Does Crime Pay in the NFL Draft?

My first article in this series, "AFC North Drafting, 2001-2012: An In-Depth Look," compared the AFC North and the New England Patriots to the league average in two areas: how good a job they did choosing players, and how well they developed the talent they drafted. You can click the link above to find out what conclusions I came to. But there are a lot of other questions one can explore with the data I compiled.

One of the things I've always wondered when draft discussions inevitably go to the question of trading up or down is, how much sense does either strategy make? I ran across an article discussing whether in the new era of reasonably-priced rookie contracts Bill Belichick, considered the king of NFL horsetrading, will change his strategy of trading down for more picks and, if anything, trade up for less of them.

The author supported the supposition by noting Belichick had actually traded away a couple of picks in the 2012 draft to move up. However, the article was written before this year's draft, and Belichick was back to his normal form, trading down to grab more picks. (Regrettably I didn't bookmark the article, and naturally I can't find it to save my life.)

But I think characterizing Belichick in this way is far too broad a brush to paint with. I decided to try and determine how much he trades, and how effective his trading strategies really are. Let's look again at the chart from the previous article, but just with the league average and New England's performance in it:

New England's best drafts by far since 2001, compared to the league average, are 2003, 2005, and 2010. (It's probably still too soon to properly evaluate 2010, so I'm going to confine myself to '03 and '05.

What I found was mind-boggling. It took me a couple of hours to sort out what happened in the 2003 draft. I expect many teams felt the same sense of "what just happened?" as I did.

The Patriots began the draft with the following picks, just for being the Patriots, as it were: Nos. 19, 50, 81, 120, 157, 193, and 234. Of those picks, they actually only used Nos. 120 and 234, and even No. 120 went around the block a few times.

In addition to these seven picks they had acquired the following picks from other teams: No. 14 (traded Drew Bledsoe, QB, to BUF for their '03 1st round pick); No. 140 (part of a 2002 trade with DAL); No. 154 (traded Greg Randell, OT to HOU in 2002); No. 225 (traded Grant Williams to Rams); Nos. 78, 239, and 2004 fourth-round pick (traded Tebucky Jones to NO); and No. 128 (traded Terry Glenn to the Packers in 2002.) They also picked up selections in most of the player trades.

The Patriots then began the draft with 14 selections, as follows:

Round 1: 14, 19

Round 2: 50

Round 3: 78, 81

Round 4: 120, 128

Round 5: 140, 154, 157

Round 6: 193

Round 7: 225, 234, 239

There were a number of complicated trades, including the one in which the Patriots traded picks Nos. 50 and 120 to the Panthers to move up to No. 45 in the second round, and then traded a bunch of stuff to the Broncos to get No. 120 back. Here's what they ended up with (original picks in brackets—no comp picks):

Round 1: [14, 19] 13

Round 2: [50] 36, 45

Round 3: [78, 81] No pick

Round 4: [120, 128] 117, 120

Round 5: [140, 154, 157] 164

Round 6: [193] 201

Round 7: [225, 234, 239] 234, 239, 243

This may not look like they did particularly well. They purveyed 14 picks, including two first-round picks, into 10 picks, with almost a third of those being seventh-rounders and only one first-round pick. However, they also got the Raven's 2004 first-round pick, the Dolphin's 2004 second-round pick, and the Redskin's 2004 fourth-round pick.

But what makes this draft particularly strong is the guys they chose. The No. 120 pick they re-acquired was used on Asante Samuel, whose Weighted Career Average Value, as per Pro Football Reference, is 72. The league average for all rounds added together in 2003 is 130.5. They also did well with their fifth-round pick (Dan Coppen) and with one of the seventh-round picks (Banty Tulla-Cain.) While the first and second round picks were not legendary, they were better than average, except the first second-round pick. The only pick who didn't pan out at all was the sixth-round pick. As a result, the Patriots accumulated a total WCAV for their 2003 draft class of 253, almost twice the league average.

Ironically, 2004 wasn't a fantastic draft for the Patriots, despite all those extra picks, but it was above average. So what about 2005?

They began with only the seven picks everyone gets, beginning at #32, and third, fifth, and seventh round comp picks. Prior to the draft they traded the third-round pick, No. 95, to the Cardinals for a player and pick No. 145.

