This will still focus on the years from 2004 to 2012. Since 2004, 287 receivers have been drafted, and 24.39% (70 players) reached the 35 yards per game played average. As the bar is raised to 45 yards, the number of players who qualify will obviously drop. We're not looking for the average anymore, but moving towards the exceptional.
In total, 36 receivers managed to reach this threshold of 45 yards per game played during this time frame. They are all listed below, and divided into 3 categories, which I will explain in a moment. As well as listing their Athletic Score (based on their combine/pro day data), and Stat Score, the list will also include what pick they were selected with, their average number of receiving yards per game played in the NFL, and what type of receiver the computer classified them as (either Big, Small, or Midget).
Group 1
Player Yards/GP Pick# Stat Score Athletic Type
| Larry Fitzgerald | 74.4 | 3 | 1.303 | 0.822 | Big |
| Lee Evans | 50.9 | 13 | 1.516 | 0.194 | Small |
| Roddy White | 68.2 | 27 | 1.093 | 0.536 | Small |
| Julio Jones | 74.4 | 6 | -0.020 | 1.370 | Big |
| Torrey Smith | 53.0 | 58 | 0.648 | 0.637 | Small |
| Johnny Knox | 49.2 | 140 | 0.111 | 0.214 | Small |
| A.J. Green | 77.6 | 4 | 0.027 | 0.127 | Big |
| Braylon Edwards | 54.1 | 3 | 1.003 | 0.429 | Big |
| Dez Bryant | 66.8 | 24 | 0.358 | 1.425 | Big |
| Brandon Marshall | 72.5 | 119 | 0.096 | 1.003 | Big |
| Calvin Johnson | 85.2 | 2 | 0.742 | 2.640 | Big |
| Greg Jennings | 68.1 | 52 | 1.414 | 0.086 | Small |
| Pierre Garcon | 46.4 | 205 | -0.073 | 0.696 | Big |
| Justin Blackmon | 54.1 | 5 | 1.239 | -0.071 | Small |
| Dwayne Bowe | 65.1 | 23 | 0.033 | 0.703 | Big |
| Marques Colston | 72.5 | 252 | 0.322 | 0.799 | Big |
| Hakeem Nicks | 67.7 | 29 | 1.023 | 0.518 | Big |
| Santonio Holmes | 59.9 | 25 | 0.331 | -0.047 | Small |
| Mike Wallace | 64.2 | 84 | -0.054 | 0.503 | Small |
| Vincent Jackson | 57.4 | 61 | 1.914 | 2.089 | Big |
| Chris Givens | 46.5 | 96 | 0.498 | -0.049 | Small |
| Mike Williams (TB) | 56.9 | 101 | 0.180 | 1.030 | Big |
| Kenny Britt | 52.3 | 30 | 1.000 | 0.678 | Big |
| Average | 62.4 | 59.2 |
Group 2
Player Yards/GP Pick# Stat Score Athletic Type
| Demaryius Thomas | 61.3 | 22 | 0.302 | N/A | Big |
| Roy Williams | 49.7 | 7 | 0.475 | N/A | Big |
| Michael Crabtree | 57.7 | 10 | 0.949 | N/A | Big |
| Average | 56.2 | 13 |
Group 3
Player Yards/GP Pick# Stat Score Athletic Type
| Stevie Johnson | 50.5 | 224 | -0.406 | -0.182 | Big |
| Josh Gordon | 50.3 | 39 | -0.999 | 0.774 | Big |
| T.Y. Hilton | 57.4 | 92 | 0.333 | -0.545 | Midget |
| Brian Hartline | 45.9 | 108 | -0.606 | -0.324 | Small |
| Sidney Rice | 46.0 | 44 | 0.751 | -0.263 | Big |
| Denarius Moore | 48.5 | 148 | -0.109 | -0.114 | Small |
| DeSean Jackson | 67.4 | 49 | 0.066 | -0.904 | Midget |
| Jeremy Maclin | 58.5 | 19 | 0.241 | -0.237 | Small |
| Antonio Brown | 54.3 | 195 | 0.490 | -0.963 | Small |
| Percy Harvin | 61.1 | 22 | -0.506 | N/A | Small |
| Average | 53.9 | 94 |
Jordy Nelson (44.9 yards/GP) and Mario Manningham (44.6 yards/GP) just barely missed making this list, but could still wind up on it eventually, as well as a few other players. The list is fluid, and over time different players could move into it, or out of it. Even by the end of the next season, I could well imagine a few names changing on this list. What I suspect won't change, is the overall picture of why they are succeeding.
Group 1 has 23 receivers in it. This group represents the receivers who were at least in the average range, in terms of college production, as well as their performance at the NFL Combine, relative to their peers. The cutoff for this group is a score that is no worse than -0.100 in either of the two categories, similar to what I showed here. As I've mentioned before, these are the receivers that conform to the computer's expectations for the most likely prospects to achieve success. Group 2 consists of just 3 receivers, who didn't participate fully, or at all, in the combine, leaving me with a somewhat incomplete picture of them. I'll deal with them, to some degree, in a minute. Group 3 is made up of 10 players who were below average either in terms of college production, or athletic ability, yet managed to succeed.
