≡ Menu

Here’s the introduction to an old fantasy football article by my fellow Footballguys staffer Maurile Tremblay:

In most fantasy football leagues, eligible players are divided into 6 different positions: quarterback, running back, wide receiver, tight end, placekicker, and special teams/defense. Imagine a league that includes a seventh position, team captain, which earns points each week based solely on the initial coin toss. For example, if you’ve got the Raiders as your starting TC and the Raiders win their coin toss, you get 30 points; if the Raiders lose their coin toss, you get nothing.

Under the current laws of probability, we can expect any particular team captain to win about 8 out of its 16 coin tosses over the course of the season, winding up with about 240 total fantasy points — so let’s use that as our VBD baseline. There will probably be one or two team captains, however, that win around 12 tosses, making them about 120 points better than average. That makes the top team captain pretty valuable!

So how long should we wait before drafting our TC1? Is the first round too early? The second?

Of course, anything before the final round is too early! Coin flips are random, so while some TCs will end up scoring many more points than others over the course of the season, there’s no way to know which ones. We should therefore be totally indifferent to which TC we end up with.

That’s not the case with, say, running backs. We may be fairly confident that Eddie George will score more points than Tim Biakabutuka. So while we have no good reason to prefer the Raiders’ team captain to the Chiefs’, we should quite rationally prefer George to Biak. And as it makes sense to spend our early draft choices filling positions where our preferences are strongest — indeed, that is the essence of VBD — we ought to generally draft our RBs before we draft our TCs.

It’s no great revelation, of course, that running backs are more predictable than team captains. A much more interesting question is whether running backs are more predictable than wide receivers, or whether tight ends are more predictable than quarterbacks. Are some fantasy positions consistently and substantially more predictable than others? And if so, which ones?

Predictability, or “stickiness”, is an essential part of a player’s value in fantasy football. The same is true in the NFL, but I’ve seen very few articles on the topic of which positions are the most predictable. Many people — Football Outsiders, Bill Barnwell, probably me or Jason Lisk at some point — have shown that field goal kickers are notoriously inconsistent from year to year, which wouldn’t surprise anyone who watched David Akers over the last two years. If you had to guess whether Justin Tucker or Peyton Manning would be a top-five player at their position next year, Manning is the obvious pick.

But what about Manning vs. Adrian Peterson? Or Calvin Johnson vs. J.J. Watt? Since NFL salaries are solely concerned with future expectations, this isn’t just an important theoretical discussion. Suppose a general manager views left tackles and pass rushers as roughly equivalent in terms of positional impact.  If left tackles are “stickier” — i.e., players like Jonathan Ogden, Walter Jones, and Joe Thomas that show elite promise stay at that high level — than pass rushers (Shawne Merriman? Derrick Burgess? Jason Babin?), then that GM should be willing to pay a top left tackle more than he would pay a top pass rusher who had a similar history of production.

Unfortunately, I can’t answer the question of which positions are the most predictable. As a very rough first cut, however, I can at least begin the discussion. I looked at all players since 1990 with at least 3 points of Approximate Value in a season and then noted how many points of AV they had the following year (if they did not play in the following year, they were eliminated from the study). I then ran a very simple regression with Year N AV as the input and Year N+1 AV as the output. The table below shows the best-fit equation for each position. To help make the numbers slightly more meaningful, I’ve also included what the projected AV in Year N+1 would be for each position for a player who had an AV of 10 in Year N. Finally, the final column shows the R^2.

PosCoeffVar10 AV PR Squared
TE0.630.768.180.39
RB0.860.697.810.37
WR1.610.637.940.28
QB1.870.648.290.27
G1.30.657.750.25
OLB2.690.558.220.24
ILB2.460.557.930.21
DT2.140.557.680.21
DE2.280.557.780.2
T2.30.567.90.19
CB2.320.517.40.18
C2.260.537.530.17
S1.990.537.270.17

There are some obvious drawbacks to this system. For example:

  • Approximate Value is an imperfect way to measure, well, just about everyone. The AV system is very good in the sense that a group of 20 players with AVs of 15 will be better, on average, than a group of 20 players with AVs of 10. But for positions where we have good statistics available, we’re better off using those metrics. And for positions where we don’t have good statistics available, AV is better than everything else (other than good game-charting data) but still doesn’t get us very far at all.
  • We could refine the formula with more inputs. A player with a long track record is more predictable than one without it, but this formula just uses one year of AV. In addition, things like age, salary (although that could become circular), draft stock, and other variables could help make this more predictive.
  • I haven’t fully fleshed this out, but I suspect there is probably some bias with the cut-offs I chose; if you switched the AV minimums to 0, or 6, the results probably change. I’m not sure if there’s a “right” number to use.
  • There are many other flaws too numerous to list.

