Draft Trade Value, 1997-2014

by Andrew Healy
On draft night, analytics can help identify the right players... in some cases. In 2013, teams should probably have liked Jamie Collins more and Dion Jordan less. This year, they should probably reconsider how high they draft Jameis Winston. But those cases appear to be the exception rather than the rule. Teams, at least recently, have not shown a consistent ability to beat the draft by finding better players than other teams find. But certain teams have been finding consistent inefficiencies elsewhere in the draft. In a world where draft picks are just lottery tickets, teams like the Patriots accumulated more tickets for years, while teams like the Raiders consistently gave those tickets away.
Here, I estimate how much value teams have acquired and given away in draft picks over the years. The amounts are large, enough to keep draft-value acquirers in the playoffs year after year, and to keep draft value giver-uppers consistently at the bottom of the league. And if the gap between teams in finding good players has largely disappeared, the gap between teams in acquiring draft pick value appears to be only growing larger.
To estimate the draft value teams gained and lost with trades, I looked at the 1,395 picks that were traded between 1997 and 2014. I am not interested in the players acquired with those picks, just the value of the selections themselves, which I estimated with Chase Stuart's chart. His chart just looks at the Approximate Value (AV) that picks tend to generate in the first five years of their careers and does not adjust for salary. Adjusting for salary, as Massey and Thaler did for the old CBA and Brian Burke just did for the new one, would make draft picks look even more valuable, since draftees are cheaper than veterans.
For each team and year, I took the total value of additional picks that teams acquired and compared it to the picks that teams traded away. By looking only at the value of picks rather than the players involved, I am just looking to identify the teams that have traded to acquire draft value, and those that have ended up trading draft value away. At the end, when I compare current general managers, I will also add in the value of compensatory picks. But my primary focus is on the extra draft value that certain teams acquire through trades.
Best and Worst Drafts by Trade Value
|
| |||||
---|---|---|---|---|---|---|
|
| |||||
NYJ | 2000 | 40.8 | NO | 2000 | -38.7 | |
KC | 2008 | 39.7 | BUF | 1998 | -36.9 | |
WAS | 2000 | 38.1 | OAK | 2012 | -31.6 | |
DEN | 2009 | 33.3 | CHI | 2010 | -30.1 | |
STL | 2001 | 32.3 | WAS | 2003 | -28.0 | |
JAX | 1998 | 28.4 | CLE | 2008 | -25.8 | |
STL | 2014 | 28.1 | WAS | 2014 | -25.6 | |
NE | 2008 | 27.3 | TB | 2002 | -24.5 | |
CIN | 2012 | 24.7 | CAR | 1999 | -24.3 | |
NE | 2011 | 24.3 | ATL | 2000 | -24.3 | |
SEA | 1997 | 23.2 | DAL | 2001 | -24.1 | |
PHI | 2010 | 22.6 | WAS | 2007 | -23.4 | |
|
| |||||
TEN | 2004 | 21.8 | MIA | 2002 | -23.1 | |
ARI | 1999 | 21.4 | MIN | 2008 | -22.8 | |
SEA | 2000 | 20.9 | DAL | 2000 | -22.5 | |
MIN | 2005 | 20.5 | NYG | 2005 | -21.8 | |
SD | 2004 | 19.9 | SD | 1999 | -21.4 | |
MIN | 1999 | 19.3 | TB | 1997 | -21.4 | |
SF | 2000 | 19.2 | ATL | 2006 | -20.1 | |
ATL | 2008 | 19.0 | ATL | 2003 | -19.9 | |
OAK | 2003 | 19.0 | OAK | 2008 | -19.2 | |
PHI | 2005 | 18.8 | BAL | 2004 | -19.0 | |
NE | 2004 | 18.7 | WAS | 2004 | -18.4 | |
NE | 2009 | 17.9 | IND | 2014 | -17.7 | |
SF | 2014 | 17.6 | ATL | 2012 | -17.5 |
To start, the left side of this table lists the drafts where teams acquired the most value in picks, according to Stuart's chart. The drafts where teams traded away the most value appear on the right. The most-depleted draft according to this measure was New Orleans' post-Ricky Williams draft (and that doesn’t even include the value the Saints also gave up in the 1999 draft, described in Chris Bouton’s great guest article) where they gave away 38.7 points of AV above replacement level, which Stuart sets at 2 AV per year. To put that in perspective, Calvin Johnson generated about that much AV above replacement in his first five years. So we can think of the Saints as giving away Megatron-like value just in those two 2000 picks they gave up to get Williams.
