SL Time Machine: More Re-Draftin’, 2000-2010

Rick Reilly has an edict that, ironically, he seems to repeat every time he opens his mouth: “One of my rules is to write sentences that have never been uttered before in the history of the English language.” I will now heed this advice with the following five words:

“Drawing inspiration from Rick Reilly…”

This is a true statement, even if a little embarrassing. I blame the lockout. Either way, Reilly’s NFL re-draft column defied the odds, somehow avoided the usual self-serving banalities, and was … pretty interesting.

Maybe it’s my unhealthy infatuation with trying to rewrite history, particularly when it comes to the Steelers and the draft (handy links here and here). Whatever, my affliction is your gain. I took Reilly’s idea, as well as my previous forays into time travel, to build a better, stronger, faster, etc., Steelers Lounge Time Machine.

Specifically, I used Career Approximate Value (CarAV), Pro Football Reference‘s methodology for determining player value based on NFL productivity. It is explained in great detail here.

(Just to be clear: CarAV has limitations, which author Doug Drinen acknowledges up front:

“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.”

With that out of the way, let’s get to it…)

So here’s the deal: I looked at every draft from 2000-2010, which also happens to coincide with Kevin Colbert’s tenure in Pittsburgh. I then ordered every draft by CarAV to see how good or bad teams were at predicting the future success of the players they selected.

(Because this turned into a much larger endeavor than I originally intended, I’ll spread the findings out over a few posts.)

Below: every first-round re-draft from 2000-2010 by CarAV. I also include the original draft position for comparison, and a separate table showing the Steelers’ selections each year.

A few things:

* Because CarAV is cumulative, the longer you play, the higher your CarAV. Which is why Steve Hutchinson’s (class of ’01) CarAV is higher than Ben Roethlisberger’s (’04). It also means that guys like Larry Foote and Antwaan Randle El are ranked relatively high compared to their draft classmates. But that’s sort of the point: longevity plus productivity makes for a good player and a solid draft pick.

* I probably should have stopped after 2007, or maybe 2008, because the results become noisy after that. For instance, looking at the 2009 draft table, CarAV says that the Pats’ Sebastian Vollmer should have been the second-overall pick. Vollmer has played well, but is he as good as Jake Long or Joe Thomas? It’s hard to say definitively since he’s only played two seasons. (I ended up including every draft through 2010 because I figured someone would ask for it. Just take the results with a grain of salt.)

* The tables that follow are, as mentioned, ordered by CarAV. That doesn’t mean that, say, John Abraham should have been the No. 3 pick in 2000 based solely on his career productivity. Clearly, teams have different needs based a whole bunch of things. Instead, this is a first-to-worst ordering of players by output. It allows us to compare, comment, and, well, waste time until we have actual football again. This seems obvious, but for completeness, I point it out it here.

* Note to nerds: you can view the data here.

Onward…

(FYI: New_Rnd = new round player was drafted using CarAV; New_Pick = new pick player was drafted using CarAV; Rnd = original round drafted; Pick = original pick drafted)

2000 (Re)Draft

Steelers’ Selections


Other notes
It’s no surprise Brady’s the first overall pick, but it appears the Ravens were spot on with Jamal Lewis at 1.5 despite my RB Fungibility Theory. Same deal with Shaun Alexander at 1.6. I’ll write about this in more detail in a subsequent post, but the Ravens know what they’re doing in the first round. More than just hitting on their first-round picks, they also have a knack for finding first-round talent in later rounds. In 2000, it was Adalius Thomas, originally a sixth-rounder.

One of the common complaints of the Colbert era is that while he’s money in Round 1, he’s average or worse late in the draft. One thing worth considering: a lot of his early selections turn out to be first-round quality. So here, in addition to Plax, Marvel Smith played like a late first-rounder.

For the Steelers: the Clancy and Poteat picks don’t seem nearly as bad when viewed through CarAV, and Haggans played like a second-rounder.

* Worst first-rounder: Jay Soward, WR, Jags. Originally drafted 1.29, now 5.162 in re-draft.

2001 (Re)Draft

Steelers’ Selections


Other notes
I remember reading recently that the Bengals gave serious consideration to taking Drew Brees with the fourth-overall pick but eventually settled on Justin Smith. It turns out, that’s exactly where Brees should have been taken. But the Bengals are the Bengals and fate intervened. Still, Cincy got three first-round talents in ’01: Chad Ochocinco, Smith and T.J. Houshmandzadeh, originally second-, first-, and sixth-rounders.

