Let’s talk about October traditions. Halloween decorations start going up around town. Suburban families start stockpiling fun-sized halloween candy, (I think the “fun” part is supposed to be seeing how many you can justify eating in one sitting). Businesses vie with each other to see what new and outlandish things they can apply the “pumpkin spice” label to.
(As an aside: am I the only one who finds it ominous how the “spice” part of that contrivance always remains so ambiguous? I’d never eat at a restaurant whose menu included items such as “seasoned meat”, “sweetened beverage”, or “prepared food”…)
One other October tradition? That’s right, it’s “Revisiting Preseason Expectations” week over here in Fantasy, in Theoryland!
Why are we revisiting preseason expectations?
After week four in October of 2013, I wanted to look at how much information early-season results really provided us. I was surprised to discover that performance through four weeks actually correlated with performance the rest of the way worse than preseason ADP did. In other words, the first quarter of the season tells us less, in aggregate, than we already knew during the offseason.
This is a surprising finding, and like any surprising finding, my first instinct is to try to disprove it. (Only once a surprising finding stands up to repeated reexaminations can we start to accept it as settled wisdom.)
I have compiled a list of the top 24 quarterbacks, 36 running backs, 48 wide receivers, and 24 tight ends according to MFL’s 2015 preseason ADP. From that list, I have removed any player who missed more than one of his team’s first four games or more than two of his team’s last twelve games. As always, we’re looking by team games, rather than by week, to account for players with a week 4 bye.
I’ve used PPR scoring for this exercise, because that was easier for me to look up with the databases I had on hand. For the remaining players, I tracked where they ranked at their position over the first four games and over the final twelve games. Finally, I’ve calculated the correlation between preseason ADP and stretch performance, as well as the correlation between early performance and stretch performance.
Here’s the data:
|Player||ADP||Early Rank||Late Rank|
The average delta between ADP and stretch performance was 6.42 spots. The correlation was 0.260.
The average delta between early performance and stretch performance was 8.00 spots. The correlation was 0.215.
|Player||ADP||Early Rank||Late Rank|
|Duke Johnson Jr||33||25||26|
The average delta between ADP and stretch performance was 19.22 spots. The correlation was 0.309.
The average delta between early performance and stretch performance was 15.37 spots. The correlation was 0.644.
|Odell Beckham Jr Jr.||3||19||3|
The average delta between ADP and stretch performance was 14.18 spots. The correlation was 0.648.
The average delta between early performance and stretch performance was 22.42 spots. The correlation was 0.632.
The average delta between ADP and stretch performance was 9.15 spots. The correlation was 0.295.
The average delta between early performance and stretch performance was 9.25 spots. The correlation was 0.559.
In 2016, preseason ADP was slightly more predictive at quarterback and wide receiver, and much less predictive at running back and tight end. Across all four positions, the correlation between preseason ADP and stretch performance was 0.548. The correlation between early performance and stretch performance was 0.659.
Rounding up all years worth of data into one handy chart, here are how things stand:
|Years||Preseason correlation||Early season correlation|
After my initial finding that preseason ADP better predicted stretch performance than early-season performance did over the 2010-2012 seasons, each of the seasons from 2013 to 2015 has given a slight edge to early-season performance in reliability. It’s possible that this represents some structural trend in the NFL. It’s also possible that data is noisy and random is random.
Either way, the correlations have been fairly close each time, close enough for me to declare that it’s not until week four that the new information we’ve gained approaches an approximate equilibrium with all of the information we had to enter the season.
One last note must be made; because I am excluding players who missed substantial time, these data are subject to selection bias. Players who lose their job due to underperformance rather than injury will not be counted, and they would surely drive the correlations for early-season performance up.
I've included all of the data used, so please feel free to peruse at your leisure and draw your own conclusions.