On the positive side of the ledger, the Patriots appear more often than any other team. The Patriots laid the groundwork for their second act in the 2010s by acquiring picks worth more than 17 AV in 2008, 2009, and 2011. The players the Patriots have picked over the years have, on average, actually underperformed their draft positions by a little bit, but the extra value in the picks the Patriots acquired gave them enough Rob Gronkowskis to balance the Ras-I Dowlings.
Best and Worst Teams Over Five-Year Periods
|
|
| |||||
---|---|---|---|---|---|---|---|
|
|
| |||||
SEA | 43.9 | GB | 31.6 | NE | 55.0 | ||
NYJ | 39.2 | DEN | 29.8 | STL | 52.2 | ||
OAK | 35.4 | CAR | 25.2 | SF | 42.2 | ||
DEN | 30.0 | NE | 22.6 | PHI | 41.7 | ||
SF | 27.0 | STL | 20.4 | CIN | 37.0 | ||
CHI | 26.1 | DAL | 20.2 | DEN | 24.8 | ||
STL | 25.4 | ATL | 18.9 | CLE | 23.4 | ||
CIN | 16.4 | DET | 17.9 | BUF | 20.4 | ||
TEN | 15.7 | TEN | 13.0 | HOU | 15.3 | ||
ARI | 15.0 | PHI | 12.1 | SEA | 10.5 | ||
NE | 13.1 | JAC | 10.9 | MIN | 7.2 | ||
SD | 11.1 | KC | 10.6 | MIA | 7.0 | ||
JAC | 10.9 | MIA | 5.7 | ARI | 2.3 | ||
BUF | 6.2 | BUF | 3.9 | KC | -0.1 | ||
MIN | 6.0 | MIN | 2.1 | PIT | -1.5 | ||
|
|
| |||||
IND | 5.8 | TB | 1.2 | GB | -1.6 | ||
PIT | 4.2 | CIN | 0.3 | NYJ | -2.0 | ||
BAL | 0.3 | PIT | -2.0 | SD | -2.5 | ||
CLE | -0.8 | ARI | -2.4 | NYG | -5.6 | ||
PHI | -0.9 | BAL | -5.0 | NO | -9.8 | ||
DET | -1.1 | NYG | -6.1 | BAL | -10.1 | ||
GB | -7.5 | IND | -9.7 | DAL | -12.1 | ||
CAR | -11.3 | CHI | -10.8 | JAC | -12.9 | ||
HOU | -15.9 | SD | -13.4 | TB | -13.6 | ||
NO | -18.0 | SF | -14.7 | TEN | -13.8 | ||
WAS | -18.2 | CLE | -15.4 | DET | -17.0 | ||
KC | -22.9 | NO | -19.1 | IND | -30.3 | ||
NYG | -23.2 | NYJ | -19.2 | ATL | -32.0 | ||
ATL | -45.1 | HOU | -20.4 | CAR | -36.4 | ||
DAL | -48.9 | OAK | -21.3 | CHI | -39.2 | ||
MIA | -56.4 | SEA | -32.6 | OAK | -47.2 | ||
TB | -61.5 | WAS | -54.3 | WAS | -51.3 |
[ad placeholder 3]
Daniel Snyder's Redskins stand out as having consistently given away draft value. While the most recent negative value might be forgivable due to the understandable Robert Griffin trade -- for a moment, forget Griffin's post-injury struggles and remember how well-regarded he was by both scouts and statistical projections -- the 2005-09 draft value giveaway was not. Oakland likewise has consistently given away draft value. From 2000-04, the Raiders were actually in the top three for acquiring value. Coinciding with their overall decline was a fall into the bottom three of draft value acquisition for each of the last two five-year cycles.
