http://www.crossresults.com has Nats predictions up also
Countdown to Nationals – Predicting the Elite Men’s Winner… With Science!
This is part of an ongoing series of national championships-related articles to get you ready for KC as a racer or spectator. You can view the rest of the series articles here.
by Jake Sisson
In an effort to most accurately predict who is going to win the shootout that will be the 2008 US Cyclocross National Championship, I turned to science. Late last night, I trotted out into the woods, grabbed a very scientific looking stick, snapped it into 6 relatively equal pieces (representing the 6 guys I felt had the best shot at winning nationals), tossed them in the air, and measured the order in which they fell towards true north. The closest twig-piece to north was to be the winner, and so on. Isn’t that science just amazing. I then promptly forgot which twig represented which rider, and decided to trust statistics instead.
If numbers make your head hurt, now would probably be the time to go out and invent your own prediction method. If they don’t, this might actually interest you somewhat. My past exploits with Fantasy Cyclocross have proven that I’m into proving analytically who is the best/worst/luckiest. So, following that trend, I set out to measure, predict and project this Sunday’s marquee races. So let’s quick review the selection process that I chose to measure the winner. First, I picked the six guys I feel have the best shot of trotting home with the title, in no particular order: Jeremy Powers (Cannondale/Cyclocrossworld.com), Jonathan Page (Planet Bike), Ryan Trebon (Kona), Todd Wells (GT Bicycles), Jesse Anthony (Jamis) and Tim Johnson (Cannondale/Cyclocrossworld). I realize that that leaves out a number of viable candidates, but when the dust settled, under the conditions I set up, none of them would have had a legitimate shot at the top. I then looked at their performances in the 20 most important races of the season. These races included all C1 races, all races included in the NACT and the USGP series. These races provided the most overlap of the top six racers.
So what metrics did we look at? First up on the list was the individual’s results in the last five races. Obviously, as with any data, a sample size that contains only 5 data points cannot really give you solid correlations, so I took a look at two other sets of data to obtain trends: performance in the last 10 races, and performances across the season. I then looked a bit more specifically, and took a look at the individuals performance in Sunday races. Finally, to match the riders up against each other, I took a look at the performances of each rider against one another. To do so, I assigned a +/- system to the finishes. If rider A beat rider B by 2 places, he was awarded a +2. If rider B beat rider A by 2 places, rider A was awarded a -2. These were then summed and averaged, to see each man’s performance against the other. Trends in performance were established graphically. The Y-axis intercept was taken as a reference point for predicted finish at Nationals. A negative slope is favorable, as is a low intercept value: y=(Slope) x + Intercept.
The reigning National Champion has been the most effective rider in the United States this year, and by a surprisingly large margin. Johnson has the best 5 race, 10 race and season average finishes of the bunch -
1.6, 2.1, and 2.1 respectively. He also has the best cumulative +/- score of the top six, a +51. The only rider he has not thoroughly dominated, however, is Ryan Trebon. I alluded to this in my weekend preview article this weekend, but to put it more explicitly, Johnson and Trebon are exactly tied on +/-. For every place Johnson has beaten Trebon by, Trebon has beaten Johnson by the exact same margin. While Trebon has won more on the season, Johnson has been the model of consistency. His worst performance in the study was a sixth place at Day 1 of the Derby City Cup. If you look at recent trends, Johnson’s 5 race trend is not exactly the best. His 3rd at Day 2 of the Portland Cup threw a wrench into his dominating set of trends. His trend in the last 5 races puts him at 2.1, while his trend from 10 races puts him at 1.222 and his season trend puts him at 1.8. On Sundays, Tim’s average finish is an astonishing 1.6, but with a trend in the wrong direction that puts him at 2.2, another worrying trend for the reigning Champion.
Ryan Trebon’s seven major wins on the season have been unmatched by anyone this season in the United
states. That’s an impressive 35% of the measured races. Trebon has also been the numerical equal to Johnson, who, it was argued, has been the hottest rider on the scene this year. Trebon has also been almost as dominating over his rivals as Johnson, with a +48 score versus the top six. Trebon also has favorable trends in the last ten races, and on Sundays. However, numerically, Trebon’s season has not been as strong as Johnson’s and doesn’t project to wrap up as well as Johnson’s. Trebon’s average finish in his last 5 races, 10 races, and across his season are 2.8, 3.2 and 2.5 respectively. His projections are also not quite as good, with only his 5 race projection (.4) better than Johnson’s. His others are 1.6 in his last 10 races and 3.1 across the season. On Sundays, Trebon averages a finish of 3.6 with a trend projection of 3.43. Trebon does, however have the best 5 race projection of any contender, which is a big boost to his National’s case.
