Expected Failed Completions is another step in using game charting to break down the air and YAC components in a successful connection. We look at a decade of results and hone in on 2015.
11 Apr 2013
Two weeks ago, I introduced an updated version of my draft efficiency model, and last week I presented franchise rankings based on the two resulting metrics, value above expectation (VAE) and return on investment (ROI). Today, we turn our attention to the best and worst draft classes (i.e., the group of players selected by a given team in a given year).
Much to my surprise, ranking draft classes turned out to be not so easy because the first question one has to answer before coming up with a set of rankings is sneaky tough: Which stat should I use to determine the rankings? VAE is a measure of total value added, while ROI is a measure of percentage value added. So what's more important, then: adding more value or being more efficient? This dilemma is not unlike trying to decide whether DYAR is a more appropriate measure than DVOA when trying to rank individual players.
Of course, as reader Kal pointed out in the comments to last week's piece, there's also something to be said for evaluating a draft class based on total value, regardless of expectation (i.e., by Adjusted Career AV per Year). Now, I obviously think VAE and ROI are much better gauges than sheer value -- otherwise, what's the point of this entire series? Nevertheless, getting a ton of value (above expectation or not) is an accomplishment in its own right, and shouldn't be entirely ignored.
This being 2013, I put the question to Twitter. And while I got good feedback, it didn't settle the issue in my mind. Adjusted CarAV/Yr (i.e., total value) will probably fit the conventional wisdom, but almost certainly overrates larger draft classes. ROI has the opposite problem: It's probably going to overrate smaller draft classes, even if we set a "minimum number of picks" limit. VAE, on the other hand, will almost certainly fit conventional wisdom more than ROI, but will also tend to overrate large draft classes.
In the end, I've decided to list the best and worst drafts according to all three measures, and let you, the readers, decide for yourselves. Of note, though: We're only dealing with the top 222 picks in a given draft, and draft classes needed a minimum of six picks to qualify for the rankings.
A few years ago, the NFL Network did one of their "Top 10" shows on the very topic we're addressing today (click here for video). Somewhat amazingly, despite not catering to an audience of NFL historians and statisticians and having no pretense of objectivity, five of the eight post-merger draft classes on their list were indeed among the top 35 in total value, regardless of expectation.
One of those five, however, is a perfect example of what happens when you don't take expectations into account. No. 2 on their list was the combined 1991 and 1992 drafts of the Dallas Cowboys. I have no qualms with the 1992 draft class being on there, as it ranked 55th in VAE (+8.5) and 119th in ROI (+35.6%) in addition to ranking 24th in CarAV/Yr (32.4). We're talking about 959 qualifying draft classes here, so I'll allow a viewership-based fudge factor; 1991 is another matter entirely. Yes, the Cowboys' 1991 draft class ranked 35th in CarAV/Yr (29.7), but it actually brought a negative ROI (-17.2%), and ranked in the bottom 100 according to VAE (-6.2). The reason for this discrepancy is that, with 11 picks in the top 110, their expected CarAV/Yr of 35.8 was by far the highest of any draft class from 1970 to 2007. (The expansion Seattle Seahawks of 1976 were No. 2, with 33.0 expected CarAV/Yr.)
With that little digression out of the way, here is a table showing the top 10 and bottom 10 draft classes from 1970 to 2007 according to Adjusted CarAV/Yr:
|Best Drafts by CarAV/Yr (min. 6 Picks)||Worst Drafts by CarAV/Yr (min. 6 Picks)|
|Team||Year||# of Picks||CarAV/Yr||Team||Year||# of Picks||CarAV/Yr|
As expected, the rankings are highly dependent on how many picks the team had, even with the six-pick minimum.
Most of the bottom 10 shows up later, so for now I'll only make one observation from that side of the table. The Minnesota Vikings had 13 total picks (in the top 222) between 1989 and 1990, and got a total of 7.1 Adjusted CarAV/Yr. What's more, only three of the 13 registered higher than 1.0: David Braxton (No. 52 pick in 1989), Marion Hobby (No. 74 in 1990), and Alonzo Hampton (No. 104 in 1990).
On the positive side of the ledger, the 1974 Pittsburgh Steelers and the 1986 San Francisco 49ers are the only two draft classes that overlap between this top 10 and the NFL Network's top 10. Of course, it's ironic that a list overly reliant on the conventional wisdom of total value would omit a Steelers draft from the early 70s that turns out to have been the most valuable of them all.
