Writers of Pro Football Prospectus 2008

26 Apr 2012

Which Team Has the Most Total Draft Pick Value in 2012?

Most everyone knows about the infamous draft trade value chart, which at this point is pretty much outdated, and universally ridiculed. Fewer people know that Doug Drinen of Pro Football Reference fame updated it back in April 2008. Even fewer people know that I extended this idea of basing draft pick values on "expected career approximate value" as part of a series of posts on Niners Nation in April 2010.

The idea is that, based on an analysis of post-merger NFL draft history, each slot is reliably associated with a given amount of expected career approximate value (AV). Therefore, if we take each teams allocated draft slots, and add up all of the expected AVs for those slots, we end up with a total expected AV for each team's entire draft.

I went ahead and applied this concept to the 2012 draft (as the picks stand right now), and came up with the table below. As the table shows, Cleveland -- mostly because they have 13 picks -- are expected to get the most value from the 2012 draft. On the other end of the spectrum, the scandal-plagued Saints are going to need a value-maximizing miracle in order to come away with a meaningful draft class this year. The beauty of all this, of course, is that we can evaluate -- say, six years from now -- how well each team did in relation to expectations.

(Of course, as I was writing this, the Browns dealt three of their picks. That trade is not included in the table below.)

Discuss.

Team No. of Picks Total Expected AV
CLE 13 25.96
IND 10 21.61
MIN 10 21.41
CIN 9 19.36
BUF 10 19.12
STL 8 19.08
PHI 10 19.04
GB 12 18.09
MIA 8 17.87
WAS 7 16.42
JAC 7 16.04
DAL 8 15.96
NYJ 10 15.91
KC 8 15.86
NE 6 15.84
PIT 10 15.37
HOU 8 15.03
TB 6 14.84
SD 8 14.84
DEN 7 14.30
CAR 7 14.24
CHI 7 14.04
BAL 8 13.77
TEN 7 13.70
SEA 6 13.59
DET 7 13.18
ARI 7 12.85
NYG 7 12.62
SF 7 12.52
ATL 5 7.79
OAK 5 7.14
NO 5 5.82

Posted by: Danny Tuccitto on 26 Apr 2012

44 comments, Last at 02 May 2012, 9:49pm by Pat (filler)

Comments

1
by Danny Tuccitto :: Thu, 04/26/2012 - 8:09pm

After the CLE trade, here's an updated total draft value table:

MIN 13 25.08
CLE 10 22.30
IND 10 21.61
CIN 9 19.36
BUF 10 19.12
STL 8 19.08
PHI 10 19.04
GB 12 18.09
MIA 8 17.87
WAS 7 16.42
JAC 7 16.04
DAL 8 15.96
NYJ 10 15.91
KC 8 15.86
NE 6 15.84
PIT 10 15.37
HOU 8 15.03
TB 6 14.84
SD 8 14.84
DEN 7 14.30
CAR 7 14.24
CHI 7 14.04
BAL 8 13.77
TEN 7 13.70
SEA 6 13.59
DET 7 13.18
ARI 7 12.85
NYG 7 12.62
SF 7 12.52
ATL 5 7.79
OAK 5 7.14
NO 5 6.82

2
by Theo :: Fri, 04/27/2012 - 4:51am

I'm afraid you're 2 days to quick with this.
Why don't you hold your horses and post it when the draft is over?

3
by Theo :: Fri, 04/27/2012 - 7:25am

I didn't watch it yesterday, but Cleveland moves up one spot for a 4th, 5th and 7th with a team that wasn't going to pick their projected player anyway??
That only makes sense if they expected Minnesota to take Richardson and if they weren't happy with Kalil on the 4th spot.

4
by Temo :: Fri, 04/27/2012 - 7:41am

There may have been another team attempting to trade with the Vikes to get Richardson. Or maybe it was just the threat of such, probably played up by the Vikings management to get Cleveland to pay up.

Either way, excellent job by the Vikings.

5
by dryheat :: Fri, 04/27/2012 - 7:57am

Reportedly the Vikings were playing the Browns and Bucs against each other, as both wanted Richardson. Good job by the Vikings and the Browns.

