Tuesday, June 11, 2013

Overall NFL Draft Success Rate

I wanted to explore this idea, of how often draft picks succeed, but there really is no way to do this that is foolproof, or won't spark some arguments.  Any measurement you use is going to have its shortcomings.  Do you base it on how many games a player started?  Plenty of guys end up starting, even if they are pretty bad.  How about Pro Bowl or All Pro selections?  Eh, that is kind of hit and miss too.  Lots of players get selected to these All Star teams, based on past reputation, rather than their current playing ability, while good players on bad teams get overlooked.

Still, if we can accept that this will be somewhat inexact, I think a  fairly decent picture of what is going on can be achieved.  No matter how I try to lay this out, there will undoubtedly be people who raise some 'what ifs' or 'yeah, buts' about the views I have.  I'm just shooting for a rough ballpark figure, to get an adequate approximation of the truth.

“I think you have to divide it into top 12 and bottom 20. If you’re in the top 12, it ought to be in the .640 range. That’s about 4.5 guys on average per year out of the seven. You measure that at the end of three years and what you are measuring is whether or not those guys become winning players, guys that contribute to wins. Bottom 20 is .571, that’s four out of seven…"- Bill Polian, former GM of the Colts

"If you look historically, teams get 2.3 (32.8% of the 7 picks) starters per draft and as a team, I think you need to strive to get 3 starters per draft (42.8% of the 7 picks) , or I should say players worthy of starting."- Mike Reinfeldt, former GM of the Titans

In both of these cases, I think these GMs are blatantly full of sh*t.  Mike Reinfeldt is closer to the truth, but is still probably overestimating a bit.  Polian's claims/expectations, on the other hand, appear to be wishful thinking.  Polian seems to factor undrafted free agents into his calculation, which you really shouldn't.  The "guys that contribute to wins" line is also extremely vague.  There is also a tendency for people to grade these things on a curve, where late round players aren't counted as failures due to the lower expectations people have of them.  Of course, when a late round player performs well, suddenly the people involved want that to be included in their resume.

For our purposes, I'm just interested in knowing what percentage of all the players selected became meaningful contributors, regardless of round.  If a first rounder busts, but a seventh rounder blossoms it all balances out in my eyes.  One of the more acceptable measures of this that I have run across was written by Tony Villiotti over at Draftmetrics.com.  It doesn't completely answer all the concerns I might have, but I think it is a good starting point.

Towards the bottom of Mr. Villiotti's paper, he lists the percentage of players at each position, and from each round, that manage to start at least 8 games per year (on average) in their first 5 years.  This is a fairly low bar to set, in my opinion, for judging the degree to which a player is a success, but is probably overall a fair compromise.  What I'm particularly interested in are the overall results listed in the final column of his chart.  He suggests that from 1991 - 2004, 21.5% of all players drafted managed to meet his 8 game per year standard.  Yes, players taken in the first round met this standard 73.6% of the time, and 7th rounders only met it 5.9% of the time.  In the end, the likelihood of a player meeting the standards required here, goes down by about ten percent for each consecutive round.  I'm just interested in the overall percentage, since I want to know how many successful players we should expect a team to accumulate in an average year.  Based on this overall 21.5% probability, this would work out to 1.5 players out of seven total draft picks, which lines up with my own observations.  Obviously, this is much lower than the estimates given by Mr. Polian, and Mr. Reinfeldt (though they were speaking about what would constitute a good draft, rather than just an average one, they are still being overly generous as I will eventually attempt to show)..

Now, even with this information, we run into some problems.  For one, these numbers could still be somewhat inflated.  I don't think there can be much doubt that the 73.6% success rate on 1st rounders is boosted by the likelihood that teams will keep trotting a player out onto the field, even when they are playing somewhat poorly, simply because of their draft status.  On the flip side, it is hard to judge the 5.9% success rate of 7th rounders, when we never know if they have been given any real opportunity to prove themselves.  In this case their low draft status, and preconceived notions of their value, works against them.  Continuing to give opportunities to high round picks, because a team still expects them to become a quality player and confirm their initial appraisal of the player's talent, obviously goes hand in hand with stymieing the success of the late round player.  There are only so many starting spots available, after all.  Either way, we'll ignore this issue, since their isn't much we can do about it.

