Tuesday, September 4, 2012

Computers Don't Kill Polls, People Kill Polls

It always annoys me when people suggest the Bowl Championship Series is flawed because "computers decide championships".  The computers are far from perfect, mainly because the BCS wants to keep it that way.  For instance, they don't take into account the score of any games, only the results (win or loss).  However, computers only make up one fraction (1/3) of the equation while the Coaches' USAToday Poll and Harris Interactive Poll (both human polls) each make up another third.

For as much as the computer rankings take the heat, they really deserve it the least.  The computers are not biased towards teams or conferences.  They rely on the far more sound mathematics than the human based polls.  Never mind that the USAToday is filled out either by coaches who don't watch many games (because they are coaching their own), or graduate assistants who only have so much other responsibility.  The Harris Poll is made up of a bunch of former players, coaches, media personalities, and other randoms who have admitted in the past to being apathetic to voting.  Time after time, the system obviously fails, but it is the computer who still take the brunt of the insults.

This week, the biggest fault in the human polls goes to a component no longer associated with the BCS, but is equally regarded: the AP Poll.  This poll made up of journalists (because who better to make decisions in this country than journalists?) is independent of others, but still equally flawed and biased.  When looking at the polls each week, I always like to see the additional teams to receive votes that didn't crack the top 25.  This week, not only was I shocked when I saw the Texas State Bobcats listed, but that they received 10 points in the poll (good for 34th overall).

My suspicions were that this was one voter who voted Texas State high in the poll (16th) versus a couple of voters having a nervous breakdown and voting for them.  Sure enough, I was right, and the culprit was Ray Ratto of Comcast SportsNet Bay Area.  The only part that surprised me is that he had been writing in the Bay Area for over 30 years with little ties to the state of Texas or Texas State.

For those who aren't sure if they've heard of Texas State before, you're correct.  This is only the Bobcats first full season at the FBS level, essentially making this past week their first official game.  Yes, they did upset the Houston Cougars, but Houston is not a highly regarded team by any means this year.  This isn't an Appalachian State, North Dakota State, or James Madison either that is joining the higher ranks.  This is a team that last year as an FCS finished with a 6-6 record.

Based on his assessment, the 1-0 FBS Bobcats are a better team than Arkansas, Wisconsin, Michigan, West Virginia, and six other Big 12 teams that inhabit that same region.  Texas State isn't his only hiccup, just the most obvious one to someone who wasn't looking for a mistake.  Of the 64 other AP voters, no one ranked Florida State (3rd), Michigan State (5th), Nebraska (10th), and Notre Dame (15th) any higher than he did, not to mention Texas (9th) who only one voted higher.  In addition, each of the other 64 voters voted West Virginia as 15th or better, yet Ratto is by far the lowest vote at a pathetic 24th.  He is one of only four not to vote for Michigan this week, and voted LSU (7th) and Arkansas (17th), far below where almost all other voters had them.

Ratto's poll alone will have little affect on the grand scheme of things, and hopefully he receives enough nasty emails that point out why he should not be allowed to vote.  However, this is not an event that won't happen again.  There will be many more horrible ballots just this year.  If it hasn't already, it should become apparent that the biggest problem in the college football system is not the computers, but the human polls.

Class of Tony Banks or Kevin Kolb?

NFL Draft Classes vary from season to season for all positions, but no position is analyzed greater than that of the Quarterback.  My goal here wasn't to look simply at how much NFL quarterback talent came out of each draft.  The 2012 of Andrew Luck and Robert Griffin III can not and should not be measured against the class of 1983 (John Elway, Jim Kelly, and Dan Marino) until the careers of all those 2012 QBs have come to an end.

I'm more interested in the flip side of the equation as in which draft classes were so terrible that simply no one was competitive enough to remain in the NFL as a fixture.  A recent "failure" of a class would likely be the 2007 class.  Most notable for being a flop was the #1 overall pick in the draft, JaMarcus Russell.  However, the disappointments do not end there.  Of the 11 QBs taken in that draft, only one remains as an NFL starter, Kevin Kolb of the Arizona Cardinals, and even he is considered a significant disappointment.  John Beck and Tyler Thigpen as backup QBs on the Washington Redskins and Buffalo Bills respectively.  Meanwhile, the top two, Russell and Brady Quinn haven't seen an NFL field since 2009, and four others haven't seen it since 2010.

