Career Wide Receiver Production: 7th Round Darlings

Categories: NFL Draft, Statistics
Comments: No Comments
Published on: March 10, 2013

People say that the quarterback position is the most difficult to project into the NFL.  I respectfully disagree.  Certainly, the quarterback position is difficult to project from the college to professional game, and we’re not very good at it.  The empirical evidence shows decision makers are essentially making a random guess when it comes to projecting quarterbacks.  The only reason first round picks seem to have better numbers is because decision makers expect them to be good and provide them more opportunities – a self-fulfilling prophecy.

But there are worse things in this world than randomly guessing.  You will be right sometimes if you’re randomly guessing.  More than that, if you keep guessing randomly over and over and over again – say once a year every April since 1936 – the number of times you’re right and the number of times you’re wrong should be essentially equal.  Random decisions are not the worst possible outcome.  It is entirely possible that you could be on the other side of random.  It’s possible to make decisions that are worse than what a random process could achieve.  Sometimes we could make better decisions if we started throwing darts while blindfolded.

As an example, consider how most lay people buy stocks.  Most people don’t want to lose their money in the stock market.  They want high performing stocks that are going up.  So they don’t buy a stock until it starts going up.  And the average person doesn’t want to be burned by market fluctuations, so they continue to watch the stock as it goes up more.  Once the stock gets sufficiently high, they buy the stock.  But this is precisely the worst time to buy the stock.  The stock has nowhere to go but down from this point and the investor is more likely to lose money.  The person would be much better off picking stocks at random because the information they are using is leading to worse decisions.

How does this relate to football?  Football decision makers drafting wide receivers and lay people picking stocks have a lot in common.  To be clear, I have no idea what information football decision makers are using when predicting wide receiver success in the NFL.  Others have studied this (that one is a pdf, be aware if you click on it) and found inconsistent results.  However, just because I don’t know the process doesn’t mean I can’t evaluate the outcomes the process generates.  So let’s do that.  Let’s evaluate wide receiver draft picks and compare that with what we would expect given a random process.

The Evidence

I started with another study of the Hall of Fame.  It seemed to work well for the quarterbacks, so why not go back to the well again?  Where were the consensus great wide receivers drafted?  Here is a histogram of all wide receivers in the Hall of Fame that played since 1945.  Dante Lavelli is not on this graph as the U.S. Army drafted him before the NFL got a chance.  Remember, a random process coupled with a self-fulfilling prophecy will create a J-shaped distribution.

If we delete Raymond Berry as an extreme outlier, the histogram looks like this.

Here we have our first strange finding in the wide receiver data.  For picks 1-80, the data show the expected pattern.  A nice J-shaped pattern as we get further on in the draft.  But what is going on with picks 81-120?  There is a strange increase in Hall of Fame receivers drafted at a point corresponding, in modern times, to the late 3rd through the 4th rounds.  This is our first evidence that wide receiver evaluation is not just random, but biased in the wrong direction.  However, it is far from conclusive.

This pattern is strange enough to demand a more complete analysis of wide receiver production.  Perhaps the bump in the late 3rd through 4th rounds represents the perceived value of the wide receiver position.  We can be reasonably sure that everyone wants a quarterback and is willing to draft their favorite quarterback with the highest pick possible.  However, the same might not be true for wide receivers.  It’s possible that wide receivers are not as highly valued compared to quarterbacks and are not taken with the highest draft pick possible.

To attempt to rule out that explanation, I took Berri and Simmons’s methodology that they used for quarterbacks and applied it to wide receivers.  Let’s examine career production for all wide receivers drafted between 1995 and 2009.  We will look at Career Receptions, Career Yards, and Career Touchdowns.  I chose to start at 2009 because the average career length of an NFL wide receiver is just a shade over 3 seasons.  Any receiver drafted in 2009 is already at the average career length for an NFL wide receiver.  I chose to stop at 1995 because it includes 15 years of data, which is a nice round number, and it was midnight when I finished entering the 1995 data and I wanted to go to sleep.  All data was downloaded from pro-football-reference.com.