Since the 3 receivers of Group 2 had above average statistical production in college, they can't be eliminated as prospects who potentially fit in the Group 1 category. Unfortunately, their lack of combine data makes it unclear as to whether they met the minimum average requirements in terms of athletic ability. In the case of Demaryius Thomas and Roy Williams, I would strongly doubt that they didn't exceed these requirements, based on what is known about their size and speed. In Roy Williams' case, we at least know that he ran a 4.48 forty yard dash, and this already starts to move him in the Group 1 direction, even if the rest of his combine data is unknown. Similarly, with Demaryius Thomas, pre-combine chatter suggested that he was running in the 4.3 range. With Michael Crabtree, I'm less confident, but he still generally appears to at least be athletically average, and does possess good size for the positon. So, with some hesitancy, I will suggest that all three probably belong more closely to Group 1 than to Group 3, for what I am about to lay out.
So, overall, what are the chances that the receiver that your team selected in the draft will end up on this list of high achieving players? Well, that depends on how you look at things. Out of the 287 receivers drafted in this time frame, these 36 receivers represent 12.54% of the total pool. So that is what your chances would be of randomly pulling one of their names out of a hat (which seems to be the method that teams employ in the draft). That would work out to an average of 3.99 receivers per draft class who are likely to become significantly above average (the average draft class has about 31.88 receivers selected). Since the actual pool of of receivers is larger than just these 31.88 who get drafted (we can't ignore the undrafted players who make a roster, though it is difficult to say how often this happens), your team's real chances of randomly picking a highly productive receiver are even lower.
Amongst this select group, between 63.8 - 72.2% fall into the Group 1 category of players (depending on whether you agree that the Group 2 players probably should be in Group 1). This is a group that the computer has shown some reasonable success at identifying as quality prospects. So, not only do the computer's favorite prospects make up the bulk of the great successes, their somewhat rare traits make them much easier to identify. As I've said before, players of this sort appear to become at least average (achieving 35 yards/GP) about 67.5% of the time. In reality, the individual odds can still be greatly affected by where a player was drafted, and the biases/opportunities that come with this. Whether looking at the average receiver (35 yards/GP) or the somewhat exceptional (45 yards/GP), Group 1 players maintain a similar share of the overall successes, and appear to be the safest bets. That their average draft position is about a full round higher than the players in Group 3 (34.8 picks higher), shows that teams aren't completely oblivious to the talent of the receivers in this group, though they still sometimes fall further than you might expect.
The Group 3 receivers make up the remaining 27.7% of the highly productive receivers. The problem with players in this group is that there is sometimes little discernible pattern to them, to help predict their future success. Their traits, both physical, and in terms of college production, can vary wildly. As it stands, about 1.11 players per draft class will find a high degree of success, while being part of Group 3. So, out of the 287 receivers we are looking at, they have about a 3.48% chance of reaching the 45 yards/GP threshold. I have no interest in dismissing the successes in this group, but without some way of reliably identifying their potential, I think they just present too much risk to invest a high draft pick in. In some cases, like with Denarius Moore of Jeremy Maclin who fall just a little bit below our targets, a look inside the smaller scores which combine to form the Stat Score and Athletic Score, can alleviate some concerns, and still make them appealing prospects. Spending a high draft pick though, would generally make me nervous. Still, that veers off of the point to some degree.
There are just a couple brief things I should mention about some oddballs from Group 3. For one, Josh Gordon's Stat Score is probably unfairly low, due to his leaving college after just one year as a starter. If he had played through his junior year, and merely replicated his adequate sophomore stats, his Stat Score would be -0.455 (since the Stat Score is based on a player's final 2 years in college). Most likely, he would have done even better, based on how other receivers tend to improve from year to year, but it's impossible to say for sure. A similar issue comes up for Stevie Johnson, who's role in the Kentucky offense shifted dramatically between his junior and senior years. He went from the 7th leading receiver (12 rec., 159 yards, 1 TD) to the 1st (60 rec., 1041 yards, 13 TD) in this time. However, in his case, his Athletic Score still would have put him in Group 3. Percy Harvin is also a bit of an oddball, as his role in college was sort of a hybrid RB/WR type. In his last two years at Florida he rushed for 1,488 yards, averaging 64.6 rushing yards per game. At the same time, he had 1,502 receiving yards, averaging 65.3 yards per game. Taking this sort of production into account would have likely changed the computer's view of him as a prospect, but the system wasn't built for such oddities. WR or RB? It was hard for the computer to tell. Harvin's speed (4.39 forty yard dash), relative to his size, was above average, but data relating to his Agility Score is unavailable. So, his placement in Group 3 is more a product of insufficient data (and general weirdness), than inadequate athletic ability or poor statistical production.