Of Pittsburgh's 15 picks in 1971, seven produced more than 3.0 CarAV/Yr. There are the obvious guys like Hall of Famer Jack Ham (No. 34), and half of the Steel Curtain: Dwight White (No. 104) and Ernie Holmes (No. 203). Less obvious are Gerry Mullins (No. 86) and Larry Brown (No. 106), who helped pave the way for Pittsburgh's running game, and Ralph Anderson (No. 126), who started at free safety in 1972 before getting traded in 1973 to make way for two-time Pro-Bowler Glen Edwards.
Finally, there's Pittsburgh's first pick in 1971, Frank Lewis (No. 8), who was the type of player that totally goes unnoticed to people like me (i.e., born in the late 70s and not a savant when it comes to NFL history). Lewis started at wide receiver in the pre-Swann-and-Stallworth days, a three-year period (1972-1974) in which the Steelers threw for a total of 5,596 yards. (For comparison, Drew Brees threw for 5,476 in the one-year period of 2011.). With Stallworth perpetually injured in 1975 and 1976, Lewis remained a starter in name only, but was overshadowed by Swann's exploits. Then in 1978, Pittsburgh traded Lewis to the media blackout known as Buffalo, where his career took off. He had the longest receiving touchdown in the league that season (92 yards), and made the Pro Bowl in 1981 thanks to 1,244 receiving yards. Until this series, I honestly had no idea Frank Lewis existed.
The 1974 Baltimore Colts had another draft class that featured seven players with over 3.0 CarAV/Yr, and further educated me about the vagaries of NFL football circa the 1970s. Wide receiver Roger Carr (No. 24) made the Pro Bowl in 1976 after leading the league in receiving yards (1,112) and yards per catch (25.9). Fred Cook (No. 32) and John Dutton (No. 5) were bookends of the Sack Pack (Mike Tanier wrote about them six years ago; of course he did).
Where things get weird is that, in an era with minimal player movement, Dutton was one of five players from this group of seven that the team subsequently traded away. Robert Pratt (No. 67) was a seven-year starter at left guard before being traded to Seattle in 1982. Freddie Scott (No. 174) didn't start at wide receiver until his fourth (and final) year in Baltimore, but saw his career take off after being traded to Detroit in 1978. Meanwhile, Noah Jackson (No. 161) and Greg Latta (No. 188) played in different leagues altogether during their rookie seasons (Canadian Football League and World Football League, respectively), both were traded to the Bears in 1975, and both became multi-year starters along the line in Walter Payton's offense.
One final draft class worth mentioning is the 1981 New Orleans Saints -- if for no other reason than that it puts a date certain on my NFL consciousness. That draft class had six players with at least 3.0 CarAV/Yr, most of whom have names I actually recognize. Rickey Jackson (No. 51) made the Hall of Fame in 2010 based on his exploits as a member of the Dome Patrol (NFL Network's No. 1 linebacker corps of all time); but more notably he helped my favorite team win the Super Bowl in 1994. George Rogers (No. 1) won Offensive Rookie of the Year in 1981, and is one of only 33 rookies since the merger to be named first-team All-Pro. Elsewhere, cornerback Johnnie Poe (No. 144), strong safety Russell Gary (No. 29), defensive lineman Frank Warren (No. 57), and tight end Hoby Brenner (No. 71) each started with the Saints for at least five seasons. (Stat of the day No. 1: Brenner made the Pro Bowl in 1987 despite only 280 receiving yards, which is third fewest since the merger.)
Aside from all the interesting minutiae I learned, the main lesson from rankings draft classes by Adjusted CarAV/Yr is that it overvalues having lots of picks -- specifically lots of early picks. The five best players from Pittsburgh's 1971 draft were among their first six picks. The 1974 Colts had six picks in the first three rounds, and four produced at least 3.0 CarAV/Yr. And even more to the point, New Orleans had five picks in the first three rounds of the 1981 draft, and those players ended up being the best five of their entire draft. To me, it's pretty obvious that total value paints part of the draft rankings picture, but there remains a lot of empty white space if you don't add in the color of expectations. I'll try to do that now.
As I said at the outset, the concern for ROI-based rankings is that they will overrate smaller draft classes. Well, as soon as I clicked "sort" in Excel, the worry became a reality, and that's why I instituted a six-pick minimum. Without it, instead of there being six draft classes at +100.0% ROI or better, there would be 11. To boot, the top three would all have come courtesy of the notoriously draft-averse 1970s Washington Redskins. Last week, I mentioned George Allen's drafts in 1972 (one pick in the top 222, +142.9% ROI) and 1975 (two picks, 156.1% ROI), but the highest-ROI draft class of them all was two years after he left. In 1979, the Redskins had two picks (Nos. 103 and 182), and got a +163.5% return on their investment.