For the Steelers: Colbert & Co. took Hampton, Bell and Okobi right where they should have, missed on Nkwenti but did well with Bailey.

* Worst first-rounder: Jamal Reynolds, DE, Packers. Originally drafted 1.10, now 6.164 in re-draft.

2002 (Re)Draft

Steelers’ Selections


Other notes
Woo hoo, the Steelers took a second-, third-, and fourth-rounder and turned them all into first-round selections! It’s hard to imagine Foote or Randle El as first-round talents, but part of that has to do with longevity as well as their relative productivity. Look at it this way: all three players contributed to the Steelers’ 2005 Super Bowl, all were playing on rookie contracts (I think), and none were making first-round money. However you label and order them, that’s a nifty draft haul.

More impressive than that: Keisel, originally the 242nd pick, is now an early second-rounder. In fact, you could argue that he’s more a first-rounder than Foote, Hope or Randle El.

* Worst first-rounder: Wendell Bryant, DT, Cards. Originally drafted 1.12, now 5.147 in re-draft.

2003 (Re)Draft

Steelers’ Selections


Other notes
In the re-draft, the Steelers have two defensive backs as first-round picks, and Taylor would be even higher if he could catch. For the purposes of re-signing him for a reasonable sum, we’ll call that a feature instead of a bug.

The latest evidence Alonzo Jackson was a stretch: according to CarAV, the Steelers drafted him 107 spots to soon.

* Worst first-rounder: Jerome McDougle, DE, Eagles. Originally drafted 1.15, now 5.164 in re-draft.

2004 (Re)Draft

Steelers’ Selections


Other notes
This is something I will have to investigate further but first impressions are that there are a lot of running backs worth top-10 selections, but most of them were originally drafted after 1.15.

I remain unconvinced that DeAngelo Hall is a top-10 selection although Chris Cooley is worthy of 1.19.

For the Steelers: Roethlisberger’s a top-3 players, which is no surprise. And for the second consecutive draft, the Steelers took a player in Round 2 that was a fifth-round talent. Starks turned out to be a second-rounder, while the front office missed on every player not named Ben or Max.

* Worst first-rounder: Rashaun Woods, WR, 49ers. Originally drafted 1.31, now 6.178 in re-draft.

2005 (Re)Draft

Steelers’ Selections


Other notes
Rodgers is only the sixth-overall pick because he had to sit behind Favre for three years. The fact that he’s already accumulated CarAv of 48 is a testament to just how good he has been in a short period. The Pats look brilliant taking Logan Mankins, and the Seahawks were right to draft Lofa Tatupu at 2.45 (which flabbergasted some folks at the time).

Despite hands of stone, Braylon Edwards has had some productive seasons, although Vincent Jackson might be more valuable. The Chiefs got Matt Cassel for a second-round pick, so clearly Belichick gave Pioli a hometown discount (*cough* collusion *cough*). This sounds like a job for Roger Goodell.

For the Steelers: turns out, B-Mac has performed like a second-round pick. In terms of CarAV, Antrel Rolle (now 2.36, originally 1.8), Carlos Rogers (now 2.42, originally 1.9), Ellis Hobbs (now 2.45, originally 3.84), and Corey Webster (now 2.53, originally 2.43) rank ahead of McFadden, but Rolle was moved to safety, Hobbs will likely retire due to neck injuries and Rogers and Webster aren’t huge upgrades. Basically: the Steelers did the best with that they had. B-Mac’s a serviceable No. 2 corner.

Kemoeatu has provided third-round productivity, as has Essex (as a backup), while Fred Gibson and the “He’ll replace Plax!” hype barely lasted through ’05 training camp.

* Worst first-rounder: Troy Williamson, WR, Vikings. Originally drafted 1.07, now 4.119 in re-draft. (David Pollack and Erasmus James were first-round picks with lower CarAV, but both suffered career-ending injuries so I didn’t include them here.)

2006 (Re)Draft

Steelers’ Selections


Other notes
Leading up to the ’06 draft I remember all the talk about the complete lack of depth among the wide receivers. Basically, Santonio and Chad Jackson were the only first-rounders, although Jackson ended up going to New England in Round 2. Five years later and this class was stacked: I’ll put Hester (although he’s more dangerous as a returner), Jennings, Colston, Marshall, and Holmes up against any WR class.

After watching Kyle Williams destroy Kemoeatu last season, I can very easily seem him as a top-15 pick.

