RotoWire Partners

According to the Data: Checking in On Preseason Consistency Correlations

Jonathan Bales

Jonathan Bales

Jonathan Bales is the author of the Fantasy Football for Smart People book series. In addition to RotoWire, Jonathan also provides content to the New York Times, Dallas Morning News, DallasCowboys.com, and NBC.

Checking in On Preseason Consistency Correlations

My first ever "According to the Data" column here at RotoWire was called Consistency, Volatility, and Predictability: Why You Should Draft Aaron Rodgers Ahead of Calvin Johnson. In that article, I discussed why understanding the consistency of each position on a year-to-year basis is crucial for fantasy owners. Even if you knew the top kicker would score 1,000 points this year, that knowledge would have been useless without an understanding of positional consistency, and thus predictability. That is, if the correlation between kickers' fantasy points in Year X and Year X+1 is zero, there's no reason to draft one before the last round, regardless of your projections.

One of the graphs I used in that inaugural post was a calculation of just how consistent each position is from season to season (below). I didn't include defenses or kickers, both of which have near-zero consistency.



Using the consistency correlations as a guide, I performed a simple regression of stats for each position and projected the final ranks for 2012. The average projected movement for the quarterback, running back and wide receiver positions (below) matched up remarkably well with past data.



Today, I want to compare the projections to the results after 10 weeks of the 2012. I've collected 2012 ADP information for each position, comparing it to the current rankings (both are point-per-reception). In calculating the results, note that I've removed the largest outliers from each sample. Thus, when calculating the average movement for the top five or 10 players at a position, I excluded the two players ranked highest and lowest. For the sake of simplicity, I included only the top five and top 10 players at each position.

Top 5 Quarterbacks
Projected Movement: 1.0
Actual Movement: 2.3
Difference: 1.3
Top 10 Quarterbacks
Projected Movement: 2.7
Actual Movement: 3.6
Difference: 0.9
Top 5 Running Backs
Projected Movement: 1.4
Actual Movement: 4.0
Difference: 2.6
Top 10 Running Backs
Projected Movement: 4.1
Actual Movement: 7.0
Difference: 2.9
Top 5 Wide Receivers
Projected Movement: 5.6
Actual Movement: 9.0
Difference: 3.4
Top 10 Wide Receivers
Projected Movement: 4.2
Actual Movement: 5.0
Difference: 0.8
Top 5 Tight Ends
Projected Movement: 2.0
Actual Movement: 7.0
Difference: 5.0
Top 10 Tight Ends
Projected Movement: 3.8
Actual Movement: 6.1
Difference: 2.3
Top 5 Kickers
Projected Movement: 13.5
Actual Movement: 11.0
Difference: 2.5
Top 10 Kickers
Projected Movement: 10.0
Actual Movement: 10.4
Difference: 0.4
Top 5 Defenses
Projected Movement: 12.8
Actual Movement: 10.7
Difference: 2.1
Top 10 Defenses
Projected Movement: 9.8
Actual Movement: 7.4
Difference: 2.4

For the most part, the preseason consistency correlations and corresponding movement projections have held up. The top quarterbacks, as expected, have been relatively consistent. The top three quarterbacks in preseason ADP - Aaron Rodgers, Tom Brady, and Drew Brees - all rank in the top six in the current quarterback rankings, with Brees and Rodgers leading the pack.

The consistency system was off a bit on elite running backs, projecting relatively little movement but seeing an average jump of four spots in the rankings. That's due primarily to Darren McFadden's poor season, though Chris Johnson has also failed to deliver on his preseason ADP (fourth among all running backs). Still, Arian Foster, Ray Rice and LeSean McCoy all rank in the top five in PPR formats, so the overall consistency isn't terrible.

Perhaps the most accurate of my projections came at wide receiver, particularly because I predicted that the top-10 receivers as a whole would be more consistent than the top-five receivers alone. Wide receiver fantasy points have a strength of correlation of just 0.40 from year to year - the lowest of any of the four main positions - and elite wide receivers have been horribly inconsistent over the years, hence my low ranking of Calvin Johnson on overall boards.

As it stands right now, the top five receivers in preseason ADP are ranked an average of 9.0 spots away from where they were drafted, highlighted by Larry Fitzgerald (16th) and Andre Johnson (28th). Meanwhile, the receivers that were drafted sixth through 10th - the ones I told you to jump on in the preseason - have moved only 5.0 spots in the rankings, on average. Four of those five players - A.J. Green, Victor Cruz, Roddy White and Brandon Marshall - are ranked in the top eight among all receivers.

My consistency projection system has performed the worst on tight ends, due in large part to Aaron Hernandez's injury and the decline of Antonio Gates. Prior to this year, elite tight ends were extremely consistent from season to season, which is why I was so high on Jimmy Graham and Rob Gronkowski. Those two are still both ranked in the top three among tight ends, but the rest of the top five has slipped.

Finally, the projections for both kickers and defenses were relatively easy. Kickers have zero predictability, and the correlation between points for a defense in one year to the next is just above zero. Thus, the major movement we've seen from ADP to current rankings is unsurprising.

Jonathan Bales is the author of Fantasy Football for Smart People: How to Dominate Your Draft. He also runs the "Running the Numbers" blog at DallasCowboys.com and writes for the New York Times.