Even if a player is managing to start games, we can't be certain that he is actually playing well.  Making subjective assessments of a player is tricky, and probably part of why so many turn out to be busts in the first place.  Still, I thought I would try and apply some subjective grades towards the team I am most familiar with, the Baltimore Ravens, to get a sense of what is going on.  To do this, I simply assigned a score from 1-5, for every player they selected from the years 2000 to 2010 (87 players in total).  While this focuses on only one team, it hopefully still serves as a useful example.

While this is a very loose and subjective approach (and quite possibly idiotic), I was trying to be as generous to the team as possible, to see the most optimistic outlook on what is going on.  Players who never got playing time or were quickly bounced from the league got a score of 1.  Players who got some playing time but were notably horrible got a 2 (WR Clarence Moore, for example).  Anybody who I felt could be perceived as at least an occasional starter or journeyman type got a 3 (DE Antwan Barnes, perhaps).  Players with a 4 or 5 grade were the obvious sorts, who had distinguished themselves in some way, either statistically, or through Pro Bowls, as well as through longevity.  I will admit though that I tended to take a point off for special team players (mainly kickers and punters) and for fullbacks, simply because of their reduced positional value.  Either way, the truth is that the scores in themselves don't entirely matter.  I just wanted to see what percentage of players could arguably be seen as average or better, even if I graded them a bit more generously than I believe that they deserve.

So, being very forgiving of the mediocre, the Ravens came up with 41.36% of their 87 selections since 2000 as being at least passably average.  Now, this whole approach is obviously somewhat foolish, and not the least bit scientific, but I just wanted to see what the best case scenario was if I only weeded out the most blatant and inarguable busts such as Troy Smith or Yamon Figurs, which make up the remaining 58.64% of the teams picks.  In reality, I suspect we could argue that their overall success rate might be closer to 30-35%, which would still be higher than league wide 21.5% I mentioned earlier.  This would still be a noteworthy improvement over the average result, for a team that is perceived to be one of the best in the draft, while still being lower that the public might expect.

As an example of what an exceptional draft might look like, I thought the Chargers' 2005 draft class might serve as a good example.  With the selections of Shawne Merriman, Luis Castillo, Vincent Jackson, and Darren Sproles, you could argue that their success rate that year was 57.14% (they had 7 selections that year).  This probably represents, or is close enough to representing, the upper limit for what teams can currently aspire to achieve.  This is obviously far from a typical result.

So, what's the point of all of this?  The reason all of this interests me, is that I suspect teams should be doing much better.  From what I have observed, teams should be shooting for a success rate closer to 70%.  While there will always be up and down periods, largely due to varying levels of talent in a draft class, but also some random luck, teams should almost never be dipping below a 50% success rate.  Based on the data available to them, from the combine and college stats, the frequency with which players at the exceptional end of the "measurable talent" bell curve succeed is much higher than they seem to realize.  I wouldn't expect a radical shift in success rate in the first 2 rounds, where teams generally do a reasonable job of selecting the obviously superior players, but more so from the 3rd round onwards.  At that point, it seems to me that their failure rate is much higher than it should be.  While the overall availability of highly talented players does diminish as the draft goes on, the ones that remain and go on to become successes, are more readily identifiable than people seem to acknowledge.  The players who reside on the extreme low end of the bell curve, in terms of athletic ability and college production, have nearly zero chance of panning out.  Investing a team's draft picks in players who have more exceptional qualities, even if they aren't flawless, at least has some plausible chance of working out, based on who we have seen emerge from mid to late round obscurity in the past.

As I continue to add to this rambling blog, I'll try to explain/justify some of my more deranged beliefs.  For now, I'm just trying to avoid having any of these updates turn into an unintentional novel, that will just put people to sleep.  Hmm, I might have already failed at that.

No comments:

Post a Comment