In the last five years, New England and St. Louis have acquired more draft value than any team did either from 2000-04 or 2005-09. The Rams acquired that value mainly through the Griffin trade. The Patriots did it with more methodical maneuvering year after year, continuing this season with the neat trick they pulled off in the Akeem Ayers trade.
The spread in outcomes actually grew wider from 2010-2014. The variance of acquired draft pick value increased by about 17 percent from 2005-09 to 2010-14. If teams are largely similar in their ability to identify good players, they appear to be growing no closer in how they value draft picks.
General Managers and Acquiring Draft Value
Trent Baalke | SF | 2011 | 48.4 | 31.1 | 8.9 | 40.0 | 10.0 |
Howie Roseman | PHI | 2010 | 52.5 | 41.7 | 0.0 | 41.7 | 8.3 |
Bill Belichick | NE | 2000 | 50.0 | 90.7 | 32.1 | 122.8 | 8.2 |
Ted Thompson | GB | 2005 | 32.4 | 30.0 | 27.1 | 57.1 | 5.7 |
Mike Brown | CIN | 1991 | 9.3 | 66.9 | 27.4 | 94.3 | 5.2 |
John Elway | DEN | 2011 | 37.7 | 11.7 | 0.0 | 11.7 | 2.9 |
Ozzie Newsome | BAL | 2003 | 37.0 | -29.3 | 61.7 | 32.4 | 2.7 |
Jerry Reese | NYG | 2007 | 6.0 | 8.3 | 13.2 | 21.5 | 2.7 |
Kevin Colbert | PIT | 2000 | 17.3 | 0.7 | 31.8 | 32.5 | 2.2 |
John Schneider | SEA | 2010 | 51.2 | 10.5 | 0.0 | 10.5 | 2.1 |
Thomas Dimitroff | ATL | 2008 | 27.3 | -8.4 | 19.6 | 11.2 | 1.6 |
Rick Smith | HOU | 2007 | 30.2 | 2.1 | 10.3 | 12.4 | 1.6 |
Martin Mayhew | DET | 2009 | 30.6 | -1.3 | 10.3 | 9.0 | 1.5 |
Jerry Jones | DAL | 1989 | 39.6 | -46.6 | 28.5 | -18.1 | -1.0 |
Rick Spielman | MIN | 2007 | 50.3 | -20.4 | 10.0 | -10.4 | -1.3 |
Mickey Loomis | NO | 2003 | 37.9 | -21.1 | 0.0 | -21.1 | -1.8 |
Bruce Allen | WAS | 2010 | 51.7 | -51.3 | 0.0 | -51.3 | -10.3 |
This comparison includes any general manager who was in place for the 2011 draft and stayed through the 2014 draft. Howie Roseman is thus included in the chart even though Chip Kelly is now the general manager. The chart suggests the very different ideas that Roseman and Kelly may have about the draft. While Kelly gave away a second-round pick in the at-best-highly-questionable Sam Bradford trade, Roseman acquired draft pick value at about the same rate as Bill Belichick has during his tenure in New England. I come up with the average excess value per year that general managers have created across trades and compensatory picks (listed here in the far right column).
[ad placeholder 4]
The top three names on the list all trade like they can't sit still. Through all that action, trading down and into future drafts, they generate the extra value that helped make their respective teams consistently successful during their tenures. Three of the next four names on the list also are moderately active in the draft trade market. Mike Brown, on the other hand, like Jerry Reese, does only a fraction as much wheeling and dealing. While it might be surprising to see Brown rank so highly on the list, most of that value was acquired in the last five years of consistent success for the Bengals.
More generally, the most consistently successful general managers on the list all acquire significantly more trade value than they give away. Ozzie Newsome is the exception, and he makes up for his trades by generating by far the most value in the NFL in compensatory picks.