On the flip side, no pick minimum would take us from zero "It's as if we weren't even there" drafts (i.e., -100% ROI) to two. The 1982 San Diego Chargers had two picks in the top 222 (Nos. 188 and 215), got nary a game from defensive back Hollis Hall, and got three seasons from punter Maury Buford, who had one in every 10 punts blocked during his rookie year.
The other awful draft class with fewer than six picks in the top 222 was that of the 1975 Kansas City Chiefs, which you might notice came just three years after the above-listed 10th-worst draft class according to total value. Quite a run they had going back then. Of the Chiefs five picks in 1975, four never played a game. The fifth, running back Morris LaGrand, had 37 yards from scrimmage in 11 games with Kansas City, but got traded to New Orleans midway through his rookie year. That offseason, LaGrand got selected by the expansion Tampa Bay Buccaneers in the veteran allocation draft, but couldn't even make the roster of (or get a midseason callback from) a team that would end the year at 0-14.
Here is the table showing the top 10 and bottom 10 draft classes according to ROI:
|Best Drafts by ROI (min. 6 Picks)||Worst Drafts by ROI (min. 6 Picks)|
|Team||Year||# of Picks||ROI||Team||Year||# of Picks||ROI|
The first thing that jumps out at me from this table is that, yeah, it favors drafts with fewer picks. The second thing is that the Philadelphia Eagles own both the most efficient and the least efficient drafts since the merger. This isn't the first time we've seen their 1992 draft class in a table, but it's also not the last time, so I'll get into the details a little later.
The 1977 draft class, however, is an incredible tale of turning lemons into lemonade. They didn't have a pick until the fifth round (No. 119), and the 7.6 expected CarAV/Yr of their six picks was the seventh-lowest among qualifying draft classes. (For comparison, that's about one-fifth of the 1991 Cowboys draft class discussed earlier.) They end up No. 1 in ROI because those six late-round picks produced the 24th-highest VAE (+10.6). Most notably, nose tackle Charlie Johnson (No. 175) was a five-year starter who made first-team All-Pro in 1980 and 1981. That pick was good enough to produce a +584.6% ROI. At No. 154, the Eagles took running back Wilbert Montgomery, who started six years, made two Pro Bowls, and led the league in yards from scrimmage in 1979.
Of the other draft classes above +100.0% ROI, Bill Belichick benefitted from two: The 1990 Cleveland Browns and the 1995 New England Patriots. The Browns/Ravens draft resulted in power running back Leroy Hoard, who made the Pro Bowl under Belichick in 1994, as well as two starting defensive ends: Rob Burnett (No. 129) also made the Pro Bowl in 1994, and Anthony Pleasant (No. 73) had 11.0 sacks under Belichick in 1993.
Five of the Patriots' six picks in 1995 brought at least +50.0% ROI, the lone exception being Dino Philyaw at No. 195 (-100.0% ROI). Hall of Famer Curtis Martin (No. 74) and center Dave Wohlabaugh (No. 112) combined for +226.9% ROI, and cornerback Jimmy Hitchcock (No. 88) produced +85.4%. All three of them were gone by the time Belichick returned as head coach in 2000, but two key components remained from the AFC Champion defense Belichick helped coordinate in 1996 (as the Tuna's "assistant head coach"): Ty Law (No. 23) brought +56.8% ROI and Ted Johnson (No. 57) brought +80.3%.
Moving over to/from the dark side of the table (Belichick segue pun FTW!), let's revisit that 2005 Minnesota Vikings draft class, shall we? Two years ago, our own Tom Gower gave Troy Williamson a pass as the biggest wide receiver bust because of the immense shadow cast by Mike Williams' 272-pound bust. After the incredible shrinking Williams had a renaissance in Seattle three seasons ago (his ROI improved to -65.9%), we can officially put Williamson and his -73.2% ROI back on the mantel.
If only it stopped there, though! Let's not forget that, 11 picks after whiffing on Williamson, Minnesota whiffed on defensive end Erasmus James at No. 18 (-67.5% ROI). In the second round, they whiffed on offensive lineman Marcus Johnson (-56.2%). In the third round, they struck out looking at the hands of defensive back Dustin Fox (-100.0%). Then, in the fourth round, the golden sombrero came courtesy of running back Ciatrick Fason (-75.2%). In 2005, it was Rob Brzezinski, not Jacque Jones, who led Minnesota in strikeouts.
