Clearly, MX outdoors and Supercross are very different types of races, and they cater to different riders' strengths and amplify some of their
weaknesses. But it's not clear just how different they are. I don't think anyone would say that EVERY rider's Supercross results would give
you a 1-to-1 prediction of his Motocross results, nor would anyone say that the Supercross races have no relevance to predicting how a rider will do in Motocross.
Supercross results predicting upcoming Motocross?
The main reason I ask this question is for predicting outdoor MX results -- can (or should) we use SX data from the prior season as part of our prediction for the upcoming MX season?
So, let's look at how a rider's results in Supercross compare to their MX outdoor results in the same year. In the graph below, each dot represents one year for one rider, and it includes all riders who had results in both SX and MX in the same year (from 2000 through 2015) in the Top 30 of both. The rider's average SX result is on the x-axis and that same rider's MX result for that same year is on they y-axis:
If a rider's dot is right on the line, then the trendline shows the relationship for all of the riders in these years, and it generally makes sense and has an R^2 value of 0.45. If we trust this model, then that means that Supercross results for a rider can explain or predict 45% of the variability in the MX results data. You might say that whatever factors cause a rider to be good at riding a Supercross course (speed/skills in general, but also ability to deal with big jumps/whoops, etc.) also help determine nearly half of their outdoor Motocross success, with the remaining factors being something else (injury, luck, speed/skills specific to MX rather than SX).
The points are generally grouped around that trendline, but with plenty of exceptions -- if the point is above/left of the line, the rider did better in SX than MX, and if the point is below/right of the line, the rider did better in MX than in SX; some of the larger exceptions have the names/results called out. Without looking to deeply into it, I highlighted three riders involved in these major deviations:
Josh Grant (orange) and Clark Stiles (green) actually had most of their years in the bottom/left area, meaning they actually had consistent results in MX that were better than their SX results. Martin Davalos (pink), on the other hand--despite his one odd season in the very top left--had results that were generally around the trendline otherwise. The same is true for Larsen (blue) and Plotts (purple), as their one odd result was much different than their others.
Based on these graphs, though, since there is a relationship between Supercross and Motocross within the same year, it looks like we should be including SX data in the MX predictions, at least to some extent. So, when final 450 and 250 Pro predictions are published this week (updates from the prelim 450 and prelim 250), some element of 2016 Supercross results will be incorporated into the projections.