The first graph shows the Average Career Receptions for a wide receiver drafted during each round of the draft during the 15 year time period we are looking at.  Unless otherwise specified, I deleted all players that were never credited with catching an NFL pass.  I have run these numbers keeping all players, and the pattern is exactly the same.  Remember that this analysis does not control for playing time, so we are expecting a J-shaped distribution.

Here is a similar graph showing Career Yards

And the same graph showing Career Touchdowns

In all three cases, the data from Rounds 1-5 show the expected pattern.  There are still players drafted in the later rounds that are making an impact, but the impacts tend to be less than those drafted earlier.

However, that wasn’t the really weird part.  The really weird part was what happened in the final two rounds.  I might not have thought much of it if I hadn’t run the Hall of Fame data first.  Let’s zoom in on career production only for players drafted in rounds 5, 6, and 7.

Here is Career Receptions

Here is Career Yards

Here is Career Touchdowns

Wide receivers drafted in the 6th round have, on average, less productive careers compared to wide receivers drafted in the 5th and 7th rounds.  This is true if we measure production based on receptions, yards, or touchdowns.  One-way ANOVAs using only players drafted in the 5th-7th round confirms this, all F’s > 15.90.  My home set up is not great for running ANOVAs, which is why I did this slimmed down version.  Next week I will put all players into an ANOVA and run some splashy post-hoc tests.  Until then, we’re stuck with this slimmed down analysis.

This is a crazy finding.  Compared to receivers drafted in the 6th round, receivers drafted in the 7th are working against the self-fulfilling prophecies of decision makers, coaches, and their quarterbacks.  They should, by all accounts, have worse careers.  And yet, they have the more productive careers than players drafted in the 6th round and are just as productive as receivers drafted in the 5th.  My first explanation for this finding was that receivers from non-FBS schools are more likely to be drafted in the 7th round.  That was not the case in this data.  80.95% of 5th round draftees, 72.5% of 6th round draftees, and 81.81% of 7th round draftees come from FBS schools (Remember we deleted receivers that either did not make a team or did not catch an NFL pass).  At the very least, this isn’t a question of source school .

What should we conclude from this?  I think we have some very clear evidence that talent evaluators should stay away from wide receivers.  The pattern at the top of the draft is consistent with a random process.  I can’t conclude it is random right now because I don’t have good numbers on wide receiver playing time.  However, the pattern is consistent with randomness coupled with self-fulfilling prophecies.  Given what we know about the randomness of selecting quarterbacks, there is no reason to assume that wide receivers would be any different.

The pattern at the lower end of the draft is even more compelling.  In this case, talent evaluators and draft decision makers are consistently wrong about players evaluated as 7th round talents.  Those players are, on average, just as good as players evaluated as 5th round talents and better than players evaluated as 6th round talents.

This leads me to believe that we simply do not know what makes a productive wide receiver.  Furthermore, something about how we evaluate wide receivers is leading us to make worse decisions.  If we could control for the self-fulfilling process effect, we might see lower rated receivers doing BETTER than higher rated wide receivers.  (see This Post for explanation of the correction)

Which ultimately leads me back to my two favorite predictions for this wide receiver draft class, Cody Wilson from Central Michigan and Brent Leonard from Louisiana-Monroe.  Both of these receivers had extremely productive college careers at schools that don’t get a lot of media attention.  Both of them are not rated highly by the draft community for physical reasons – Cody Wilson because he is short for an NFL receiver and Brent Leonard because he is “not fast.”  But I ask you, who cares?  Fast receivers do not do any better than slow ones.  And prioritizing fast receivers might be leading us to draft worse receivers in the high rounds.

Remember, when I started this I said I don’t know what process evaluators are using to grade wide receivers.  However, after looking at the data, I am confident that they should stop whatever it is they are doing.  The decisions being made late in the draft are worse than randomly guessing.  At the end of the day, decision makers would be better off throwing darts at the names of draft eligible wide receivers blindfolded.

No Comments - Leave a comment

Leave a Reply


Welcome , today is Sunday, November 19, 2017