There were 20 draft classes with at least six picks in the top 222 that failed to result in a single positive-VAE player, only one of which shows up in the bottom 10 overall: It's those 1992 Eagles again! (Details soon, I promise.) In contrast, there were only three draft classes that didn't have a negative-VAE pick. I mentioned the 1990 Browns earlier; the other two were the 1998 Cowboys and the 1983 Bears.
After ragging on the wildly overrated 1991 Cowboys draft class, it's only fair to give their 1998 class its propers. It just missed making the tables, but it produced the 21st-best VAE among qualifying draft classes (+11.0), as well as the 13th-best ROI (+81.9%). The worst pick was Greg Ellis, who made a fine 12-year career for himself, thereby exceeding the expectations of a No. 8 pick (+0.5 VAE). The best pick was Darren Hambrick (+4.1 VAE) at No. 130, although he "Attitude Era-ed" himself out of the league after only five seasons. The other four picks were five-time Pro-Bowl left tackle Flozell Adams (+2.3 VAE) at No. 38, and three players who had mild success after leaving Dallas: safety Izell Reese (+1.7 VAE) at No. 188, offensive lineman Oliver Ross (+1.0 VAE) at No. 138, and serial killer Michael Myers (+1.5 VAE) at No. 100.
Unlike draft classes for the 1990 Browns and the 1998 Cowboys, the 1983 Bears draft class shows up in the VAE table below and in the NFL Network's Top 10. Incredibly, Chicago got Hall of Famer Richard Dent at No. 203, and his +4.7 VAE ranks just outside the top 100 most valuable picks since the merger. Sixteen picks later, they took Mark Bortz, who played his entire career in Chicago, was an 11-year starter at left guard, and made two Pro Bowls. In addition, they got half of their Super Bowl-winning secondary: safety Dave Duerson (+2.3 VAE) at No. 64 and cornerback Mike Richardson (+1.9 VAE) at No. 33. And even their two first-round picks brought positive VAE: No. 6 pick Jimbo Covert was a two-time All-Pro left tackle, and No. 18 pick Willie Gault was the favorite deep threat of Tecmo Bowl players across the land.
Here is the table showing the top 10 and bottom 10 draft classes according to VAE:
|Best Drafts by VAE (min. 6 Picks)||Worst Drafts by VAE (min. 6 Picks)|
|Team||Year||# of Picks||VAE||Team||Year||# of Picks||VAE|
That 1983 Bears draft class contributes to why I like this list more than the others: It's a good mix of conventional wisdom and whatever is the opposite of conventional wisdom. It's the only list with three members of the NFL Network's top 10 (1974 Steelers, 1986 49ers, 1983 Bears). It's also got that 1971 Steelers draft class that contributed heavily to their first two championships, and the 1987 class that helped bring the franchise out of a 1980s abyss. In addition, it's got the 1990 Green Bay Packers draft class that spurred their min-90s run of NFC dominance.
At the same time, it's got a few draft classes that ended up being for their franchises a lot of sound and fury signifying nothing. For instance, the best picks in San Diego's 1975 draft class made their names elsewhere. Running back Rickey Young was the 164th pick (+4.3 VAE), but only started three years with the Chargers before being traded to Minnesota for Pro Bowl guard Ed White. As a member of the Vikings in 1978, he led the NFL with 88 receptions.
Hall of Famer Fred Dean (+3.6 VAE) went to San Diego at No. 33, but then, like many Chargers of the day, went to San Francisco a few years later, and helped the 49ers start a dynasty. The third "value that got away" was ninth-round linebacker Larry Keller (+1.8 VAE), who started for two years with the Jets after never playing for the Chargers. (Stat of the day No. 2: Third-rounder Mike Fuller is one of only 38 non-kickers/punters since the merger to have made an extra point.
And now we come to the elephant in the room, which I hinted at in last week's piece, and have downright ignored so far today despite it showing up on the positive side of all three top 10 lists: The 1971 and 1972 Cincinnati Bengals. At first, when I saw this result, I was apprehensive. There had to be a glitch in the system somewhere. Then I remembered that the Bengals of the early 1970s were in the same division as the dynastic Steelers, and also had to compete with mini-dynasties in Miami and Oakland. Basically, the Bengals were just the first in a long line of post-merger franchises whose only affliction was horrible timing. Until this past season, the modern-day Ravens may as well have been a direct descendent.
In five seasons from 1972 to 1976, the Bengals went 46-24, but only made the playoffs twice. In 1973, they got in after a 10-4 regular season, but a Week 5 loss to the Steelers in Pittsburgh meant they didn't win the division, and therefore had to travel to Miami to play the defending- and soon-to-be-repeating champs. In 1975, they went 11-3, but had to play their first playoff game in Oakland because Pittsburgh went 12-2. In 1976, they missed the playoffs at 10-4 because -- here's a surprise! -- the Steelers won their head-to-head tiebreaker by virtue of a regular season win ... in Pittsburgh.