In this quick study, QB Rating was used to grade how bad each draft class was.  I say "how bad" because itt isn't a great measure to determine how good some classes were.  Of the 32 years analyzed (1980-2011), no one would argue that one of the very best if not the best was 1983.  However, by QB rating, it ranks only 8th overall in the past 32 years.  Instead, it was beat out by a far lesser class like 1991.  The rating for 1991 is higher because a lone player, Brett Favre, weighed the class by himself since so many others didn't last long in the NFL.  Similar results can be found for 1998 where Peyton Manning and Matt Hasselbeck far out weigh the poor play of Ryan Leaf and Jonathan Quinn.

While it might not determine the best classes, I feel it does a good job of pointing out the classes that had little to no NFL talent and that is really the goal here.  The 1991 class might have been poor overall, but an NFL fixture like Favre was discovered.  Most classes will generally find one NFL starting QB that stands the test of time for seven to ten years at least.  One last catch is that these QB ratings are for the entire careers of the players in those draft classes.  Therefore, the more recent ones, think 2008-11, are going to likely vary significantly in the years to come.  Meanwhile, the previous ten years will likely have little to no adjustments while prior to that will likely never change due to the majority of those players having already retired.

The career QB ratings of each draft class can be found in this chart with the most notable classes labeled individually:
You'll notice the first couple seasons were rather poor.  In fact, 1980 and 1982 were two of three worst seasons on record.  However, as mentioned, things rebounded the following season with the tremendous class of 1983.  In addition to the well known names of Elway, Marino, and Kelly, you had solid to decent contributors like Ken O'Brien and Tony Eason.  The decade following the 1983 draft was rather steady in terms of production.  You had a high point with the aforementioned 1991 draft including Brett Favre, and a low point in 1986 where Jim Everett and Mark Rypien were the best the college world had to offer.  Note that in the mid 80s, the NFL had competition with the USFL.

If was after this 11-year steady period that the draft classes took an unexpected turn for the worst.  Between 1994-97, three horrendous classes were produced.  The lone exception, 1995, was still worse than each of the class from 1983-93 with the exception of 1986.  Most of the seasons still produced NFL regulars.  Names like Trent Dilfer, Steve McNair, Kerry Collins, and Jake Plummer before his sudden retirement were long time starting QBs.  The exception was in 1996 where the NFL Draft produced...Tony Banks?  Bobby Hoying?

The 1996 class is definitely the low point.  Banks is the only player to throw for more than 10,000 yards in his career (15,315).  Danny Kannell is behind him with 5,129 yards. To compare, Cam Newton is already at 4,051 yards in his career following only his rookie season.  Sam Bradford has eclipsed that mark already in only two seasons.  The percentage of passes completed from that draft class is a lowly 53.4%, second worst only to 1980.  One piece of trivia, Banks set the record for most fumbles as a rookie in 1996 before having the mark eclipsed by Kerry Collins and Daunte Culpepper in the future.

Despite how bad things got through the four years, the more remarkable part is the rebound from 1998-2001.  Those four seasons make up four of the seven best totals.  Again, not a great measuring stick for "the best", but it shows the remarkable improvement compared to the previous four seasons.  Since then, results have been extremely variable year to year.  2002 was among  the poorer seasons led by David Garrard and Joey Harrington followed only two years later in 2004 by one of the very best.

More recently, we've been plagued by the 2007 class, and to be honest, the 2010 class.  Sure there was Sam Bradford who had a great rookie season, but his sophmore campaign was as poor as his health.  He certainly isn't terrible, but thus far has produced middle of the road results while being the only remaining starter from a group including Tim Tebow, Jimmy Clausen and Colt McCoy.

The 2011 class has showed far more promise in their rookie seasons, and it will be interesting to see if they can take that any further this year.  In fact, the 2011 draft class nearly surpassed the 2010 class in half as many seasons.  They could pass the terribly 2007 class by the end of this season.