Here's what they ended up with, after a good bit more horse-trading. Original pick is in brackets—comp picks in italics:

Round 1: [32] 32

Round 2: [64] no pick

Round 3: [95, 100] 84, 100

Round 4: [133] 133

Round 5: [168, 170 ] 170

Round 6: [206] no pick

Round 7: [246, 255] 230, 255

Again this doesn't look impressive. They began with 11 picks, plus the player (Duane Starks) they got from Arizona. They ended up with seven picks, and only the third round and seventh round picks were higher than what they had in the first place. But they also, true to form, acquired three 2006 picks—the Ravens' third-round pick [75], the Lions' fourth-round pick [106], and the Raiders' fifth round pick [136.]

Fortunately for the Pats they struck pay dirt on almost every pick. The only complete busts were the fifth and seventh round comp picks. Pick No. 230 was Matt Cassel. Pick No. 100 was Nick Kaczur. The player they picked up from Arizona, DB Duane Starks, didn't produce a lot in his year in Boston, but he wasn't a total loss either. The pick they got in that trade (145) was parlayed into Detroit's 2006 fourth-round pick. All in all, an interesting argument for trading.

So let's look at the two years on the chart above which were distinctly below average—2006 and 2007. Did Belichick change his trade strategy? We already know he began 2006 with three extra picks. Here are the original picks they had, including the picks traded for in 2005:

Round 1: 21

Round 2: 52

Round 3: 75 (from Ravens in 2005), 86

Round 4: 106 (from Lions in 2005), 118

Round 5: 136 (from Raiders, 2005), 152

Round 6: 191, 205, 206

Round 7: 229

You know the drill—that's almost certainly not where they picked, for the most part. That is 12 picks. What did they end up using? (If different from actual pick, original picks in brackets. Comp picks in italics.)

Round 1: 21

Round 2: [52] 36

Round 3: [75, 86] 86

Round 4: 106, 118

Round 5: [136, 152] 136

Round 6: 191, 205, 206

Round 7: 229

That's right. Astonishingly, the Patriots made a single trade—they traded their second-round pick and the third-round pick they had received from the Ravens for the Packers' second-round pick.

The Packers took Greg Jennings at No. 52, the first pick the Patriots traded away, and a reasonable if not amazing DB at 75 with the second Pats pick. The Patriots picked Chad Jackson, a WR who can fairly be called a bust. For once, the trade partner with the Patriots had the last laugh. The 2006 actually wasn't a bad draft, despite the first-round pick, a running back, not being fantastic and the second round being a dud. The problem for New England came because so much of the "value" of their picks was accumulated on other teams. In fact, 49% of the total WCAV for the drafted players was earned on other teams, as compared to the league average for 2006 of 81%.

Another problem in assessing this draft is that one of the fourth-round picks was kicker Stephen Gostkowski. As noted elsewhere, for reasons which completely elude me, PFR refuses to assign any value whatsoever to kickers and punters. Sebastian Janikowski, the Raiders' much-derided first-round pick back in 2000, has probably scored more points for the Raiders than any other one player since that time. For all I know, he's scored more points than any single position group, such as quarterbacks. His WCAV is 0. It makes no sense. In Gostkowski's case, he's still on the team, and has a lifetime 84.2% field goal conversion rate, which is actually better than Janikowski's.

So I'm not entirely certain it's fair to call this draft a flop, as we look at it more closely, although it certainly isn't one of Belichick's "value" drafts. It perhaps didn't help the Patriots as much as it might have, either.

How about 2007? Two below-average drafts in a row are pretty astonishing for the Patriots (in fact, 2008 wasn't very good, either.) This wasn't just below average, though—it was legendarily bad. So how did it go down?

The Patriots originally had the following picks:

Round 1: 28

Round 2: 60

Round 3: 91

Round 4: 127

Round 5: 165, 171

Round 6: 202, 208, 209

Round 7: 238

We see it is the usual double-digit number of picks for the Patriots. Here's how they used them. (If different from actual pick, original picks in brackets. Comp picks in italics.)

