TE positional advantage is bogus

[Ka Boom!]

We hear all the time how Graham/Gronk/Thomas provide "the best" positional advantage in fantasy football. I disagree and will use statistics to make my case.

Anyone making the assertion that the top 3 TE's give a huge positional advantage must account for the large number of wide receivers that are in lineups every week. I play in 12 team, 3 WR, 1 TE, 1 flex PPR leagues. Obviously my analysis won't apply to every league. However, 12 team 3 WR, 1 TE, PPR is fairly standard. Since the flex can be used in any position, I've excluded it from analysis.

How big of an advantage does having one of the 3 stud TE's give an owner versus a stud WR? To answer that question, I've compared the big 3 TE's versus the next 9 TE's. Since there are 3 times as many WR's as TE's, I've compared the top 9 WR's versus the next 27 WR's.

Here are the top 3 TE through week 12 (I've subtracted an average week for any Steeler or Panther):
180.4 (Gronk), 162.3 (Graham), and 154.6 (Thomas). This averages to 165.7. The next 9 TE's average 120.21. The positional advantage of the top 3 TE's is 45.56 points through 11 weeks (12 if you count the bye).

The top 9 WR's average 202.15 points. The next 27 average 141.31 points. The positional advantage of the top 9 WR's is 60.84 points through 11 weeks.

The top 9 WR's have a 60.84 point advantage compared to a 45.56 point advantage for the top 3 TE's.

Interestingly enough, the top 3 QB's have a 67.84 point advantage over the next 9 QB's.

The next time anyone claims that Graham, Gronk, and Thomas give the best positional advantage in FF, there are 7 appropriate responses:

[Pac_Eddy]

The next time anyone claims that Graham, Gronk, and Thomas give the best positional advantage in FF, there are 7 appropriate responses:

I've got two comments on your logic.

1. Part of the advantage of the big three TEs is we know who they are. Almost everyone had them as the top three and they are. That consistency is huge. You compared these known three to the points of the top nine WRs. Are they who we thought they'd be? No Calvin, Julio, Keenan, Alshon or Marshall. Try the same results with the preseason top nine WRs by ADP.

2. One partial season is a small sample size.

[snitchinsider]

This fails to account for many things:

1. How many premier receivers can be found on the wire versus TE
2. How predictable or unpredictable the top talents at each position group are (Predicting the top 3 TEs is fairly solid barring injury, predicting the top 9 receivers is absolutely not)
3. All three of the top 3 TEs this season have missed time or have been not playing "normal" snaps due to injuries, which softens the disparity... note that despite such fact, they are still the top TEs.

[John Paul]

It depends on lineup and scoring system, no one belief trumps that. Your league's requirement for 3 WRs changes the stats, so you've adjusted correctly.

I do have a question though, how does your math work out for 2013, 2012 and 2011(that's when this TE advantage took off)? Is this assessment consistent in your league or just so far in 2014?

[joeday]

It depends on lineup and scoring system, no one belief trumps that. Your league's requirement for 3 WRs changes the stats, so you've adjusted correctly.

I do have a question though, how does your math work out for 2013, 2012 and 2011(that's when this TE advantage took off)? Is this assessment consistent in your league or just so far in 2014?

Thats a good question, because looking at just 11 games in 1 season and not looking back to when Graham and Gronk came on to the scene could make a flaw in the OP's argument.

[FuzzyD]

I appreciate the quantitative approach you've taken to this question. However, I think you're missing an important part of it.
Don't you find it interesting that the pre-season consensus top-3 elite tier of TEs turned out to ACTUALLY be the top-3 elite group. The ability to accurately project Graham-Gronk-Thomas as top-tier adds value. I'm guessing that the top-9 WR list you analyzed differs dramatically from the "consensus top-9" from before the season. In order to benefit from the "point advantage" of a group, you need to forecast those players. Easier to do at TE (at least in recent history).
The top tier WR group seems to be a lot more fluid than top tier TE. We will almost surely see new names next season in the elite WR class. Not as likely, however, that we see a new TE crack the top-3.

[Ka Boom!]

The next time anyone claims that Graham, Gronk, and Thomas give the best positional advantage in FF, there are 7 appropriate responses:

I've got two comments on your logic.

1. Part of the advantage of the big three TEs is we know who they are. Almost everyone had them as the top three and they are. That consistency is huge. You compared these known three to the points of the top nine WRs. Are they who we thought they'd be? No Calvin, Julio, Keenan, Alshon or Marshall. Try the same results with the preseason top nine WRs by ADP.

2. One partial season is a small sample size.

Good points. I know about the small sample size. If you or anyone else wants to run with the idea and make a better analysis, go for it. I'm lazy and don't feel like doing the work.

As far as knowing the big 3, I see your point. Doesn't change my mind, but I see your point.

[John Paul]

I also prefer to do this sort of analysis on PPG. You can remove injury games on the positions and get a clearer look at what advantage you are targeting.

Side note: I use the same logic in trying to get Watt in IDP leagues, he is the Jimmy Graham of DLs.