The best GMs in the NFL have been the ones who continue to stockpile draft value. Where other GMs may be seduced by the sizzle of a flashy running back, the best GMs can get past our natural tendency to undervalue assets when it is unclear exactly what you will eventually get. It would be hard to be less sexy or more vague than "a future fourth-round pick." While that means yawns to most of us, it spells opportunity for the GMs whose teams are the most consistent winners.
Comments
35 comments, Last at 10 May 2015, 12:02am
#1 by 0tarin // Apr 24, 2015 - 6:56pm
While it's long been known that Ozzie loves himself some compensatory picks, I had no idea as to the scope of it. I wonder if this lends some credence to the idea that he's generally a "more successful" drafting GM than many others. An argument could be made that if he's consistently getting higher compensatory picks for players he lets go, he's generally getting more players that are of higher value (or at least perceived value).
Of course, that assumes that these players are originally drafted by him, which I feel like is largely true, but I certainly haven't put any research into the claim.
#4 by DrP // Apr 25, 2015 - 1:38am
Not being interested in the actual player selected is a major flaw. MT and the Harvard draft chart are based on the AV for the AVERAGE player selected at each draft pick.
If the team had a player evaluation as better than average, then the team ought to be willing to pay more to get that player. Conversely if the evaluation is less than average, then the team ought to look at another player or try to trade down.
Further, one should be careful to note that MT says that teams do a good job of evaluating talent overall. They do note that there is a slight problem of picking the best versus the next best player. Yet not being able to differentiate between player A and player B does NOT mean that teams cannot evaluate the difference OVERALL among all players.
This means that when talent is distributed, the players tend to cluster together in groups. In statistics we note effects BETWEEN groups and WITHIN groups.
Those WITHIN groups have similar talent, so it makes sense that it is hard to differentiate the talent level. If they had large differences, then they would be in different groups.
All things being equal, teams should try to maximize their draft points. Yet that is analogous to saying it is good to save money. Then one can spend it on the things one wants.
Draft points are the COST, but the VALUE is in the players actually selected.
I am yet to hear the announcer say the starting lineup consists of +50, -20, +10, etc draft points instead of actual players by name.
#5 by bengreen424 // Apr 25, 2015 - 6:18am
This diatribe looks very familiar...
"Again the MT work and Harvard draft chart are based on the AVERAGE player taken at a specific draft pick. If the team has a better than average player evaluation, they should be willing to pay more in draft points to get him. If the player evaluation of a given player is worse than average, then they should look for another player or try to trade down..."
(http://www.advancedfootballanalytics.com/index.php/home/research/draft/248-special-offer-the-buffalo-bills-visa-card)
I knew I recognized it. That huffy "but the Harvard draft chart says" is hard to forget. The points have all been made and rehashed, though perhaps not so shrilly. From what I can tell, they failed to resonate with most of the football analytics community. Cahse Stuart's model remains the go-to version,
But maybe another slight rewording - WITH EVEN MORE CAPS - would change all that.
#7 by DrP // Apr 25, 2015 - 9:18am
Yes the EMPHASIZED points are made on several sites that I frequent because the authors generally do a good job. Yet they have a fundamental misreading of Massey-Thayer that undercuts their arguments.
It took years for many folks to recognize that trading down was a good thing and GENERALLY it is. I give Massey Thayer a lot of credit. The Harvard draft chart, the new CBA, and the rookie salary schedule reflects their work.
Yet COST in draft picks is only a part of the equation. The VALUE of the SPECIFIC PLAYER must also be included.
Perhaps there would be less snark if there was a better understanding of the difference of the general case and the specific players involved. It took years for the general concepts of MT to be widely disseminated. It will take some further time to learn the broader implications.
#32 by Scott C // Apr 29, 2015 - 1:26pm
Pro Tip:
On the Internet all-caps is considered YELLING AND SCREAMING.
Italics, or bold are considered emphasis.
In fact, the html tag 'em' stands for emphasis and produces italic text, while the html tag 'strong' produces bold.
#9 by jonnyblazin // Apr 25, 2015 - 6:14pm
"Your point doesn't follow logically. If you consistently draft above average, more picks mean more above average players."