8
by Kulko :: Fri, 04/27/2012 - 9:04am

Wouldnt that be good job by the Vikes and Tampa. The woinner of an Auction is always guaranteed to have lost value.

11
by DEW (not verified) :: Fri, 04/27/2012 - 10:48am

How about good job Vikes, since the loser of the auction, by definition, doesn't get what they actually wanted in the first place. The Browns overpaid and the Bucs had to settle for Plan B...whereas the Vikings get a handful of draft picks and still get to draft the guy they wanted in the first place.

13
by tuluse :: Fri, 04/27/2012 - 11:04am

If Tampa really wanted to trade up to the 3rd pick to draft a running back, they won without knowing it.

25
by JIPanick :: Sat, 04/28/2012 - 7:52pm

Yeah, no kidding.

26
by Whatev :: Sat, 04/28/2012 - 7:58pm

Sure, but winning because your rivals fell flat on their faces is hardly cause for congratulations.

28
by Mr Shush :: Mon, 04/30/2012 - 6:27am

I know this discussion is ongoing elsewhere, but I really don't think that's automatically true now that rookie salaries are slotted (and indeed rookie contracts shorter, meaning that the ability of RBs to play at a high level as rookies should be more highly valued).

14
by dryheat :: Fri, 04/27/2012 - 11:12am

If Cleveland really wanted Richardson, giving up a pittance to ensure you're getting him is a win.

Likewise, if Richardson was #1 on Tampa's board, it's bad management by Tampa, assuming the trade demands were about equal and the Vikings would be okay dropping to 5.

If Tampa preferred Claiborne (probably not, given the trade down) or Kalil as much or more than Richardson, it's a win-win-win situation.

15
by commissionerleaf :: Fri, 04/27/2012 - 12:50pm

Since I view taking a running back anywhere before about #25 as basically wrong, period, obviously I think Cleveland got hosed here. Running backs not of McFadden/Peterson talent are fungible, and even those players are less valuable than high-end linemen or defensive players. And it isn't like Cleveland didn't need Blackmon every bit as much as they needed Richardson.

Tampa probably should have taken Claiborne instead of trading down, but they are positively lucky they did not get this deal.

19
by dryheat :: Fri, 04/27/2012 - 1:49pm

Well, that's an altogether different discussion. I wouldn't have drafted Richardson there either. That, however, has nothing to do with Cleveland's manipulation of the draft board in order to get the man they wanted.

17
by Aaron Brooks Go... :: Fri, 04/27/2012 - 12:53pm

Likely, yes.
Guaranteed, no.

29
by Mr Shush :: Mon, 04/30/2012 - 6:30am

Only if you assume that the values of the commodities traded are stable. The Browns had more draft picks than they could realistically fit on their roster, so those late round picks were worth less to them than to the Vikings. Richardson was (in their opinion and mine) worth more to them than any player they could have taken at #4, but less to the Vikings than Kalil (thanks to Joe Thomas and Adrian Peterson). This trade was a win-win, as most trades ought to be - and most trades not involving the Raiders are.

7
by Theo :: Fri, 04/27/2012 - 8:46am

Ah yes. Of course.
Thanks.

27
by db :: Sat, 04/28/2012 - 11:08pm

Or if they thought that another team would make the trade. Yes?

6
by Ewout (not verified) :: Fri, 04/27/2012 - 8:05am

As far as I can tell the "Approximate Value" metric is descriptive, it describes how well a player has done in the past.
Is there any evidence that this metric is predictive of team success?
Is there any evidence that past "Approximate Value" of players on a team is a better predictor of team success than for example last season's wins?
If not then I'd say this AV-based draft value chart should also be 'universally ridiculed'.

9
by tuluse :: Fri, 04/27/2012 - 10:18am

Is there any evidence that having more or higher draft picks is a better predictor of team success than the last season's wins?

This chart is trying to measure how much value a team has in draft picks, not how a team is going to do the next year.

16
by Ewout (not verified) :: Fri, 04/27/2012 - 12:52pm

The only relevant measure for how much value a team has in draft picks is by how these picks will influence future results.
How can you argue that those two are unrelated?

The only way to decide which draft value table is better is to correlate them with future team results. Saying that an approximate career value based chart is better than the Jimmy Johnson chart (or any other chart) without supporting this with relevant data makes it merely an opinion. And everybody has an opinion...