[joeday]

I appreciate the quantitative approach you've taken to this question. However, I think you're missing an important part of it.
Don't you find it interesting that the pre-season consensus top-3 elite tier of TEs turned out to ACTUALLY be the top-3 elite group. The ability to accurately project Graham-Gronk-Thomas as top-tier adds value. I'm guessing that the top-9 WR list you analyzed differs dramatically from the "consensus top-9" from before the season. In order to benefit from the "point advantage" of a group, you need to forecast those players. Easier to do at TE (at least in recent history).
The top tier WR group seems to be a lot more fluid than top tier TE. We will almost surely see new names next season in the elite WR class. Not as likely, however, that we see a new TE crack the top-3.

To your point, since Gronk and Graham came on the scene in 2011, the only WR that made the top 3 scoring more than once at the end of the season was Calvin Johnson. In 2011 it was Calvin, Welker, Cruz...in 2012 it was Calvin, Marshall, Bryant...in 2013 it was Demaryius, Antonio, Gordon. 2014 is shaping up to change that trend though, at least I hope since in one of my main leagues I start Dem and Antonio lol. But players like Cobb, Hilton, Sanders, Maclin, Tate, Kelvin, Evans are all top 12 WRs, none of which cracked the top 12 in any previous season (since 2011).

[John Paul]

The top three TE all also benefit from good offensive systems with a QB that can get them the ball. However, Brees, Brady and Manning are in the twilight of their careers. So looking ahead, try to get the TE that fits the athletic profile that is matched up with a younger QB that can/will get him the ball. Players that I think fit in that category:

Kelce
Eifert
ASJ

We haven't seen enough from Osweiler or Garoppolo to project a major decline from Thomas or Gronk. Graham has a couple more years with Brees IMO.

[Telperion]

The biggest flaw in your analysis is that you assume a linear regression whereas PPG is best fit to a logarithmic model.

Simplified, the difference between TE3 and TE7 is much greater than TE5 to TE9 despite being only 4 apart.

TE has a steeper curve. I can graph this later today.

[Csl312]

Good points. I know about the small sample size. If you or anyone else wants to run with the idea and make a better analysis, go for it. I'm lazy and don't feel like doing the work.

Not trying to be mean or anything but it is a little bit unrealistic for you to make an argument, claim you are going to use numerical analysis to back it up, then expect someone else to prove your point for you. Especially when the others you are addressing don't agree with your conclusion. I understand not having the time/desire to broaden the analysis but why post here in the first place then?

[DFHustle]

The biggest flaw in your analysis is that you assume a linear regression whereas PPG is best fit to a logarithmic model.

Simplified, the difference between TE3 and TE7 is much greater than TE5 to TE9 despite being only 4 apart.

TE has a steeper curve. I can graph this later today.

I was going to make the same comment. I'd also rather see a PPG analysis as well since missed games can heavily effect total season points. Also, each week should be treated as an individual point for each player in both the TE and WR groupings. From this distribution, which may or may not be a normal distribution, you can then compare elite vs. non-elite performance more properly. By using the averages, you are assuming the distribution is normal; the median might be more appropriate.

Also, forcing groups can skew data. For example, what if the 4th highest TE was actually just a few total points behind the 3rd? Since you force that into the 2nd grouping you are corrupting the average of that group. Things should be done somewhat blind.

There are many flaws in your original analysis and it's always easy to find data to support any already biased opinion of what you want it to be. I have no idea what the answer is, I might mess around with this tonight when I get home as I love statistical challenges.

[dlf_jules]

We hear all the time how Graham/Gronk/Thomas provide "the best" positional advantage in fantasy football. I disagree and will use statistics to make my case.

...

The next time anyone claims that Graham, Gronk, and Thomas give the best positional advantage in FF, there are 7 appropriate responses:

I respect the thinking behind this project, but there are a few important methodological flaws in your work. And when you conclude a post the way you did, you'd better be darn sure that your conclusion is right and your reasoning is perfect. I hope you keep up this type of work and thinking, but also remember not to overstate your case.

FWIW, I've done some half-hearted analysis of this issue, and I tend to agree that TE positional advantage is at least overstated. I hope to do a more thorough post this offseason.

[dlf_jules]

I have no idea what the answer is, I might mess around with this tonight when I get home as I love statistical challenges.

I hope you do; it would save me some work. Here are a couple thoughts (some/all of which you may already have considered):

The trickiest part of this analysis is figuring out what the baseline should be at each position. In a 12-team QRRWWWTF league, there are about 14 TEs, 42 WRs, and 28 RBs starting each week. It's tempting to set the baselines at TE14 and WR42 (RBs are a higher degree of difficulty because so many have week-to-week value). But, especially in a dynasty league, there are usually only between six and eight teams in contention most years. Those teams hold a disproportionate number of the top players, so the baselines should probably be higher -- something in the range of TE7-9 and WR21-27.

Another tricky part is figuring out what we mean by positional advantage. Value over baseline (often called "VBD" based on David Dodds's Value Based Drafting principles) is a simple measure, but I'm not sure it's the best. Another method (one I've seen Adam Harstad use) is to take the average difference between Player X and all other players at or above baseline. So Gronk's positional advantage would be the average of [Gronk - Graham, Gronk - Thomas, Gronk - Olsen, Gronk - Bennett, ..., Gronk - TE9 (or whatever your baseline is)].

Hope this helps!