Not really, it depends on the pool of players you have to choose from at a given pick. More lower round picks won't help if the pool of players you are selecting from don't have to potential to become above average.
#23 by jonnyblazin // Apr 28, 2015 - 1:07am
"Do you have any evidence anyone is able to do so?"
What a tedious question. I think football evaluation is a skill, and some people are better at that skill than others. For instance, a 2 year old is not as good at evaluating talent as 1980's Bill Walsh.
#24 by Thomas_beardown // Apr 28, 2015 - 1:30am
Good point. Allow me to ask another question. Do you have any evidence that any GM in the salary cap era has been better than average at identifying when a draft is top heavy and trading up to take advantage?
#28 by jonnyblazin // Apr 28, 2015 - 3:28pm
Obviously I don't have teams' draft boards, so I can't access that evidence. But I think it's fairly well established that teams develop tiers of prospects (sometimes "1st round" or just a top tier), and try to accumulate picks in a way that ensures they will have the opportunity to choose from an optimal pool of players. Sometimes that would involve trading up, sometimes trading down.
#21 by ChrisS // Apr 27, 2015 - 12:01pm
In behavioral economics/psychology there is a well known problem that afflicts almost all people, it is known as Optimism Bias. And it causes people to be over-optimistic in their abilities and therefore everyone thinks they are above-average at player evaluation.
#33 by jonnyblazin // Apr 29, 2015 - 3:58pm
First off, that table isn't really that accurate. Ozzie has been in charge of the Ravens draft since 1996.
Second, how is it possible that the Ravens have so many comp picks? They are well known for re-signing the best players they drafted and letting good players they drafted walk. So the only way you can generate a lot of comp picks is by drafting so many good players that you don't have cap space to resign them.
#35 by chemical burn // May 10, 2015 - 12:02am
Well, this makes me think: forget about finding a GM who fits the criteria, is there any criteria that would satisfactorily show a GM is in fact "better at drafting." I'm not sure if anything criteria will be satisfactory. Here are the issues:
Counting the players to simply stick on the roster isn't enough - it's going to be easier to stick to a bad team with tons of holes like the Jaguars than a good with few holes like the Patriots. Just measuring draft success by looking draftees in terms of volume will penalize GM adding to successful teams and reward GM's building up losers.
Measuring Pro Bowls and other faulty accolades would cause GM for teams that traditionally get adequate pro bowl representation like the Cowboys and Steelers to look better than GM's for teams like are traditionally over-looked by Pro Bowl voters - for example, Travis Frederick was not a better pick than Lavonte David, but if we consider only Pro Bowls, the GM who picked Frederick will be judged better than the GM who picked David.
A team's win/loss record has as much to do with coaching, cap management and free agency.
A player's development had as much to do with coaching, especially clear in terms of position groups. For example, you are more likely to be successful as a wr or lb in Pittsburgh or a QB working for Andy Reid than as a CB in Detroit or a QB in Jacksonville. If Andy Reid can't make you look good enough as a QB to flip you for a second rounder, you are hopeless (e.g. you are Mike McMahon) - therefore if you fail as a WR in Pittsburgh (e.g. you are Limas Sweed) you are NOTABLY bad pick in a way that if you fail as a QB in Cleveland you are not. If a GM hands Pete Carroll a CB he can't turn into a starter, that GM has really messed up, regardless of the draft position of the player.
If a player goes on to have a productive career elsewhere apart from the team that drafted him, should them GM be penalized or rewarded? Reggie Nelson was a bust for Jacksonville, but productive for Cincinnati - the who picked him successfully identified a talent capable of playing well in the NFL, isn't that all that can be asked of him?
To me, I think all you can do is look at which GM has few 1st and 2nd round busts who contributed nothing to their team, consistently finds starters in the lower rounds, has built a team with a reasonably consistent amount of success and done so working with multiple coaches (this will unfairly exclude some GM's, but it addresses the issue of coach vs. GM in terms of a team's success and player development.) Newsome mainly fits the bill, although I'm not sure what the threshold for 1st/2nd round busts should be. Polian definitely fits criteria. Bill Parcells as well. So far, so good: some of the greatest GM's qualify... although the criteria is soft, but kinda by design. I'd be curious to know if anyone has something better...