In some cases, like with DeSean Jackson, T.Y. Hilton, and possibly Denarius Moore, their Athletic Scores would appear better if I finally made the Midget class of receiver an official classification, but as it stands they are being measured against conventional Small receiver types. The Midget class appears to be identifiable (largely due to their exceptional speed, and high 2nd Gear score), but there just haven't been enough successes in that category for me to feel entirely confident about establishing it as its own 'type'. Honestly, in the past 10 or so years, Desean Jackson is the only example I can come up with, who has maintained a high degree of productivity for a player of his size, though a handful have worked their way into the lower tier 35 yards/GP group. For this reason, I am very interested to see what will become of the much hyped, and similarly tiny, Tavon Austin, who would appear to need to defy some serious odds to justify his draft position (8th overall pick).
If it's starting to seem like I could make excuses for the majority of the receivers in Group 3, that's because I probably could. For the most part, they all showed some signs of potential, but perhaps ran into one hurdle that they didn't quite clear. Still, I think their scores are more appropriate for mid-to-late round draft picks, where these risks are more acceptable. The real statistical outliers in this list are Brian Hartline and Antonio Brown, for whom the computer never would have predicted significant success.
Beyond appearing to be much safer picks, the Group 1 players also seem to be outperforming the Group 3 receivers, by an average of 8.5 yards per game. That is a 15.76% improvement over the Group 3 average. So, not only are Group 1 receivers easier to identify, but when they succeed, they do so to a higher degree.
Are there some obvious ways to criticize what I've just tried to lay out? Sure. Suggesting that the Group 2 receivers probably belong in Group 1 is a good example. That little maneuver swings 8.3% of the players on the list towards my chosen group. You just have to make up you own mind as to whether that is fair, and if these players likely do possess above average athleticism. Making excuses for why a player like Percy Harvin didn't score well is another (though I did put him in Group3, so it was of no real benefit to me). Not fully delving into the Midget class of wide receivers, to explain exactly why I think they are also an identifiable receiver type, could also be seen as a bit of a dodge (though I do get into this a bit at the bottom of the post about the 2nd Gear, and will probably pursue it in the future). All I can say is that trying to boil this all down to two basic sets of numbers isn't ideal, but fits what I think is a practical limit for discussion in a blog. There are even more numbers within these numbers, and sometimes going deeper into the rabbit hole does provide better answers. I'm shooting for a sort of relative simplicity here. Hopefully, whoever is crazy enough to read this will make some allowances.
As always, I'm not trying to suggest that a player can't succeed with below average combine results, or mediocre college productivity. I'm just suggesting that teams who constantly pick such players (and many teams do), are probably taking an unnecessary risk. As much as some scouts may have a good eye for talent, constantly believing that your unmatched wisdom alone can overcome the 87.46% probability that your selection is incapable of becoming an above average receiver (and a 75.61% chance they won't even be average), seems a bit foolish. Yes, there may be 1.11 somewhat hard to identify gems that exist in an average draft class, but without having a clear and reliable way of pinpointing who they are, such speculation is very risky. What happens when you gamble on your gut? Well, just looking at the first 2 rounds from this same time frame gets you something like this:
Player Yards/GP Pick# Stat Score Athletic Type
| Darrius Heyward-Bey | 37.0 | 7 | -0.313 | 0.872 | Big |
| Anthony Gonzalez | 32.7 | 32 | -0.779 | 0.129 | Small |
| Michael Clayton | 31.1 | 15 | 0.083 | -0.546 | Big |
| Arrelious Benn | 23.3 | 39 | -0.360 | 0.923 | Big |
| Troy Williamson | 23.1 | 7 | -0.231 | 0.593 | Small |
| Mark Bradley | 22.5 | 39 | -1.304 | 0.475 | Small |
| Craig Davis | 21.5 | 30 | -0.423 | N/A | Small |
| Brian Robiskie | 13.1 | 36 | -0.335 | N/A | Big |
| Sinorice Moss | 11.4 | 44 | -0.739 | N/A | Small |
| Limas Sweed | 3.5 | 53 | -0.368 | 0.571 | Big |
| Average | 21.9 | 30.2 |
Is it a coincidence that I limited this list to just 10 busts (or moderately disappointing, in Heyward-Bey's case), which averages out to 1.11 per year? No. For every player who is a pleasant and somewhat unforeseeable surprise, there will be at least one highly drafted receiver like these guys, who will be taken about 63.8 picks (or 2 full rounds) ahead of the actual gem they were looking for. Do you really believe that your team's scouts, and their magical guts, can tell the difference between Brian Robiskie and Antonio Brown?
I'm not saying all of this as part of some pointless attempt to get people to subscribe to my peculiar methods. I am only bringing this up because I feel some people are overly dismissive of the importance of statistical production, and the data from the combine/pro days. However you reach your conclusions, based on the information you acquire, is your business (as long as it works). On the other hand, disregarding the information that is available to you, and placing your faith in the wisdom of your gut, frustrates me. Watching teams continue to invest picks in players that poorly fit the typical models of successful players from the past simply frustrates me. Accepting a high bust rate, frustrates me.
I'm starting to think that this blog is turning me into a loon. Oh, well.
No comments:
Post a Comment