Fantasy football matchup analyses are, by their very nature, frustrating exercises in making sense of a random world. Two weeks ago, who could have foreseen LeGarrette Blount posting only 11 carries for 19 yards against a Carolina Panthers run defense that was the least efficient in the NFL through 12 weeks according to Football Outsiders' DVOA metric (explained further here)? Similarly, how is it possible that Calvin Johnson could catch only three passes for 29 yards against a Minnesota Vikings pass defense that ranked 30th in DVOA?
Several readers have lamented the fact that there are so many quarterbacks in the plus-minus tables each week. With most people's fantasy playoffs in full swing, and minimal opportunity to improve rosters in any meaningful way, I figured I'd take this opportunity to explain.
In essence, what it comes down to is a combination of two things. First, quarterback scoring from week to week is much more predictable than that of other positions. Second, the defensive matchup is much more predictive of weekly quarterback scoring than that of other positions.
Regarding the predictability of scoring, what we're talking about here is that there's less week-to-week variation for quarterbacks relative to how many points they score. In layman's terms, as compared to running backs, wide receivers, and tight ends, there's much more signal than noise when it comes to fantasy quarterbacks.
This is pretty easy to show using fantasy statistics from 2011. I went ahead and looked at the weekly scoring for quarterbacks ranked seventh to 18th in fantasy points per game this year, as well as for running backs ranked 13th to 30th, wide receivers ranked 13th to 42nd and tight ends ranked seventh to 18th. I chose these cutoffs because they correspond to the caliber of player typically presented in our plus-minus tables each week.
I included only players who have played 10 or more games. Also, for the qualifying players, I counted only scoring for games in which they were active. In other words, if a wide receiver was active for a game but didn't score a point, that zero counted. However, if he was inactive, that zero-point game didn't count. The reason for both of these choices was that I wanted to make sure I measured game-to-game variability in as valid a way as possible (given the sample size).
So with the gory details out of the way, to put a number on predictability, I calculated the average (i.e., signal) and standard deviation (i.e., noise) for each player across the games he's played so far this season and then calculated his individual signal-to-noise ratio, which is just his average divided by his standard deviation.
After that, I calculated the average signal-to-noise ratio for each position. It's not important that you know the mathematical vagaries of this, but the basic idea is that a higher signal-to-noise ratio means greater predictability. For the purposes of this column, it doesn't matter whether you know what a signal-to-noise ratio of 3.0 means; just that you know 3.0 is better than 2.0.
As is clearly shown from the bar chart, weekly scoring this season has been much more predictable for quarterbacks than for the other three positions. More predictability means that our statistical system for creating the plus-minus tables has the most potential for quarterbacks.
The other part of the reason for quarterback prominence in our tables is that defensive matchups capture more signal (and less noise) for quarterbacks than for running backs, wide receivers and tight ends. In essence, our system is best able to predict weekly quarterback performance.
If we put these two ideas together (i.e., predictability and predictiveness), we end up with tables that are saturated with the position about which our system has the most confidence. We prefer being right, after all.
I hope that clears things up for those who were curious. Now let's turn to the specific player matchups that we'd like to highlight this week.
Taking a look at the plus-minus tables, this is one of those weird weeks where it seems like the worst quarterbacks have the best matchups, and vice versa. I'm going to go ahead and assume the vast majority of you aren't, for example, considering starting Jake Locker or Dan Orlovsky with your fantasy seasons on the line, so I'll instead discuss four quarterbacks who might actually enter into your lineup decision-making.
Rex Grossman (plus-10 points)
Aside from his Week 13 game against the New York Jets, the last month has seen Grossman average about 20 fantasy points per game (286 yards, two touchdowns, one interception). Not bad for a guy who Mike Shanahan thought was a worse option than John Beck.