Round 1: [28] 24 (from the Seahawks, for Deion Branch)

Round 2: [60] nada

Round 3: [91] also nada

Round 4: 127

Round 5: [165, 171] 171

Round 6: [202, 208, 209] 180, 202, 208, 209

Round 7: [238] 211, 247

Since they had one first-round pick they traded away the other to the 49ers for their fourth-round pick [No. 110] and 2008 first-round pick [No. 7]. I don't know why teams even take the Patriots' calls... And speaking of that, the Dolphins then took the Patriots' second and one of their seventh round picks in exchange for Wes Welker. That seems to have worked out okay for the Patriots.

In Round 3 the Raiders gave up their 2008 third-round pick [No. 69] and their seventh-round pick [211] for the Patriot's third-round pick. they also traded WR Doug Gabriel to New England for pick No. 165.

Finally, we have the test of Belichick's "finding gold in the later rounds" theory. The Patriots had seven picks after No. 130. As you can see on the chart, the answer is, "it's a crapshoot." The total WCAV accumulated for the team by those seven picks is one. (A couple of the fourth-round picks went on to produce something for other teams, bring the WCAV up to nine.) Yes, essentially the entire value of their 2007 draft was produced by their first-round pick, and Brandon Merriweather's WCAV of 22 is less than half of the WCAV of the first-round pick for Cleveland, and just over half that of the Bengals and Steelers. Only the Ravens' first-round pick in 2007 is equally unimpressive, if we compare it to the AFC North.

Of course, to be fair we have to add in the value of the players they traded picks for, and it makes quite a difference, as you can imagine. If you add in the value Wes Welker brought to New England, suddenly this goes from a poor draft to an above-average one. Or so it would appear. I can't really say that at this point, because one would have to also add in the value of all players traded for by all the other teams. I may do that at some point. But it is fair to say the Welker trade takes this from a poor draft to at least an average one.

And at last we come to the comparison with the Steelers. I apologize to the rest of the AFC North. The other members of the division will appear again in the next article in the series, but trying to figure out Bill Belichick's trades wore me out...

Let's look at the chart again with Pittsburgh added in:

Conveniently, Pittsburgh's two best drafts are in years where the Patriots had less strong years—2002 and 2007. Curiously, the Steelers' two worst drafts were also in years when the Patriots were below-average—2006 and 2008. We'll go with those, however, and see where they lead us. I'm going to run the same exercise for the Steelers, and I don't know for sure but I'm guessing it will be a lot easier.

In 2002 the Steelers began with the following picks:

Round 1: 30

Round 2: 62

Round 3: 94

Round 4: 128

Round 5: 166

Round 6: 202

Round 7: 242

There was a single change to this, in the seventh round—the Steelers traded Kris Brown to the Texans for their seventh round pick [212]. This was not the pick with which they took one of the all-time great seventh round picks, Brett Keisel. They used that pick for LaVar Glover, a DB who did exactly nothing for anyone. The rest of the draft was classic solid Steelers drafting, with the best two picks being Larry Foote and Keisel. However, the only "bust" was the afore-mentioned seventh-round pick, and the sixth-round pick wasn't anything to write home about, either. But there was nothing fancy about any of the picks.

How about 2007? Here goes:

Round 1: 15

Round 2: 46

Round 3: 77

Round 4: 119, 132

Round 5: 156, 170

Round 6: 192

Round 7: 227

The Steelers did actually make a trade this time—they traded with Green Bay to move up for Daniel Sepulveda. Here's how it went down:

Round 1: 15

Round 2: 46

Round 3: 77

Round 4: [119, 132] 112, 132

Round 5: 156, 170

Round 6: [192] no pick

Round 7: 227

Once again, nothing fancy. The one trade they made doesn't show up on the books, again because of PFR's non-rating of punters, but Sepulveda, despite being oft-injured, was an above-average punter. Which makes this draft possibly look even better.

Now for the down years. Here's the Steelers' 2006 picks:

Round 1: 32

Round 2: 64

Round 3: 96

Round 4: 129, 131, 133

Round 5: 164. 167

Round 6: 201

Round 7: 240

Here's what they actually did. As you can see, they did a lot more trading than the previous two years we looked at.

Round 1: [32] 25

Round 2: [64] no pick

Round 3: [96] 83, 95

Round 4: [129, 131, 133 ] 131, 133

Round 5: 164, 167

Round 6: 201

Round 7: 240

The first-round trade was for Santonio Holmes. The Steelers traded their first-round pick [32], their third-round pick [96], and their fourth-round pick [129] to move up seven slots. Apparently intoxicated by the unusual feeling, they also traded their second-round pick [64] for two third-round picks.

