It started out as a fluff piece. Er, wait, we don't have fluff pieces on Behind the Steel Curtain. That is for other Sites Which Must Not Be Named. But if we did, this would have been one of those. Because of a piece Neal Coolong posted a bit over a week ago, "Predictions made on the draft are often forgotten, but shouldn't be," I thought I should pony up and have a look at how well the players from my previous years' mock drafts have done. (For those of you unfamiliar with my mock drafts, I use a metric I devised, called "Best Looking Player Available.")
But, being me, I felt the need to back it up with some actual statistics. So more than a week later, with my eyes bleary from squinting at spreadsheets, here is the initial result. My BLA post long ago went out the window. Instead, I tried to come up with a way of comparing drafts directly to the rest of the league.
One of the big difficulties, of course, is the choice of metric for evaluating players. Some evaluating sites, such as Pro Football Focus, have detailed evaluations on each player at each position, but these evaluations only go back to 2008, and so aren't useful when one is trying to get a long-term picture. Some sites, such as Football Outsiders, don't do detailed evaluations on every player, or for that matter, every position. So other than the raw NFL stats, really the only option is the Pro Football Reference Weighted Career Approximate Value. Every player is covered, and even better, can be found in sortable draft tables, which made the data entry easier. So PFR it is. It isn't perfect—in fact, one glaring omission, in my opinion, is that they don't assign any value at all to members of the kicking team. This means the Raiders, who drafted Sebastian Janikowski in the first round, get no credit at all in the 2000 draft for a player who has probably made more points for the organization than any other player during the same period of time. Fortunately, it doesn't come up very often. (This also affected the Steelers during this time period, who drafted punter Daniel Sepulveda in the 4th round in 2007.)
Another problem with comparing drafts is figuring out how much value would be "average" from a given pick in a given round, in a given year. In theory, each team has one pick from each round, and those picks have a certain value. In actual fact, I can tell you, having typed in the picks for each team for the past 13 years, it is very rare for a team to have a "classic" draft in which they have a single pick in each round. Obviously it isn't fair to compare the outcome of the draft for a team who had two first-round picks (something which is more common than I would have guessed) to a team who had none (also fairly common.) There are reasons teams make these trades, and one would assume they think they are gaining an advantage thereby, but nonetheless for purposes of straight comparison it's necessary to have a way to assign a theoretical value to a team's picks.
One could use the generally derided Draft Value Chart, but it is derided because it is considered to be way out of date. However, five years ago one of the PFR guys decided to update the chart, using the above-mentioned Weighted Career Average Value for players for the previous 20 years. In fact, it turned out to not be as different as you would expect, which goes to show that Jimmy Johnson's instincts about value were pretty sharp.
Either of those metrics, however, value a pick at the beginning of a round much more heavily than one at the end of the round, particularly in round one. There does indeed seem to be a pretty sharp drop-off between the first few players taken in round 1 and the rest of the draft. The drop in the subsequent rounds is much less steep, and by the time you get to the later rounds there isn't much to choose between them.
Therefore, I decided for the purposes of the analysis to value a first-round top six pick at 1.5, the rest of round one at 1.0, a round 2 pick as .61, a round 3 pick as .45, a round 4 pick as .34, and the rest of the draft as .2. There are differences, certainly, between a 5th and a 7th round pick, but not large enough to make it worth the hassle of separating them out. In case you're wondering, those proportions were taken from PFR's work in which they added together the value for each draft position (in other words, say round 1 pick 14). Generally speaking, an "average" 2nd rounder is going to be worth a bit less throughout his career as an average first round pick, and so on. By the time you get to the later rounds, generally speaking they contribute later, if at all, and at a lower level. As the PFR guy who decided to define what a "bust" was said,
It's kind of silly to call any player a bust if he was selected after Pick 64. If he's only supposed to be —at most—32% as good as the #1 pick, can a team/fan really be upset if he doesn't even turn out to be that good?
This is from an article called "Logarithmic Decay: How the NFL is like a Sprite." Here's the chart he plotted:
This shows extremely well just how steep the drop-off is, and how it levels out. Definitely worth a read if you are in the least interested in such things.
Using these figures, I added up the value of each draft, using the PFR draft charts. I also added together the value of each team's picks, because it obviously makes a difference whether you had two first round picks or started with round 3. I then averaged the numbers for the whole league for that year, and then ranked the AFC North teams by where they fell in the spectrum. Here are the individual charts first, and then all four teams superimposed:
Since these charts are based on the Weighted Career Average for players, obviously the last several years aren't going to look the same in a few years, as some players fall out of the league and some improve. These charts also don't say too much about the teams themselves, because these are just the drafted players, who may or may not be playing with the team who drafted them. So, for instance, San Diego got the credit for Eli Manning, and, for that matter, Drew Brees. And as we've all noted, it is increasingly the case that the team who drafts and develops a player may well not be able to afford him when he becomes a free agent.
Thus this article is only looking at how well a team drafts. These numbers don't necessarily say much about how well a team performs. It is telling that this year's Super Bowl champion had drafts which ranked at or below the league average for the past four seasons.
The charts above compare to the league average. Now let's look at the rankings for each team (lower is better):
Finally, here is a chart showing which teams, league-wide, had the best and worst ratio each year of draft pick value to draftee value, (which I'm calling Draft Efficiency) and who had the highest and lowest total pick values each year:
That's all for the moment. I welcome your comments as to where I went wrong : ), and I'm working on a series of ratings by position to try and pin down where the value might be.