The Saints have a knack for finding first-round offensive linemen late in the draft. In fact, they have three first-rounders in the ’06 re-draft, and Jahri Evans and Colston have been more productive than Reggie Bush, originally taken second overall.

For the Steelers: nothing’s changed — Smith and Reid remain wasted picks. Colon’s now a third-rounder (which lends credence to the argument that he probably wasn’t one of the NFL’s best right tackle’s in 2009), and the front office would have had more success throwing darts at the draft board with selections 4-7.

* Worst first-rounder: John McCargo, DT, Bills. Originally drafted 1.26, now 5.155 in re-draft.

2007 (Re)Draft

Steelers’ Selections


Other notes
Some surprising names here, starting with Patrick Willis. He gets lost in San Francisco and you forget how good he really is. Lynch, Griffin and Weddle seem like stretches, as does Jamaal Anderson. This could be early signs that CarAV becomes less stable as we approach 2011.

Fun fact: Brady Quinn falls from 1.22 to 5.150. Sounds high.

Still, Woodley has played like he’s worth the No. 6 pick, and along with Timmons has worked out well for Pittsburgh.

Gay is now a third-rounder, Spaeth is 19 picks worse, and McBean is slightly more productive (although that was with the Broncos after the Steelers released him). Sepulveda’s CarAV is 0, but that’s because I don’t think it accounts for kickers or punters.

* Worst first-rounder: Justin Harrell (Packers) and Jarvis Moss (Broncos) share the honors. Harrell was originally drafted 1.16, Moss 1.17, and now are 6.161 and 6.162 in the re-draft.

Because you’re no doubt wondering: JaMarcus Russell, the new face of “Biggest Bust Ever” (Ryan Leaf finally catches a break), ends up at 4.105, primarily because he was afforded so many opportunities to fail.

2008 (Re)Draft

Steelers’ Selections


Other notes
Keeping small-sample caveats in mind, the Ravens have two top-5 players, and the Saints discover another late-round lineman with first-round skills.

For the Steelers: Limas somehow avoids the seventh round (but still ranks higher than Bruce Davis!), Tony Hills does not, and Ryan Mundy provides some value as a fourth-rounder.

* Worst first-rounder: Vernon Gholston, DE, Jets. Originally drafted 1.06, now 4.102 in re-draft.

2009 (Re)Draft

Steelers’ Selections


You could make a case that Wallace is the No. 1 player in this draft although No. 6 ain’t bad. Early returns are that Ziggy has been a second-round talent, Urbik a bust, Lewis a reach and Burnett exactly what the team thought. David Johnson is now a fourth-rounder.

* Worst first-rounder: Andre Smith, T (Bengals), Aaron Maybin, DE (Bills), and Peria Jerry, DT (Falcons) share the honors. Smith was originally drafted 1.06, Maybin 1.16, and Jerry 1.24. Now they 4.103, 4.104, and 4.105 in the re-draft.

2010 (Re)Draft

Steelers’ Selections

All bets are off. I’ll get behind the top-4 selections but can’t see tight ends going fifth and sixth (to the same team, no less … although it if were going to happen it would be the Pats). Also: Colt McCoy is now a first-rounder. We have reached the event horizon, folks.

Based on one season, Manny is now a second-rounder, Antonio Brown a third-rounder and Sylvester a fourth-rounder. Take that for what it’s worth, which at this point isn’t much.

This entry was posted in Analysis, Draft, Random Junk and tagged , , , . Bookmark the permalink.
  • t1mmy10

    this is very interesting results. except for one year, colbert only hits on the occasional prospect after round 3 or 4. i just wonder how he compares to the rest of the league in that regard.

    i realize that you do point this out in your intro & the AV rating system is probably as good as any while being easy to use, but it’s really called into question in my mind when Chris Chambers, one 1,000 yard season in his career & no pro-bowls, is ranked above casey hampton, the anchor to our defense for the past 10 years. i get the impression that this system strongly favors the positions were it’s easy to measure stats.

  • Anonymous

     One thing that struck me is how sitting behind a good player affects your positioning. CarAV is based on contribution, not talent, and it really shows up when a good young player rides pine behind a good veteran. Still, from the perspective of the team doing the drafting that’s something that should have been considered.

  • Cols714

    I think we can put to rest the “Colbert is not that great after the first round” myth. Just looking at this he seemed to do pretty well at finding players who had good productivity after the first round.

    I realize that people are not going to be happy until every 5th rounder becomes a starter, but I think he’s done a pretty good job.