We can’t just put aside those significant caveats, but what can we learn from this quick study? If those caveats don’t change things, then that would make tight end the most predictable positions on offense. I think most would expect quarterback to top the list, as no one has any doubts as to whether Manning, Brady, Aaron Rodgers, or Drew Brees will all be very good in 2013.   But once you get past the top quarterbacks, things begin to change.  I’m not sure I’m convinced that tight end happens to be the stickiest position, but that result doesn’t offend my sensibilities.

I'm not mad at ya.

I'm not mad at ya.

On the other hand, I would have assumed that AV would tell us that lineman are consistent. Truth be told, I don’t know if we’re ever going to get very far in determining how consistent offensive linemen are from year to year. I generally don’t believe any one offensive lineman is all that valuable, so to the extent that lineman are not very sticky, that would further depress their market values. It’s easy to think of the players like Ogden, Jones, and Thomas, but take a look at the Pro Bowl tackles form 2002 to 2011.  There are several one- or two- year wonders in there, and also average lineman who suddenly started making Pro Bowls once their teams got great quarterbacks.

On defense, linebackers would come out as most consistent while players in the secondary have wildly fluctuating values. Are cornerbacks inherently more inconsistent? In 2010, the top 10 corners according to Pro Football Focus were Antoine Winfield, Brandon Flowers, Champ Bailey, Josh Wilson, Ronde Barber, Joe Haden, Devin McCourty, Tramon Williams, Sean Smith, and Brent Grimes. Putting aside Bailey and Barber due to age, the rest of the group didn’t do a great job of holding their value. Winfield, Flowers, and Haden remain very good, but McCourty struggled in 2011 and moved to safety last year while Williams has regressed and has been part of bad Packers secondaries the last two years. Grimes has dealt with injury issues, Wilson has been an average player, and Smith graded out as terrible in 2011 and below-average in 2012. #11 on the PFF list from 2010 was Charger Antoine Cason, who hasn’t come close to playing at the same level since, and #12 was Vontae Davis, who the Dolphins traded before the 2012 season.

Defensive backs are a lot like offensive lineman in that they’re links of the same chain.  Strengthening the weakest link is probably more important than having a single star (although one great cornerback can help out the rest of your secondary more, in my opinion, than one great lineman can help a line), and perhaps what’s going on here is that defensive backs are simply more subject to the randomness of team defense (which is a large component of AV) than other positions. Given the pass-happy nature of the league, teams should have more than enough film to grade defensive backs, and the sample sizes aren’t small like they are with kickers.  So I’m not sure what the deal is there.

What do you think? How can we measure the stickiness of positions? Would you agree that defensive backs are the least consistent starters from year to year? If so, why?

I also think it’s worth mentioning that I’m interested in the actual playing level of a player and not necessarily his production. Receivers are very dependent on their quarterbacks and offensive lines (and other receivers), so their production can wildly fluctuate even if they maintain the exact same level of play; to me, that’s not a sign of them being “inconsistent” (ala Larry Fitzgerald) but rather just a the reality that their production is heavily tied to their environment. Similarly, a cornerback could look great with a lethal pass rush one year and get exposed if his defense can’t get to the quarterback the next; but if the corner’s actual playing level was the same each year, I wouldn’t say he changed as a player.

I don’t have any answers here, but I do think the idea that players who play positions that are the most team dependent — generally those that line up farthest from the ball — can conflate the issue of stickiness.  Of course, a player whose production is heavily dependent on his teammates probably deserves a lower salary cap value for reasons besides how easy it is to predict his future level of play.

{ 5 comments }
  • Ben April 9, 2013, 12:19 am

    Hmmm with the rules the way they are now it seems like offense is more important in determining how a game plays out ie defense is almost dependent on the quality of offense they face. From this perspective offense would be more consistent from year to year. Also, I don’t think this offense is 100% the reason for winning and losing, good defensive play helps, but I think offense is far more important, a good passing offense will shred the best defense in the league. It is for this reason I think defensive players are extremely overpaid from a value perspective.

    Reply
  • Danish April 9, 2013, 5:11 am

    How far along is the snap data thing? Number of snaps may be indicative of “actual level”. That way “scouting” or charting is indirectly incorporated through the coaches evaluation. So the assumption basically is if Player X has Y snaps in year N, and somewhere around Y snaps in year N+1 then he, on average, played at the same actual level.

    Allthough there will be problems with elite players and QBs – they’ll get max-snaps unless they are injured.

    The same could be done with games started as input, I suppose, but the variance in that sample is probably to small, ie. you’ll get a lot of 16s.

    Reply
    • Ben April 9, 2013, 8:18 am

      AV is a better measure of value than any of those

      Reply
      • Danish April 9, 2013, 5:39 pm

        Absolute value, sure. But here we’re talking relative value. AV is very much inluenced by the performance of the sorrounding cast of players, something we should seek to avoid. Snapcount doesn’t seem to me to be quite as sensitive to sorrounding talent.

        All of this is likely academic since the data likely isn’t available.

        Reply
  • James April 9, 2013, 10:28 am

    Good post. This is where I was headed with my most recent comment on the NFL draft chart post.

    Reply

Leave a Comment