You could for example argue that better players stay in the league longer, so in terms of approximate career value they get a double bonus. But a long career doesn't benefit the picking team, as long as the guy serves out his rookie contract. So you could argue that an approximate career value based draft table undervalues the later picks and that the Jimmy Johnson chart is a better fit.
To quote the author of the article: "AV is flawed in several ways".

At least the draft pick value chart the NFL guys currently use was subject to some evolutionary pressure, so it won't be far off. Before dismissing it we might want to try and understand it first.

18
by tuluse :: Fri, 04/27/2012 - 1:13pm

The value in draft picks is by the potential for changing performance in the future not by actual change.

"At least the draft pick value chart the NFL guys currently use was subject to some evolutionary pressure, so it won't be far off. Before dismissing it we might want to try and understand it first."

It was also made in the early 90s, think things might have changed since then?

22
by Intropy :: Fri, 04/27/2012 - 3:32pm

Yes things have changed. The biggest I can think of is rookie pay. Until basically right now, I think the chart had too steep a discount especially near the top of the draft.

23
by tuluse :: Fri, 04/27/2012 - 3:39pm

The salary cap a whole has changed making all draft picks more valuable because teams can't always afford to keep their veterans.

Those are just the big easy things to identify. There has been 20 years of societal change as well.

24
by Intropy :: Fri, 04/27/2012 - 4:59pm

Totally agreed. I'm just thinking that we're actually seeing the chart get closer to the right valuation as compared with, say, five years ago. Versus 20 years ago? So much has changed.

10
by Joseph :: Fri, 04/27/2012 - 10:47am

The only problem I have with this (I read Doug's post back then) is that it only takes one player to make the avgs. look ridiculously small. For example, for my Saints, it only takes 1 player starting most of the games during his rookie contract to get that AV, regardless of whatever the other players do.
Even for Indy (#3 overall), if Luck is 1/2 the player they think he'll be, he should meet that projection (or at least I think so, from what I remember from AV). If we were to compare this chart in ~5 years, the Colts' draft would look to be a "success", as it met the baseline. IMO, if Luck is only average, and the Colts get "nothing else" out of this draft, they will still be drafting in the top-10 for the next couple of years. However, if Luck turns out to be a top-10 QB, and they get a couple of other starters, and the lower round picks at least contribute on ST for 2 or 3 yrs, THEN this will be a "successful" draft.

12
by tuluse :: Fri, 04/27/2012 - 11:03am

Yeah, the main lesson here is that a huge percentage of draft picks are useless.

It would be interesting to see the standard deviations.

20
by AnonymousA (not verified) :: Fri, 04/27/2012 - 2:59pm

There's a very real chance that draft pick value is not a normal distribution. You can measure the standard deviation, and it's likely to be pretty high, but that's sorta like answering "how much is this car worth?" with "well, it's 17 feet long!"

A histogram of the data would be interesting, partially because it would reveal the shape of the distribution.

21
by Ewout (not verified) :: Fri, 04/27/2012 - 3:08pm

Danny, help me out here. What exactly is you draft value table based on?
Is it "expected career approximate value" or "Average Weighted Season AV" or none of the above?

The posts on Niners Nation that you link to are using the averaged measure. Doug Drinen's article doesn't use expected career approximate value either, but introduces some weighted average.

I'm a bit confused. I mean, I've got a bone to pick with all of them, but I'm just not sure what exactly to aim my arrows at :-)

30
by Pat (filler) (not verified) :: Mon, 04/30/2012 - 10:11am

I don't know why this hasn't been brought up in terms of a draft chart, but the basic idea here is a bit flawed. For the sports statisticians it should really feel like when people were comparing DVOA in a small sample size versus DPAR/DYAR. The problem is that you have to remember that a draft pick costs a roster spot, and that roster spot could be filled with a replacement-level player.

It's a lot like the reason why businesses are choosy about hiring: hiring the wrong person costs you money, because you had to train them, and waste resources on them, and they didn't produce any value for you.

In the same way, a 7th round pick doesn't cost you nothing. It costs you a roster spot, which could've been filled with someone else with more value.

