Statistical Skier

As promised from last time, a quick look at some similar data broken down by race length. I’ve filtered out the occasional “strange” race length, to focus on just the standard 5, 10, 15, 30 and 50 kilometer distances. The metric is the same: the median number of skiers within a fixed percent of the winner. However, we should be very aware in this case that the median is really nearly the raw data itself. How many 30k mass start races are there per season, anyway? It would be weird to have more than four, and you could only have two.

Here’s the graph for only the interval start races (click for full version):

Notice that some interval start race lengths either didn’t occur in a particular season, or were phased out altogether. So the lines may behave strangely, or simply end. Also note that I omitted the smaller percent back values, as they are less useful for interval start races.

The apparent sudden increase in competitiveness of the women’s interval start 30k races, before they were ended completely, is entirely due to a single (and only) race in 2007-2008. So really, those numbers had been holding steady for quite some time. The men’s data displays its own volatility for the same reasons (single races of a particular length in a given season) but the numbers are all generally higher than for the women.

If you figure the men were seeing ~15 people within 3% of the winner, while the women have been seeing ~5 people within 3% of the winner, that’s a pretty major difference in practical terms. Again, think of it as a game of musical chairs. When the music stops for the men, there can be as many as three times as many skiers vying for a place to sit!

As I laid out in my previous post, I think this difference is mostly a numbers game (ha!). Namely, while 3-4 decades seems like a long time for women’s skiing to have developed, it really isn’t, at least compared to the cultural and practical infrastructure built up around men’s athletics over previous centuries. So if you don’t like this state of affairs, send some money to Fast And Female, be patient, and check back in 20 years or so. I’ll guarantee you things will be better.

Finally, here’s the same graph but for mass start (and pursuit) races: Continue reading ›

On various occasions I have looked at FIS point race penalties, in a very general, roundabout way.  Most recently, I looked at whether the FIS points you get in a non-World Cup race accurately reflect what you would have scored if you had raced in a World Cup race.

The answer appeared to be: kind of.  At least for the OPA Cup and Scandinavian Cup circuits.  Specifically, we do see roughly the sort of relationship we’d hope to, though there is considerable variation from skier to skier.  This is an important example of the difference between something being true on average, and being true absolutely in every case.

One interesting extension of that post is to look at top North American races and see if we find a similar relationship.  The methodology is exactly the same, only the races I’ve used have changed.  I looked for skiers who did at least one “minor” race (Super Tour, Nor-Am, US/CAN Nationals, Continental Cups) and at least one “major” race (World Cup, Olympics, World Championships) in a single season.

This leaves us with an even smaller collection of athlete/season data points than before: generally only 5-15 skiers per season.  We also have the same problem of small numbers of races.  A skier may have done a bunch of domestic racing and then only one World Cup in a season.  So all the same caveats apply.

Here’s the graph:

As you can see, there are many fewer data points here.  However, the end result is basically the same.  The overall correlation between median FIS points has, if anything, improved over the OPA/SC cup case.  However, the high degree of variability has remained as well.