What went wrong? Well, in a year in which the drop-off between the second round and third round was much larger than usual, and the drop-off between the first round and second round was much lower than usual, the Steelers traded out of the second round (admittedly the very end of it) for two third round picks, neither of which were terribly successful. The fourth-round picks were okay, and they got nothing—zip—in terms of WCAV, at least, from any of their fifth through seventh round picks, in a year when rounds five and six were better than the 10 year average. Not too surprising it wasn't a great year.

Here's a graphic comparing the 2006 value per round with the 10-year average:

Hindsight is always 20-20, but if the Steelers had picked the player the Giants did with the No. 32 pick, they would have taken a lesser player than Holmes. They might also not have Lombardi No. 6. As for the second-round pick they traded away, the Vikings took Tavaris Jackson with it, which wouldn't have helped the Steelers much. I'm sure there were lots of other fabulous players on the board at that point. Isn't it fun to play "what if?"

Finally, 2008. Ugh...

Round 1: 23

Round 2: 53

Round 3: 88

Round 4: 123

Round 5: 156

Round 6: 188

Round 7: 232

The Steelers made no trades until the fourth round, at which point they traded away pick No. 123 to the Giants for Nos. 130 and 194. The Giants picked up a pretty meh player with No. 123. The Steelers got one of their better players in this draft, Ryan Mundy. (They also took Tony Hills with No. 130.) After that they stayed put until the seventh round, where they traded pick No. 232 to the Falcons for 5'8" DB Allen Rossum.

Rossum only stayed in Pittsburgh for that season, but earned his keep, at least according to PFR. Or, at least, his Average Value of 2 for 2008 is better than that achieved by, wait for it, ANY OTHER PLAYER in the Steelers' 2008 draft, with the exception of Rashard Mendenhall and Ryan Mundy. (He tied Dennis Dixon, who has a lifetime CAV of 2.) However, it isn't quite that simple—the average WCAV for a seventh-round pick in 2008 (on the team drafting him) is 3.3. So while Rossum was only in Pittsburgh for one year, and achieved a great deal more CAV in Atlanta before coming to Pittsburgh, the Steelers traded a potential 3.3 points of team value for 2 points. However, given how few hits and how many misses the Steelers made that season, it's difficult to say they made the wrong choice.

The original question, now lost in the mists of time and the plethora of data, was, do draft trades pay off? The question isn't whether any particular trade pays off, but whether utilizing trades as a core part of your drafting philosophy is going to pan out, more often than not.

Needless to say, we can scarcely answer this question with the amount of data I have presented. But we probably already know the answer to the question. I suspect the answer is, it depends a lot on the person doing the trading. One thing which the limited data would not really support as a good option, though, is the idea of trading down for multiple picks in the lower rounds. There are two examples of this with the Patriots, 2002 and 2007. Ironically, one of those was a phenomenal draft and one was a below-average one, even adding in the value of traded players. The question is, does the greater number of chances of hitting the lottery in the lower rounds balance out with the much greater chance of hitting it big in the top rounds?

Just as an illustration, here's the average WCAV of players picked in each round, for both years:

It's amazing how similar the two charts look. The 2007 starting point is lower, which I'm guessing is almost entirely because most of the successful players from the 2002 draft have played out the prime of their careers, where many of the 2007 players will still be peaking. I'm guess that in four or five more years they will be almost identical.

One curiosity is the sixth round. Note that in both cases it is substantially lower than the seventh round, which is counterintuitive. I decided to redo the charts with the numbers adjusted for how many players were actually drafted in each round. Typically there are a ton of comp picks in the seventh round, and I wondered if that makes the difference. Let's see:

My hunch was correct. By far the most players in any given round in both years were drafted in the seventh round—50 of them in 2002, 45 in 2007. Adjusting for this makes a difference, but not as much as you might suspect. So is the sixth round typically the weakest round in the draft, per player taken? The only way to find out was to run the rest of the numbers, so I did. After adjusting each round of each year for the number of players chosen, I averaged each round:

As you can see, the two years we looked at above were anomalies—on average, a sixth round pick is ever so slightly better than a seventh round pick. As to whether it makes a lot of sense to trade down (as a common strategy) to grab a bunch of late-round picks, though, I would say no. It is extremely rare for a first-round pick to be a total bust, and, despite Limas Sweed, it's pretty rare for a second-round pick to be a bust, either. Therefore, any given first or second round pick is pretty likely to be right around the league average. But when you get to the late rounds, it is typical for the majority of the picks to be at or below average, with just a small number of very good players raising the average. Here's an example, at random:

I had the draft page open for 2001, so that's what I picked. (Furthermore, this gives the maximum amount of time for a late-round pick to pan out.)

