[Update at end of post]
Thank you everyone that took some time to read the introduction to DPI, in which I talked about what this new metric is meant to represent and the principles that I tried to infuse into its creation. Now, I want to go into the concepts behind DPI and how they were assembled to enumerate the worth of a player based on this performance and draft position. There are two components to DPI, and in this post, I will detail the major one, Draft Performance Ratio (DPR), which of itself, can be an interesting tool to evaluate a whole draft class in lieu of individual players.
The driving force behind DPI is Approximate Value (AV), which is a stat developed by Doug Drinen, founder of Pro Football Reference. AV attempts the, no doubt, monumental task of placing a single value on a player’s contributions on the field over a season. You can get more into the nitty gritty of Approximate Value here. Using AV fulfils my desire for DPI to be a numeric indicator that can compare individuals or groups of players, however, I quite understand that such a reductionist philosophy may be an acquired taste. From PFR:
AV is not meant to be a be-all end-all metric. Football stat lines just do not come close to capturing all the contributions of a player the way they do in baseball and basketball. If one player is a 16 and another is a 14, we can't be very confident that the 16AV player actually had a better season than the 14AV player. But I am pretty confident that the collection of all players with 16AV played better, as an entire group, than the collection of all players with 14AV.
Essentially, AV is a substitute for --- and a significant improvement upon, in my opinion --- metrics like 'number of seasons as a starter' or 'number of times making the pro bowl' or the like. You should think of it as being essentially like those two metrics, but with interpolation in between.
Agreeably, AV has already been extensively coupled to the draft process via its use in developing alternative draft scales to Jimmy Johnson’s ubiquitous chart. See OCC’s related recent article here. In that post, OCC linked to a draft value methodology developed by Chase Stuart at Football Perspective who summed the AV accumulated between 1980-2007 for each draft pick in their first five years to produce his draft chart. After some minor correction to values of his final few draftees, DPI incorporated this data set that I call Expected Approximate Value (ExpAV). Per Chase Stuart:
Using each player’s career AV makes sense on some level, as the drafting team gets the chance to have a player for his entire career. But the real value in the draft –especially now thanks to the new collective bargaining agreement — is the ability to get a player for cheap on his rookie contract, which expires after (at most) five years.
Stuart’s evaluation matches my expectations for what a player’s DPI score represents. Also, as an aside, when I was a TA in grad school, I had a bit of a rep as a guy that was known to grade strictly, so instead of using the total AV, I’m going to subtract any AV points the player accumulated anywhere but his drafting team in the first five years of his career. This is Draft Approximate Value (DrAV),and it’s a number I can use directly from PFR.
The 5 year cap merits further exploration. The downsized rookie wage structure under the new CBA offers teams an opportunity to uncover fantastic value with good picks. With a team option on the fifth year of 1st round draftees, the potentially most valuable picks in a team's draft class, it behooves teams to draft with even greater discernment and develop their players so as to retain them at below market values. So rookie wages are the carrot; whence the stick? I'll double dip into Football Perspective on the matter of taking advantage of the rookie salary cap in this great article.
Teams that draft well will not have to replace as many drafted busts with veteran free agents. That means they'll have a higher ratio of first-contract players to free-agent acquisitions, and will therefore get more production per dollar overall. Since total dollars are relatively fixed, that means they'll get more production overall - more wins, more playoff appearances, more Lombardis. If you miss on your high draft picks, you're going to have a difficult time making up for that by paying five times as much for the same production from a veteran. Add in a salary floor that all teams must abide by, and the price of free agents is likely to soar over time as the number of bidders in the market increases (no team will sit on the sidelines for long).
We can now contrast the AV a draftee has accumulated with his drafting team (DrAV) with AV that an the average player that was selected at the same pick generated in his first five years. I call this ratio Draft Production Ratio, or DPR.
DPR = DrAV ÷ ExpAV
DPR is a team-oriented metric since I have used DrAV instead of AV. Any AV accrued after a draftee is released from his original team does not benefit that draft class. General Managers and FOs are penalized for not drafting well and not developing their draft picks, which is how I suppose smart teams navigate the flatlands of the current salary cap era.
Here is the relationship between ExpAV and the DrAV for each pick in the 2008 class. Just as you’d think, expected success decreases non-linearly reflecting the dropping probability of finding a good player later in the draft:
DPR can be considered an alternate method of evaluating draft classes beyond simply looking at aggregate production. Regardless of whether a team had two first round picks or none, you can look at it as a measure of how well it drafted based on the picks it had. And since 2008 just happens to be 5 drafts ago, let’s go ahead and do that for the Dallas Cowboys Class of 2008:
Dallas’ 2008 class got solid contributions from most of it’s members, and it’s top picks even surpassed their ExpAV, slightly. It is worth pointing out again that each player’s DrAV is only counted for Dallas, so even though Martellus Bennett and Erik Walden have both been handed brand new long-term contracts after being let go from Dallas, their more recent success is of no value to the Cowboys, and the team’s draft class is judged accordingly. The DPR for this class is the result of dividing the cumulative DrAV by the cumulative ExpAV, i.e. 82/68, to give 1.31, which means that this group exceeded expectations by 31%.
How does that compare to the rest of the league? Glad you asked.
Based on DPR, the clear winner of the ‘08 draft was Indianapolis, who benefited from great production by later picks like Jacob Tamme and Pierre Garcon. But the Colts’ lack of a 1st round pick, reflected in their low ExpAV, tells you they overachieved significantly on low expectations. In that sense, it’s worth looking at both Cleveland and Atlanta who ended up with a DPR of 1.78 but with a significant difference in their ExpAV -- the Falcons had two 1st round picks and a total of 6 picks in the first 3 rounds whereas the Browns did not make their absolute first pick till the 4th round. Clearly, Atlanta had the better draft, but DPR asks us to acknowledge that Cleveland was able to extract equal value from their lower round prospects.
Perhaps unsurprisingly, the sad-sack Jags finished dead last in the 2008 DPR rankings, and with a ratio of 0.38, it’s not even close. But there’s an unfortunate parallel still to be made: with Dallas’ recent FA disinterest in Victor Butler, and his subsequent signing with New Orleans, the last member of the Cowboys 2009 draft class was wiped from the roster. This development confirms their paltry 2009 DPR of 0.31 is locked in with no further room for improvement. Nothing new, per se, but DPR is just another way to look at how awful that draft really was.
Well, ranking teams by DPR is fun and all, but the endgame was always to find a way to judge the picks themselves. For this purpose, DPR falls woefully short. Choice and Scandrick have a larger DPR than their 1st round counterparts Jones and Jenkins respectively. Were they really that much better players or picks? Of course not.
Therefore, it becomes necessary to consider the value of a pick relative to where the pick was made allowing me to evaluate players across the entire draft on the same scale. That scale is DPI.
[Update] Sir Robin, in the first comment, posted a link to a SportsonEarth.com article ranking the 10 best drafts of the decade. So I went ahead and did the math for each of those classes to give their DrAV, ExpAV and DPR below in the interest of seeing how the numbers matched the article's rankings. I have sorted the table by raw DrAV, which most likely matches the author's intent. Keep in mind the 5 year cap.
Dallas #1! Still too early to judge the 2012 drafts, but they obviously show promise. In terms of value reflected by DPR, Dallas '05 and New Orleans '06 were sparkling drafts. But this table also shows that Indy '03 and Green Bay '09 don't quite deserve to be on this all-decade list. In summary, this is a great example of being able to use DPR to compare draft classes across the years.
Next post, I will discuss the second component of DPI, which weighs a player’s DPR by their draft position.