Todd Wells’ measurements have been skewed a bit because of the slow start to his season. Illness slowed him down in many of the first few weekends’ racing, and he skipped a few of the early weekends, no doubt
feeling the strain of an Olympic campaign. Recently, however, Wells has made his case for being the best rider in the United States. Wells’ average finish in the last 5 and 10 races – 2.4 and 2.6 – is better than Trebon’s in both categories. Wells has certainly improved from his slow start, having the highest projection of everyone if you look across his season. He projects to 0.3. Before you start yelling about “how can you project to less than 1?!” remember these are arbitrary and we’ll compare them later to make a final ruling. Lately, however, Wells’ metrics are far more pedestrian, projecting to 3.6 across the last 5 races and 2.1 across the last 10. Good, but not great. Why is Wells a top contender you ask? Wells is easily the most dominant on Sundays, projecting to a -0.04! Granted, he’s being influenced by poor results in the early season, making his improvement drastic, and that’s what the trends are measuring. But if you’re trying to improve your form for a specific point in the season, that’s how you go about it, right?
Jeremy Powers has taken home a number of impressive wins this season, but the data doesn’t favor Powers escaping with the victory at Nationals this year. Powers’ trend in his last five races has been positive and
does not project well (3.8). Neither does his 10 race projection (3.3), and while his season projection is better, it is not that much better (2.7). His average finishes have also hovered around the 3.0 mark for the entire season. Sundays have been somewhat kind to Powers, with his Sunday average finish his strongest in the data, with an average finish of 2.9, but his performance of late has not boded so well, with a projection score of a dismal 4.2. Powers does, however, own a better +/- rating than Todd Wells (-24 to Wells’ -36) but Wells’ poor early season is mostly to blame for that figure. Powers has not been dominated by any one person, however, and his season ratings against Trebon (-9) is one strong point in his case to be National Champion.
Of the six men who could win the National Championships, Anthony’s resume is probably the weakest. Anthony has no wins to his name this season, and his best performance is a handful of second places.
Anthony’s average finishes are good, but not exactly top caliber, with a 5 race average of 4.8 and ten race average of 4.2. Anthony does not have more than 10 registered results from this season, due to his breaking his wrist and missing a good deal of the early season. Anthony’s 5 race projection, however, is positive, but only projects to 4.2. His ten race and season performances are not as promising, however, with his 10 race projection only equating to a 4.6. Sundays have, however, been quite kind to Anthony, as his average finish on Sunday is a 4.5 and he projects out to a 2.75. Anthony has only finished four Sunday races on the season so his data isn’t the most concrete. Anthony has a fairly bad +/- score of -47. Anthony is certainly the man to beat for many years to come, but this year will not be his year.
Jonathan Page presents a quandary. Page has only one result against America’s top competition, which means there’s really no basis to make comparisons relevant to this discussion. I could try to pull out a metric from his results in Europe, but the competition is worlds different from what the competition will be here in America, and that’s no real use to us. At Cross Vegas, Page came in fourth, in his only race against the top level of American ‘cross racing, but that was in September, and so many things have changed since then. So Page will come in as something of a wild card, and he could be the one to take this year’s National Championship based on his track record and his experience.
So let’s take stock of everything presented here:
In order of 5 race Projection: Trebon (6), Johnson (5), Wells (4), Powers (3), Anthony (2)
In order of 10 race Projection: Johnson (6), Trebon (5), Wells (4), Powers (3), Anthony (2)
In order of Season Projection: Wells (6), Johnson (5), Powers (4), Trebon (3), Anthony (2)
In order of Sunday Projection: Wells (6), Johnson (5), Anthony (4), Trebon (3), Powers (2)
Johnson – 21
Wells – 20
Trebon – 17
Powers – 12
Anthony – 10
Of the Americans, Tim Johnson has a slight advantage over Todd Wells as favorite for the National Championship. Will Johnson take home the title, or will Wells be able to out-duel the incumbent? Will Trebon be able to prove all the data wrong and take home his second title in the last three years? Take a look at the data and decide. Either way you decide to choose, we’ll know better on Sunday.
Have you subscribed yet? You're missing out if not. Get all-original content and your cyclocross fix throughout the year with a subscription and Issue 23 back copy, with features on Lars van der Haar, Jonathan Page, Elle Anderson and more!