In Round 4 35 players were chosen. The "average" total WCAV (not just on the team but for their careers) was 13.4. Of those 35 players, six earned a WCAV of zero. Nine more players earned a WCAV of 5 or less. Nine more players earned a WCAV of between 6 and 13. So 87.5% of the total value of the class came from the other 11 players.

In Round 5 32 players were chosen. Of those, 10 accumulated no WCAV at all. The average WCAV for that round was 5.3, and seven players accumulated a WCAV of 5 or less. A little over half the players accumulated 90% of the total WCAV.

Round 6 saw 36 players chosen. 14 of them accumulated no WCAV. A good percentage of the 14 never made it through training camp. The average WCAV was 3.2—11 players earned a WCAV of 3 or less. So 78.3% of the WCAV for the round was earned by less than a third of the players chosen.

In Round 7, 17 of the 45 players chosen earned no WCAV at all. For some unknown reason the seventh round was very strong that year, so the average WCAV was 7.0. An additional 16 players earned a WCAV of 7 or less. So 89.2% of the class WCAV was earned by a bit more than a quarter of the players chosen.

You can see that, unlike the top two rounds, your chances are not great that you will end up with even an average player. In 2001 you had the following percentages:

Round 4: You had a 17% chance of a complete bust, a 51.5% chance of a player at the league average or below, and a 31.5% chance of getting a better-than average player. The best player by far in the class was tackle Ryan Diem, whose WCAV of 70 is distinctly better than the WCAV average of 54.8 for the first round. On the other hand, he was the only player to clear that mark. On average, an achieving fourth-round pick is going to be "worth" about a quarter of an average first round pick.

Round 5: 31.2% chance of an essentially worthless player; 21.8% chance of a player who would accumulate, at maximum, a lifetime WCAV of 5 or less, and a 53% chance of getting a player who would be better than a 5 lifetime. The average fifth-round pick is only going to be about 1/10 of the average value of a first round pick. (The actual numbers for 2001 best picks: first round: LaDamlian Tomlinson, lifetime WCAV 126: best fifth-round pick, Russ Hochstein, lifetime WCAV 21.)

Round 6: About 39% of the players were worth nothing in terms of WCAV; another 30% or so will earn the average or less. That means your chances of getting a substantially better than average sixth-round pick are about 1 in 3. (The best 2001 sixth-round pick was Cedrick Wilson, with a WCAV of 22.)

Finally, the seventh round. About 38% of the players accumulated no WCAV. Another 36% accumulated the average or (mostly) less. Your chances of getting a good player in the seventh round were about 1 in 4. The best 2011 pick was T.J. Houshmandzadeh, with an accumulated WCAV of 56.

So unless you think you are an extraordinary evaluator of talent (I would guess Bill Belichick does think this, with some justification) and you are lucky besides, the numbers don't favor trading heavily into the lower rounds. As we saw, trading down heavily into the lower rounds wasn't exactly what the Patriots were doing anyhow. They were collecting extra higher-round picks in subsequent years and scoring big on small downward adjustments. At least that's what I see.

And naturally one has to consider that there are only so many roster spots. This is the same for everyone, of course. But I would think it makes more sense, at least for most teams, to stick with a smaller number of higher picks and pick up a bunch of UDFAs to fill out the camp bodies. Your chances of getting lucky with a UDFA is probably not that much less than it is with the players in the fifth and sixth rounds, and in the meantime your chances of getting a good pick are much better in the higher rounds. It almost argues for a strategy of trading up for fewer picks.

Which brings us around to the original question—has Bill Belichick changed his strategy, given the rookie salary cap? There's too little evidence to tell, yet. But I'll be watching...