  • EasyLikeSundayMorning

    Ryan, this is really interesting stuff. I’m guessing that CarAV has it’s flaws, as any rating system does. But it seems directionally correct. If it isn’t a massive amount of addition data entry to figure it out, in your folllow up, maybe you can address a question related to the issue Cols raises: how does Colbert stack up to other teams (eg, by round and overall)?

    My guess is that he’d come out looking very good. In the 11 years he’s drafted for us, Colbert has gotten 1 strong contributor once (08), 2 strong contributors 5 times (01, 03, 04, 06, 09), and 4-5 strong contributors 5 times (00, 02, 05, 07, 10). In the 6 years where he got only 1-2 strong contributors, he got 5-6 elite players (Casey, Troy, Ike, Ben, Santonio, Wallace) and 4 other starters (Bell, Starks, Colon, Ziggy). Plus, he drafted at least 30 guys who still deserve to be on NFL rosters.

    Overall, that seems like a very good haul from the draft. Add in some strong UDFAs (Harrison, Parker) and you have the core of an elite team so that we haven’t had to overpay to bring in free agents. That’s a great blueprint for being a perennial Super Bowl contender in the free agent and salary cap era.

  • EasyLikeSundayMorning

    Ok, here’s a little math.  By dividing CarAV in each draft by the number of seasons since each draft, Colbert’s picks have averaged 16.9 per year. Two drafts were significantly above this (23.6 in 2002 and 23 in 2007) and two drafts were significantly below this (12.6 in 2001 and 8.7 in the putrid 2008 draft).

    The closest to average was 2005, when we got Heath, McFadden, Essex and Kemo. This gives us a sense for the (imo high) standard that Colbert has set. 

    On this measure, the drafts with 4-5 strong contributors fair better than those with 1-2 elite players.  To me, this means you shouldn’t pay a steep price to trade up unless you think you already have a strong and deep roster.

    • Anonymous

      Don’t you have to weight that according to the number of draft picks (Not the same as actual players chosen, see Intropy above)?

  • EasyLikeSundayMorning

    Ryan, In case it isn’t obvious, I really like how you put this data together.  It’s like a cat toy and I’m Will Ferrell at his SNL audition (http://www.myspace.com/video/vid/10918502).

    Not surprisingly, Colbert has done very well in r1.  Based on our r1 draft slots, we should have expected to have gotten 405 in CarAV but have gotten 449.  That’s like an extra r1 pick above what we should have expected.

    Most of our r1 picks were within a handful of points for their expected CarAV.  Only the injured and adult-onset-diabetic Simmons was significantly below (-11) and Troy and Ben were significantly above (+21 and +22).  

    Pouncey was +5, which is really high for one season.  He had the same value as Ryan’s redraft picks at 1.17 and 1.24 combined.

    Looking past r1, Colbert also drafted 7 guys who are r1 picks in Ryan’s redraft.  The more I look at it, the stronger Colbert’s drafting looks…

    • Anonymous

      Your comment about Troy bring another point to mind. When judging player’s performance value, you should use their draft positions. But when judging a team’s selection value you shouldn’t use the draft position consumed, you should use the initially available position, which in his case was the sum of 27, 92, 200. Those picks would be valued at about 55 CarAV on average, which would make him a +20 if he retired today. available position, which in his case was the sum of 27, 92, 200. Those picks would be valued at about 55 CarAV on average, which would make him a +20 if he retired today.

      • EasyLikeSundayMorning

        Good point.

  • JschericH

    A lot of teams take players with char concerns in those later rounds. We handcuff ourselves by avoiding those players that fall. Our round 3-7 picks are usually under achievers at big schools or small school players with holes in their games. I don’t have any problem with this strategy. In Colbert we trust.

  • ROb D

    Great stuff to read..I always thought that Colbert was unfairly slagged in some quarters for not finding gems late in the draft. No system of evaluation is perfect and I am sure  CarAV has flaws. But I really believe the Steelers are amongst the top 5 teams with regard to drafting wisely. And we know they aren’t drafting in the top 15 every year to boot.

    They miss on players like everyone else but its not exactly an exact science. And you can’t hide players from the talent evaluators like the old days. You add in their UDFA record and you can see why this team is always fielding a contender.

    Very enjoyable read…thanks.

  • http://pulse.yahoo.com/_FCTLJGMUSHGHK7NLZ3VRTGGNP4 Matthew

    And here I was thinking I’m the only one who thinks Rick Reilly usually comes off as a clueless moron.

