In a lot of ways it's a simple question: how much does the starting spot* matter for determining where the rider finishes? But, it's an important one, and one where correlation doesn't necessarily mean causation.
Starting Spot vs Finishing Position
*The riders' position "after the holeshot", "coming through the first turn", "at the holeshot marker (?)", "after the starting section (?)", whatever you want to call it -- I'm not sure what the best way to refer to it is. This question (and the terminology), get brought up in comments on the Atlanta recap. It's an awkward thing to describe -- terminology suggestions welcome.
There are many ways to visualize it, so here are three graphs with different ways of comparing the "Starting Spot" with the "Finishing Position". Please take into account that these are only for 450 Supercross in 2016 (because the data for previous races is very hard to get. Maybe someday.), and only Main Events. We'll end with the most boring chart (in my opinion), since that one shows the R^2 and leads to the next topic.
2016 450SX Starting Spot vs Finishing Position, heat chart:
Here we see the effect of the relatively small data set, in that a few starting spots vary wildly from the others -- 4th did much
2016 450SX Starting Spot vs Finishing Position, by color and distance:
2016 450SX Starting Spot vs Finishing Position, trendline:
According to that R^2, where the riders emerge from the first turn explains about 74% of the variance in terms of where they finish the race. But to say that the starting spot "determines" the finishing position would be jumping to a conclusion. For one thing, the riders that get the better starting spots are often the best riders, so the real determining factor is the riders' skill level -- them getting the better position out of the first turn is just a by-product of their superior skill.
So, this isn't really news: getting a better spot in the starting section of the race correlates with how well the rider does overall in the race. Although, seeing that the starting position correlates so highly with the end result is noteworthy -- off the top of my head, I probably would have figured that starting and ending positions (over just 8 races, a fairly small sample) would have had an R^2 of 50% or less, so the 74% indicates there's a stronger correlation than I would have expected.
(Rocky Mountain ATV/MX actually looked at a similar topic a few months ago, I noticed, though they focused on the one rider who got THE holeshot. But very interesting nonetheless.)
In addition to the correlation, we have some more to look at. How does this relationship map out for each rider? First, we can look at where each rider started and ended each race (in 2016). First the top half, according to average finish position:
Now the bottom half:
And finally, we can look at each rider on average over their races in 2016, which shows how well (or poorly) they started and how well (or poorly) they finished, with the slope of the line indicating how much they increased or decreased by the end of the race. Including only riders with 3 or more races in 2016:
I think we see about what we expect -- the riders we think of as faster have upward slopes; slower riders, downward. For instance Cole Seely and Chad Reed have gotten pretty good starts and slightly better finishes. Jason Anderson has had not great starts but good finishes. Mike Alessi, Justin Bogle, and Phil Nicoletti, to name a few, have had not great starts but worse finishes. One surprise is Davi Millsaps, who's slope is slightly negative, meaning on average he's lost ground through the races: in both Anaheim 2 and Glendale he got the holeshot, then finished 7th and 5th, respectively.
So, some interesting graphs, which summarize how meaningful it is to get a good position when getting through the very first section of the race. As the season goes on, we can update this with more data, and possibly look at the Main Event vs the Heat and Semi races.