Most people who've attempted to alter the "NFL trade value chart" have always neglected to take into account the cost in terms of a roster spot of the player.

A better way to do this would probably be something like:

1) Calculate the AV distribution on a typical NFL team. That is, calculate AV for each team, rank them, average, etc. Basically what you want is to know how valuable the 'bottom' roster spot is, the 'next to bottom' roster spot, etc.

2) Take the number of draft picks that each team has, and add up the AV for an equal number of roster spots. If you've got 10 draft picks, you add up the bottom 10 roster spot values. This is the immediate cost of the draft class.

3) Calculate the average number of years retained for each draft pick (with a maximum of 5), and add up the number that are kept for 1 year, for 2 years, for 3 years, etc. Repeat the above analysis and sum. (The reason you limit the 'cost' to 5 is that anyone retained for more than 5 years is no longer on their rookie contract, and therefore was chosen to be kept by the team). This is the future cost of the draft class.

4) Add career AV for each draft pick. Subtract total cost (immediate + future).

There are a few things you can immediately see about this kind of analysis that are much better than existing methods: First, draft picks are not additive in value. You can't draft 100 7th round picks, because they end up taking up your entire roster, which is dramatically more valuable than 100 7th round picks.

Second, early-round picks are worth much more than their career-AV expectation because they only take up 1 roster slot, and any career AV they accumulate past their rookie contract has basically no cost to the team.

The "draft picks are not directly additive" is something that all of the teams know instinctively (teams in general are not willing to take a bajillion late round picks for an early pick), but no one in sports analysis has taken into account yet.

31
by foolrider :: Mon, 04/30/2012 - 12:23pm

These types of comments are why I read the comments on FO

32
by RichC (not verified) :: Mon, 04/30/2012 - 3:00pm

"In the same way, a 7th round pick doesn't cost you nothing. It costs you a roster spot, which could've been filled with someone else with more value."

Nitpick here, but a very large percentage of 5th/6th/7th round picks never make the roster. They only make the roster if they prove in training camp that they're better than a typical 5th/6th/7th or replacement player: IE, they don't cost anything unless they MAKE the roster, and if they've made the roster, they've generally already proven to have some value (be it special teams or whatever).

Also, I think the roster spot concern is only an issue for about half the teams in the NFL. There are plenty of teams that have guys on the roster who are replacement level or below (like the colts right now).

33
by Dean :: Mon, 04/30/2012 - 3:32pm

Not to nitpick on your nitpick or anything, but those 5/6/7th round picks who don't make the roster still cost probably ten grand or so in dead money from any miniscule signing bonus they get.

34
by RichC (not verified) :: Mon, 04/30/2012 - 5:16pm

I could be wrong, but I don't think most of those guys actually get signing bonuses.

And even if they do, they don't count against the cap because they won't fall in the "top 51" rule.

36
by Pat (filler) (not verified) :: Mon, 04/30/2012 - 6:01pm

Signing bonuses always fall under the cap once the season starts. It's money that gets paid out, period.

35
by Pat (filler) (not verified) :: Mon, 04/30/2012 - 6:00pm

What you'd do there is just take how often a 7th round pick makes the roster as the effective "number of players" that the 7th round pick takes up. In other words, if you've got 4 7th round draft picks, and 30% of them make the roster, they effectively count as 1.2 players. That sort of thing.

However, there is one other thing to consider which is that it also matter what position they play - you can't practically keep more than 3 QBs, for instance, so drafting 3 7th round QBs in the hopes of finding Tom Brady is pretty pointless - you don't have the resources to evaluate them.

IE, they don't cost anything unless they MAKE the roster

They still do take up a cost, but unfortunately it's a lot harder to measure. They still take an 80-man roster spot no matter what, for instance.

I think in the end the cost to the "training camp resources" is probably bigger than the roster spot, but I can't think of any easy way to measure that. In any case, you have to take into account the opportunity cost of drafting a player. Otherwise you get bizarre results like the Massey/Thaler paper. The roster constraints on a team (especially if you think about it by position) are strong enough that I'm surprised that more teams don't try to consolidate picks to move up.