#11 by Andrew Healy // Apr 25, 2015 - 6:40pm
Player-for-pick trades are also included. At some point, I may look into the value of the players included in those trades, too. Here, I'm just interested in teams' tendencies to acquire and trade away draft value
#12 by Jerry // Apr 26, 2015 - 2:33am
I appreciate what Andrew's trying to do here. The one thing that occurs to me, though, is that very few draft pick(s) for draft picks trades happen in a vacuum. They happen during the draft when a team is looking to move up and take a particular player. So an alternative method of valuation would be to replace the estimated AV of the first pick with the AV of the player selected there. If the team moves up one pick, then the AV of the player selected with the following pick can be used, too, since both teams knew exactly who they were going to draft with those picks. Once another team is interspersed, going back to estimated AV makes more sense. Even using the drafted player's AV may not be perfect, since a team could be looking to fill a specific hole, or the best choice available at that point could be worse than the estimated value of the pick, but that's making things much too complicated.
#15 by Andrew Healy // Apr 26, 2015 - 7:14pm
Good point. I think there's something to that. And it's worth checking out. Clay Matthews is one example of trading up clearly being a good idea. And I think trading up to try RG III was a good idea at the time for WAS given what a franchise QB is worth. My guess is this wouldn't change much except maybe for franchise QBs. For every Clay Matthews that teams trade up for, there's probably a Mark Sanchez where the trading up didn't work out. But that is just a guess right now.
#17 by mehllageman56 // Apr 26, 2015 - 9:11pm
Trading up for Darrelle Revis turned out to be a great idea as well, Too bad two years later the Jets took that trading up idea and shot it with steroids, so that their entire 2009 draft consisted of 3 players. By the way, the Jets only gave up their first and second round selections in that year to move up to the fifth pick, throwing Kenyon Coleman, Abram Elam and the preseason immortal Brett Ratliff in as well. So the trade itself wasn't as bad as the actual pick.
#20 by Thomas_beardown // Apr 27, 2015 - 11:59am
I think the analysis of franchise QBs is wrong. Yes, they are very valuable, but you have to account for risk. There is no way your scouting department is so good that you can justify 3 first round picks (at least one of which is top 10 and more probably will be) on that risk. Washington could have sat tight at 6th and drafted Tannehill and had all those picks to help surround him with talent.
#14 by Bright Blue Shorts // Apr 26, 2015 - 9:54am
Would be interesting to see the average draft position for teams included, to see whether that influences how good teams are at this?
- Oakland who are listed 3rd best for 2000-04 were one of the best teams in the league for the first four of those years. When they declined they apparently became bad at getting value.
- On the other hand, New England apparently trade well but have usually been drafting bottom four.
#16 by Andrew Healy // Apr 26, 2015 - 7:21pm
My guess here is that the causal direction goes the other way. Poorly-run teams, like the Raiders towards the end of Al Davis's run, trade away draft value. Losing that draft value hurts the team in the following years. But it's hard to separate the impact of losing draft value from the other impacts associated with being poorly run. In other words, trading away picks willy-nilly is as much a symptom of mismanagement as a cause of future losses.
#18 by Independent George // Apr 27, 2015 - 9:54am
Well-run teams also typically have fewer holes to plug, and therefore are less inclined to trade up to get a specific player rather than taking the best player available and then finding a way to use him.
#19 by nickbradley // Apr 27, 2015 - 11:48am
Great Analysis - thanks. You should probably take a look at actual player selection though. I took a look at draft effectiveness since 2010, and while Trent Baalke -- whom you have as #1 on this chart -- has made some brilliant moves, the picks he actually made were kinda lousy. Maneuver all you want to get an extra second round pick, but if you blow it in Vance McDonald, well...
http://www.ninersnation.com/2015/4/26/8500673/was-the-2012-49ers-draft-the-leagues-worst-this-decade