Right or Wrong?: 2016 450SX Pre-Season Predictions
WAY, way back in January, we published predictions of how the 450 Supercross season would end up. It was so long ago, I forgot what that format looked like -- it had quite a few colors, which is a bit much. Although I kind of like having the predicted total points at the end of the lines, but the numbers on the axis at the bottom are pretty useless and kind of an eyesore. Take a look:
The format was later changed to make it a bit more square and to show the change from week to week. But enough about what the graphs looked like, how about the predictions?
Overall, pretty good for a first try, judging by the table at the top of the page. In the Top 5, four were dead-on. On one hand, picking Ryan Dungey to win and Ken Roczen in 2nd wasn't exactly a wild and crazy forecast. But on the other hand, plenty of people saw Roczen as the favorite, and many had Eli Tomac winning it all or at least in 2nd. Having Tomac predicted 4th and Chad Reed predicted 5th were also exactly right.
At the time of the original predictions, I was thinking "Man, Dungey and Roczen sure are predicted a lot higher than the rest of the field -- I wonder if that's too extreme?" But, that turned out to be true -- possibly even understated for Dungey. Dungey was predicted to have a 43 point lead over 2nd and 82 points over 3rd. It ended up being 60 and 76. Roczen was predicted to have a 39 point lead over 3rd and 40 over 4th; that ended up being 16 and 41, and had he not DNF'd in Vegas (i.e. gave up), then that 16 would have been closer to the predicted 39.
Jason Anderson was the big surprise, from a numbers standpoint. Even though people around the sport said he looked much faster while prepping for 2016, which we mentioned in the prediction, even they wouldn't have pegged him all the way up at #3. These predictions didn't incorporate 250 SX results, and if they had, I think that Anderson would have been much higher, probably around Marvin Musquin in 7th, but still not nearly as high as 3rd. Next time around we'll try to incorporate those 250 results in some way.
Missing from the actual Top 5 was Trey Canard, who was predicted to finish 3rd but ended up 9th. That brings us to one of the issues with the predictions -- injuries. These predictions didn't take potential future injuries into account, which was noted at the time when the predictions went up. So when Canard missed 2 weeks, he was deviating from his pre-season prediction. Same goes for Justin Barcia (missed 9 weeks), James Stewart (10), Davi Millsaps (6), Blake Baggett (8), Broc Tickle (10), Andrew Short (15), Wil Hahn (8), and Dean Wilson (15), along with some others with lesser injuries. I'd love to layer in the probability of being injured for each rider's prediction, but first we have to establish that injury is a "skill", so to speak, and isn't just random. To look at how these riders did on the weeks they were riding and not injured, see the "Per-Week" section below.
Next up is Marvin Musquin, who was also predicted exactly right, in 7th. This prediction did not have much data behind it, and getting it exactly right was probably a bit on the lucky side. If he doesn't injure his wrist near the end of the season, he very well could have moved up several spots. Cole Seely, on the other hand, had plenty of data behind his prediction (9th), yet it was off by 3 spots as he finished 6th. Seely just hadn't performed very well prior to 2015, which dragged his prediction down. His 2016 season was slightly worse than 2015 on a "Per-Week" basis (see chart below), but not by as much as predicted.
Justin Brayton also finished much higher in the standings (8th) than predicted (15th), though some of that may be due to riders above him being injured -- there were 7 riders above him with significant time lost to injury. On a per-week basis he performed about as expected:
For instance, Brayton was predicted to earn 11.1 points per week, but he earned 11.8, for a difference of 0.7 points in his favor. Most riders were within a point or so of their predictions. Dungey, of course, was even better than predicted, and so was Jason Anderson. A couple others to highlight, focusing on the positives, were that Davi Millsaps and Blake Baggett had results that ranked them 10th and 12th, respectively, when the did race, even though they finished 14th and 17th overall due to injury. Broc Tickle was 14th while Wil Hahn was 16th, even though they finished 19th and 20th with their injuries. Another thing to mention is that when the riders miss a lot of time because of injury, any one race with a particularly bad result--such as the crash that injured them--usually has a disproportionate effect on their per-week results.
Last, we'll look at those who the predictions missed entirely. Justin Bogle had no history at 450SX, and was left off even though he managed to finished 11th overall. Mike Alessi would have been predicted at 29th, yet he surprised many by finishing 12th. Christophe Pourcel had no history at 450SX, though he finished 15th. And Vince Friese was projected all the way down at 44th, but finished 16th.
So, there we have it. The accuracy of the full-season predictions is in the books for Supercross 2016 -- season one at MotoXGraphs! Predictions for the coming outdoor season will be posted soon, and hopefully prove to be even more accurate than the ones for Supercross, and they also should involve all Pro classes and the upper amateur classes. Stay tuned.