  • Ted

    Wow, awesome read and research, Ryan; the kind you cannot find anywhere else on the Web. As a pro football historian and HOF fanatic, Pro Football Reference is one of my favorite sites.

    CarAV is an interesting concept that provides some measure of comparison and is likely better than any system I could come up with. While you do note the system’s flaws, however, you then seem to discount those flaws when talking about Steelers’ draft picks. You correctly noted how it puts too much empahsis on career longevity.

    But some of the Steeler rankings show why this system should not be used when analyzing past drafts. Putting Kendall Simmons as a second-rounder and Willie Colon as a third-rounder is a prime example. Colon did not play his first year and then missed all of last season with an injury. But he was by far the best offensive lineman we had in 2009 for a good offense and did so while playing a valuable position at RT on a mediocre offense.

    Simmons, in contrast, is arguably the only real bust first-round pick of the Colbert era. For 1.5 years he was average at guard and was horrendous for the rest of his career, while playing the easiest position on the o-line at RG. He kept starting when healthy because the Steelers had even worse players on the bench. But Simmons was the weakest link on an offensive line that was the weakest part of a decent offense. For that he is rewarded the same amount as the best player on a beter o-line, assuming that player did not go to a Pro Bowl.

    I know you said not to look at 2010 CarAVs yet, but please do and you can see why the system is fatally flawed. Jason Worilds’ only real playing time on defense last year came on obvious passing downs in the second half of the Miami game after Woodley went down. Worilds had no defensive stat line in that game, although he did provide a nice outside speed rush. He also saw a few plays of defensive action late in a few meanlingless blowout games, picking up his two sacks for the season in the fourth quarters of a 38-13 win at Tampa and 35-3 over the Raiders. Those two boosted his CarAV, but the statless Miami game was his only real contribution to the Steelers’ 2010 defense.

    However, most of his 17 tackles came on special teams, where he was so invaluable that he did not even make the dress roster for the Super Bowl despite being fully healthy. In hindsight, the faster and more mature Worilds should have dressed over veteran Keyaron Fox.

    Regardless, Worilds had almost no impact in 2010 for one of the NFL’s top-ranked defenses and probably watched more than 98% of the team’s defensive plays from the sidelines. Still, through this system he gets credit just for being on the roster and loosely affiliated with that great defense. The folks waving towells in Section 522 had just as much an impact on that defensive ranking as Worilds. He will hopefully become a great player and one day a starter, but Worilds did very little on the field last year.

    Nevertheless, Worilds still gets a second-round grade through the CarAV system because he saw action on a below-average kick coverage unit (sadly good from a Steelers’ perspective, though) while officially being listed as a linebacker on the roster. In contrast, Antonio Brown only gets a third-round grade in the same draft.

    You see Worilds played in 15 regular-season games compared to only nine for Brown, who was battling Sanders and Randle El for spots on the dress roster each week. Still, Brown had 16 regular-season receptions before emerging as a key contributor in the playoffs. The Steelers do not make the Super Bowl without Antonio Brown on the roster in 2010. He was that important. However, having Arnold Harrison, Frazier, Thad Gibson or anyone else as the Steelers’ No. 9 LB instead of Worilds last season would have made no difference.

    And Brown’s impact on STs was obviously much greater than Worilds. He averaged 23 yards per kick return and we won the Tn game in large part due to his opening KO return for a TD.

    While the Steelers ranked in the top two in total defensive and defenisve scoring last season, they were a mid-tier offense, which hurt Brown in the CarAV, while rewarding Worilds. No one can honestly defend a system that computes Jason Worilds as having a better NFL career than Antonio Brown at this point.

    • EasyLikeSundayMorning

      There’s no such thing as a flawless rating or ranking system. It is always easy to find fault with a specific rating or ranking and try to use this to damn a whole system. There will always be “noise” in the raw and analyzed data. To me a few debatable points up or down on Simmons, Colon and Worilds seem like issues within an acceptable margin of error.

      To me, the question here is: is the system accurate enough to draw overall conclusions across Colbert’s 11 previous drafts? The answer seems to be yes, unless you or others can show some significant, systematic problems. As you said, it is likely better than anything you (or any of us) could come up with. So, I prefer to focus on the big picture patterns it tells us rather than fixate on the inevitable debates on the margins.

    • Cols714

       Gee, no one could have guessed that you would use this data (which Ryan admits is flawed and gets worse as the drafts become more recent) to remind us that you hated the Jason Worilds pick.