That kind of weird wild stuff ca. 1972 shouldn't take away from the Bengals' draft classes, however. Of their 15 picks in the top 222, 10 produced at least +2.0 VAE. There's obviously Ken Anderson (No. 67 in 1971), who is to statheads what Mick Tingelhoff is to Peter King: the "he" in "How the hell is he not in the Hall of Fame?"
But there's also Tommy Casanova (No. 29 in 1972), who was a six-year starter at safety, returned punts, made the Pro Bowl three times, and was voted first-team All-Pro in 1976. His +6.1 VAE ranks as the 23rd-best individual pick since the merger. Stan Walters (No. 210 in 1972) started at tackle for three years until getting traded to Philadelphia in 1976, where he subsequently made the Pro Bowl in back-to-back seasons (+5.6 VAE). Fred Willis (No. 93 in 1971) was Cincinnati's starting fullback (+4.2 VAE) until being traded to the Houston Oilers for Charlie Joiner in 1972; and of course, Joiner went on to join the Hall of Fame thanks to what he did with the Chargers, not the Bengals.
I could go on, but one final high-VAE pick worth mentioning was strong safety Neal Craig (No. 171 in 1971), whose career in Cincinnati was progressing nicely until the combination of ownership revenge and depth-chart dynamics found him banished to Buffalo in 1974. And in true Bengals fashion, the emerging backup who replaced Craig, Lyle Blackwood, was himself traded to the Baltimore Colts in 1977, where he proceeded to lead the NFL in interceptions his first year there.
Thousands of words ago, I said I was going to present the numbers, and let the readers decide. Well, you can do your deciding in the comments section, but here's my take. If we look at the totality of above stats, I think it's pretty clear which draft classes should be in the conversation for absolute best and worst since the merger.
In terms of absolute worst, the 1992 Eagles draft class is in a league of its own. It's the only one to show up in all three tables, and it's dead last in two of them. As already mentioned, it was one of 20 draft classes to feature zero positive-VAE picks. But wait, there's more!
Of Philadelphia's eight picks in the top 222, none had an Adjusted CarAV/Yr over 0.5, and five couldn't even break loose from zero: running back Siran Stacy at No. 48 (-2.7 VAE), running back Tony Brooks at No. 92 (-1.9 VAE), return specialist Jeff Sydner at No. 160 (-1.2 VAE), guard William Boatwright at No. 187 (-1.1 VAE), and linebacker Chuck Bullough at No. 214 (-0.9 VAE). Incidentally, Boatwright has the ignominious distinction of not even having his own PFR player page -- and yet he was somehow the Eagles third-best pick behind Bullough and fifth-round defensive back Corey Barlow (-1.0 VAE). For the sake of Eagles fans reading this, I'll stop there instead of bringing up their literal waste of a pick at 102nd overall.
In terms of absolute best, things aren't as clear, although it seems to me there are four worthy candidates: 1986 49ers, 1975 Bears, 1971-1972 Bengals, and 1974 Steelers. The first and last of those are in all three tables, as well as among the NFL Network's top 10. I tend toward ranking them as the top two -- in no particular order -- because it's hard to argue against Pittsburgh's four Hall of Famers, and San Francisco's six picks in the third and fourth round were Tom Rathman (No. 56), Tim McKyer (No. 64), John Taylor (No. 76), Charles Haley (No. 96), Steve Wallace (No. 101), and Kevin Fagan (No. 102). That's just ridiculous value (+16.9 VAE) to get in the span of 47 mid-round picks.
The Bengals are obviously my "wild card," pun very much intended, which leaves us with the 1975 Bears. The two words almost always associated with that draft class were "Walter Payton:" He returned +4.0 VAE as the fourth overall pick. However, five other words we should associate with the value of Chicago's 1975 draft class are "Joe Theisman of Testicle Trauma," aka fourth-rounder Virgil Livers. In a five-year career, Livers started at cornerback each season, and ranked ninth in punt return average as a rookie. As the No. 83 pick in an average overall draft, that's good enough for +4.2 VAE. I'm not silly enough to argue that Livers' VAE makes him a more valuable pick than Payton, but I will argue Livers deserves to be known for something other than an unintended PSA for male athletic supporters.
That's it for today, and that's it for VAE/ROI splits for franchises. The plan for the remaining two weeks before draft day is to go granular, culminating with the best and worst individual picks.
86 comments, Last at 01 May 2013, 8:10am by dmstorm22