Notable Quarterbacks By Draft Class:

1980:  Marc Wilson (Oakland Raiders), Mark Malone (Pittsburgh Steelers), Eric Hipple (Detroit Lions)

1982:  Jim McMahon (Chicago Bears), Mike Pagel (Baltimore Colts)

1983:  John Elway (Baltimore Colts), Jim Kelly (Buffalo Bills), Tony Eason (New England Patriots), Ken O'Brien (New York Jets), Dan Marino (Miami Dolphins)

1991:  Todd Marinovich (Los Angeles Raiders), Brett Favre (Atlanta Falcons)

1994:  Trent Dilfer (Tampa Bay Buccaneers), Gus Frerotte (Washington Redskins)

1995:  Steve McNair (Houston Oilers), Kerry Collins (Carolina Panthers), Kordell Stewart (Pittsburgh Steelers)

1996:  Tony Banks (St. Louis Rams)

1997:  Jake Plummer (Arizona Cardinals), Danny Wuerffel (New Orleans Saints)

1998:  Peyton Manning (Indianapolis Colts), Ryan Leaf (San Diego Chargers), Brian Griese (Denver Broncos), Matt Hasselbeck (Green Bay Packers)

2000:  Chad Pennington (New York Jets), Marc Bulger (New Orleans Saints), Tom Brady (New England Patriots)

2004:  Eli Manning (San Diego Chargers), Philip Rivers (New York Giants), Ben Roethlisberger (Pittsburgh Steelers), Matt Schaub (Atlanta Falcons)

2007:  JaMarcus Russell (Oakland Raiders), Brady Quinn (Cleveland Browns), Kevin Kolb (Philadelphia Eagles), Trent Edwards (Buffalo Bills)

2008:  Matt Ryan (Atlanta Falcons), Joe Flacco (Baltimore Ravens), Chad Henne (Miami Dolphins)

2011:  Cam Newton (Carolina Panthers), Jake Locker (Tennessee Titans), Christian Ponder (Minnesota Vikings), Andy Dalton (Cincinnati Bengals)

Monday, September 3, 2012

Competition vs. Punching Bags

Now that nearly all teams have completed one game of the 2012 season, I think it is worth it to mention the teams who consistently schedule a real team for this time of year, and not simply a punching bag.  Looking through past schedules, there are only a dozen teams across the nation who have scheduled a BCS team to open their schedule consistently over the past three schedules.  Not surprisingly, only one of these programs is an actual BCS program itself, with the majority of teams being from the lower level MAC, Sun Belt, and Conference-USA.  Despite the low percentage of teams that consistently face these teams to open the year, it often seems worth it since I feel these games often provide the best chance for an upset.

For instance, Boise State has done this for years, and has come out on top more often than not by preparing more during the off-season for that one game.  Boise State has faced a BCS team the past four seasons, and has won three of those meetings.  Unfortunately for them, their winning streak came to an end this weekend when they lost as Michigan State.  Their wins included Oregon in 2009, Virginia Tech in 10, and Georgia in 11.

There are only two programs that currently have opened the season against BCS opponents, and hold a perfect record while doing so.  The longest winning streak is held by BYU who like Boise State went into their four consecutive opener against a BCS opponent.  They defeated the Washington State Cougars 30-6 on Friday after seasons of winnings against Oklahoma in 09, Washington in 10, and Ole Miss in 11.

The other program on this list sporting a winning streak is the aforementioned, single BCS team, Northwestern of the Big 10.  Their streak is only at three games currently with wins against Vanderbilt in 10, Boston College in 11, and most recently Syracuse.  On a side note, LSU beat a non-BCS team in North Texas, but had a streak running into this season at 3-0 themselves including wins over Washington, North Carolina, and Oregon.

It is worth noting that the majority of teams who consistently open against a BCS teams lose.  Western Michigan currently holds the longest streak, and is 0-8 over that time.  Memphis and Western Kentucky both snapped their streak of opening against BCS teams, but were 1-7 and 0-6 respectively.  Miami (OH) and San Jose State are 0-5 and 0-4 themselves.

Props to the teams who consistently schedule these tougher teams, and then are competitive with them each season.  Added props to teams like BYU, Northwestern, and until this year Boise State and LSU who consistently beat these teams, particularly Northwestern and LSU who earn wins that really do not count much to their overall season in the grand scheme of things.  As for BYU and their streak, the next two seasons they have Texas scheduled for Week 2 of the season with no opponent designated for Week 1.