Thurman Thomas:
- 1984: 205 carries for 4.1 ypc
- 1985: 327 carries for 5.1 ypc
- 1986: 173 carries for 4.3 ypc
- 1987: 251 carries for 6.4 ypc
So... Per your logic:
205 carries = 4.1 ypc
956 carries = 5.1 ypc
Wait... What?
Or any way in between:
205 carries = 4.1 ypc
532 carries = 4.7 ypc
705 carries = 4.6 ypc
956 carries = 5.1 ypc
I get your point. If you cherry-pick a certain player's highest ypc season, their overall mean will be lower than their highest individual season. But if you assume that the historical accident that Ron Dayne's freshman campaign was his highest ypc season (although not highest overall carries per season nor his highest number of carries per game) it makes it simple to assume that all future seasons would be lower.
But then you look at Thurman Thomas and his highest number of carries per season (327) was also his second-highest ypc average, and that his second-highest number of carries per season was his highest ypc average. If I followed your logic and saw the first three seasons, I'd assume that his sophomore campaign at 5.1 ypc was the outlier, and that his junior campaign was reversion to the mean. If he had left college for the NFL after his junior year, none of us would have known he was due to explode with a 6.4 ypc campaign as a senior.
Yet his senior campaign was SO good that one year brought his overall mean from 4.6 ypc up to 5.1 ypc, equal to his BEST individual season previously.
You don't know which season is the outlier, so you can't assume that future seasons not played are lower ypc than even good previous seasons.
Here, look at it like this, because it applies to Thomas. Maybe you'll see this as a different point, but it's all related:
with 4 years worth of carries, we get a great idea of the player's personal mean, right? Dayne's was 5.8, Thomas' was 5.1, and Sanders' was somewhere below 7.6. We can be much more confident in Dayne's and Thomas' because of the volume of carries. We cannot be confident in Sanders' because of the relatively few carries. I THINK we can agree on all that.
.
Now with that luxury of having confidence in a player's true mean, we can make predictions season-by-season.
Thomas started out with 4.1 ypc his FR season, on a good amount of carries (205). Looking back, that was low for him, so we'd expect that to increase his 2nd year. Whether it jumps up TOWARDS 5.1 or past it, we can't really say, but we can be confident it would jump up from 4.1, and it did - 5.0. Even with all the additional carries that year (327), since we know his true mean, we could be confident his ypc would increase from year 1.
So he jumps up to 5.0 ypc, which was predictable. But what wold we predict now? It doesn't matter much, as he tore his knee and had under 200 carries.
Skip to his 4th year, where his ypc jump to 6.4. That makes sense, because he didn't continue on to build on his 327 carries from year 2, it was scaled back to 251 carries.
.
My overall point is that in the case of Thurman Thomas, independent of coming off an injury, if he had carried the ball 300+ times, that 6.4 ypc would have dropped towards his 5.1 career ypc average. It's plain as day.
YPC averages tend to vary in a zig-zag, using the career ypc average as the baseline. YPC tends to zig-zag WITHIN the career arc of a bell curve. If a player has a high number of carries one season, he'll tend to benefit in a ypc bump the next season with fewer carries.