37
by Intropy :: Mon, 04/30/2012 - 8:48pm

You also need to consider the average utility vs draft position curve, which differs by position as well as the fact that you do not know which prospect will be the best player. Say you need a corner, but only one corner, and you only have one roster spot available. Now maybe the obvious strategy is to trade your entire draft for the #1 overall pick and then take the best CB prospect. Or maybe you do best if you stand pat and draft the best CB with each pick and take the best of those 7 once you've done better judging. Or maybe you trade back an pick 20 CBs later in the draft and then take the max of those.

38
by Pat (filler) (not verified) :: Tue, 05/01/2012 - 10:03am

No, see, that's the point: you can't take 20 CBs later in the draft and take the best of them. You simply don't have that option - you don't have any way to evaluate 20 CBs during training camp, and so in the end, you'd probably take the ones you had the highest draft grade on, in which case... why did you draft the others? Oh, and now you also didn't have 19 other players to evaluate.

Teams don't typically draft more than about 10 players: instinctively, the GMs realize that they've got limited resources and do the best they can with those. The problem is that most draft value charts don't take that into account.

39
by Intropy :: Tue, 05/01/2012 - 8:49pm

20 is obviously a high number and meant to illustrate the concept. Replace it with five, which is a totally reasonable number to evaluate. The point stands. Picking the best prospect is not obviously more likely to yield better average results than picking the max of 5 lesser prospects of various ability given the margin of error on measuring prospects.

41
by Pat (filler) (not verified) :: Wed, 05/02/2012 - 9:42am

I disagree. First off, I think the idea that you can make a good enough evaluation in 1 training camp of rookies of equal pre-draft grades to reliably pick out the one that will have the highest career AV is almost certainly wrong. This could be determined from data pretty easily, though. How often does a team release a player who turns out to be better than a player they retained?

Imagine that you move down from a round 3-4 pick and pick up 5 round 7 picks, with the intention of keeping only the best one. You would've filled those extra 4 slots anyway with UDFAs, which cost you no draft resources.

So you had two options:
1) Choose 1 player 100 slots earlier, and 4 players 16 slots later. Take the best one (likely the player 100 slots earlier).
2) Choose 5 players.

You can't have a draft value system that values bundling up a bunch of late round picks for a higher pick. It just doesn't work. There's basically no marginal value over UDFAs to the very late round picks, and if you don't take into account the fact that for each draft pick you add (i.e. go from 5 total draft picks to 6) you take someone away, then you're just not valuing the pick properly.

And for the higher draft picks (like rounds 3, 4, 5) the cost of that roster spot becomes very real, especially if you treat it by position, because the team will almost certainly hold onto the player into the season (and over multiple years).

40
by Joseph :: Wed, 05/02/2012 - 1:45am

Pat,
IMO, teams that are good (playoffs for at least 2 of the last 3 yrs.) have fewer roster spots "available", and thus should try to consolidate their draft picks to get "impact" players at "need" positions, while teams who "need a lot" should try to draft as many different players as possible. The problem is that we all know that only about 1/2 of all players drafted actually stick for more than 2 yrs. So, those good teams risk drafting players who actually don't do much for them by putting all their eggs in the same basket, so to speak. If you draft 10 or more players, the chance that 3 or 4 become at least solid contributors is pretty good, and a couple of others may play for a year or two.
One example that stands out to me is the 2006 Saints. They drafted Bush in the 1st round, Antonio Freeman in the 4th, and Pierre Thomas was an UDFA. Three RB's, and PT has played more like a 1st round pick, Bush like a 4th rounder, and Freeman like the UDFA. (If you don't remember Freeman, it's because he got cut because of PT--then he went to the Rams, and I don't think he played more than 2 yrs.) In other words, more players drafted = the chance to evaluate players better, and the chance to continually improve the ~10 worst players on your roster (or at least replace them with someone younger, cheaper, and with possibly more potential).
IMO, since the best teams usually have better depth, they are always looking to improve those "worst" players on their roster, and thus are always looking for a high-ceiling UDFA who may turn out to be something. They probably go to camp with 45 veteran roster spots set in stone (unless there is a season-ending injury to a presumed starter/key backup), 5-6 for draft picks, and 2-3 for UDFA's/other team's camp cut/late trade. IMO, if you have less than 45, you prob. don't have enough talent on the roster; if more, you prob. don't have enough youth, and are going with late-career vets making the minimum over guys you can develop who may actually do something for your team for the next 2-3 yrs.

