

    @Dakota

    As I mentioned above, everything you describe is technically possible, as I’ve basically replicated the entire FIS results database (not entirely, but pretty close). And those sorts of athlete development questions are certainly on my queue. Obviously, they get more complicated, though, and I’ve had to balance the database building+cleaning vs analysis time commitments. In particular, I like the idea of looking at whether older athletes are more successful at longer races. I suspect any difference you’d see would be small, but it would be interesting if it existed at all…

    As to the question of looking only at WC/OWG/WSC races, my strongest argument for that really is that those are the races where they, by definition (kind of), have a penalty of zero. I have some pretty deep suspicions of whether FIS points are really such a great way to compare performances between different races. Looking at only those races with a penalty of 0 makes for the “safest” comparisons. So I’m just being very conservative, that’s all.

    @JimGalanes version 2:

    The question of whether athletes who start on the WC with relatively high points ever develop into top athletes is a good one. I’ll have to think about a systematic way to approach it for aggregate data, but as a quick example, Beckie Scott did it, so it’s at least _possible_. Her average distance points (in big races) dropped from over 100 to around 10 over the course of her career. Obviously, though, that’s just one skier.

    Anyway, thanks again for all the comments!

Leave a Reply