42
by Pat (filler) (not verified) :: Wed, 05/02/2012 - 10:10am

So, those good teams risk drafting players who actually don't do much for them by putting all their eggs in the same basket, so to speak. If you draft 10 or more players, the chance that 3 or 4 become at least solid contributors is pretty good, and a couple of others may play for a year or two.

If you draft 10 players, you have to shove room for them on your roster. That's a lot of players. Certainly, for the next year, the middle-round players will add negative value to the team - they're almost certainly going to be worse than an available veteran. The late-round players who make the team also probably will be about the same. In the long term, they may add value, but this is, in some sense, the point: you can't say "a draft pick is worth X value" without taking into account what adding that draft pick costs.

In fact, that kindof argues for 2 draft charts: an "immediate value" chart and a "long term value" chart. The career AV added by players is dramatically different by what round they're in: a 1st round pick has a good chance of adding value to the team this year. A fourth round pick does not, but he may add value in years 2 or 3. This likely would vary by position as well, but all of these things could be done in a better draft value chart.

The problem is just looking at the collection of draft picks that a team has and trying to add up all their value blindly. Even ignoring that the teams have different strengths, different number of roster spots available, etc., a team that drafts a bucketload of 3rd round picks may end up with a lot of "value" by a simple chart, and then they'll go 0-16, the coach will be fired, and most of those 3rd round picks will stagnate as things go crazy.

That's an exaggeration, right, but hopefully it makes the point - it's all stemming back to the fact that you can't just assume that 4 players sum up to 1 player, even with equal career AV. You have to take into account the additional roster cost first, and then possibly any distribution differences between when the career AV comes.

In other words, more players drafted = the chance to evaluate players better,

Right, but your example wasn't quite what I was talking about. First, a 1st round RB is just insane nowadays. I don't know why it still happens. Running backs have short careers and stars come out of a UDFA not dramatically less than a 1st round pick. Second, you've got 3 players, a high-round, a mid-round, and a low-round.

I'm talking about the cost of forgoing a high-round pick in favor of many low-round picks. I don't know why it isn't obvious that this strategy doesn't work. First, it pretty much doesn't happen in the NFL. Second, when it does, it's usually a disaster. Effectively, this is what the Redskins were doing in the mid-2000s: they were constantly trading away picks to move up and so averaged about 5-6 draft picks/year. Now, you might say "WTF, that's exactly the opposite of what you're saying" - but really, it's not. Effectively they were left with 1 high round pick, and lots of late round picks and UDFAs. I don't believe this strategy makes sense.

I'm not suggesting trading all players away for a single high-round pick is what you should do. Taking into account roster considerations means that there's likely a roughly optimal number of draft picks an average team should take, and then of course the highest value is when those draft picks are all highest. Taking fewer players (trading away to get a high draft pick, for instance) probably won't get you enough return to make up for the players you traded away (and replaced with UDFAs) and taking more players (with average lower draft picks) means you likely won't be able to retain enough of them to reap the long-term benefits.

43
by tuluse :: Wed, 05/02/2012 - 11:12am

In fact, that kindof argues for 2 draft charts: an "immediate value" chart and a "long term value" chart. The career AV added by players is dramatically different by what round they're in: a 1st round pick has a good chance of adding value to the team this year. A fourth round pick does not, but he may add value in years 2 or 3. This likely would vary by position as well, but all of these things could be done in a better draft value chart.

This seems pretty insightful. Most people only value draft picks on the long term value, or at least only take into account risk and possible value.

There is another question, which is how many players can a team afford to keep on the roster that provide little to no value? This is team specific too. If you've just gone 2-14 and need to add talent at any cost the answer is probably higher than a team that's 10-6 and trying to make a playoff run.

44
by Pat (filler) (not verified) :: Wed, 05/02/2012 - 9:49pm

It's absolutely team specific, but you could at least estimate it based on the average AVs at the bottom of the roster for all teams. That's basically what I was suggesting - a "perfect" method would be team dependent, but in some sense you kindof need a baseline anyway.