Twitter-ific Preview: MAC and the Independents
I, for one, was a bit surprised by the MAC projections my predictor model spit out. It projects 6 teams (almost half the league) to finish even in conference play, and 10 of the 13 teams to finish within a single game of the .500 mark. Who do they think they are, the ACC? I thought I'd use the extended part of this post to do a little more in-depth explanation of why this is so.
For starters. the MAC West is projected to be much stronger than the East. Of their 18 games against East foes, the system projects the West to win 12 of them (or 2/3rds). That's why every team in the West is projected to have at least a .500 conference record. Now I'll go team by team and offer a little insight into the projection model.
Western Michigan--Broncos were 6-2 last season, but won three league games by a combined 16 points (one in OT). Their two losses (to Central Michigan and Ball State) were by double digits. The projection model is also wary of a defense, average by league standards last season, that returns only 3 starters.
Northern Illinois--Huskies paired a weak offense (10th in the conference in yards gained) with the league's best defense to go 5-3 in 2008. The model forsees slight progression by the offense (7 starters return) coupled with a bigger regression by the defense (4 starters return).
Eastern Michigan--The system likes the pedigree of the head coach (Ron English) to help the defense go from awful to presentably bad and the 9 returning starters on offense to help the Eagles improve by 2 games in the standings.
Ball State--The system hates the losses on offense (NFL-caliber quarterback and four offensive linemen including 3 All-Conference players) and sees the defense declining slightly. A 4 game decline may seem a bit steep, but remember the Cardinals play in the much stronger MAC West.
Toledo--The system likes the veteran offense to go from bad to average and the defense to maintain its position (8th in the league last season in yards allowed), making for a 2 game improvement.
Akron--Zips had only the 6th ranked offense last season, but 2nd best in the East. The system sees them as the best in the East this season with 9 starters, including a senior quarterback returning. It predicts only marginal improvement from a bad defense (10th in yards allowed last season) for a 1 game improvement.
Kent State--The best offense in the East loses its quarterback, but brings back 4 starting offensive linemen and starting running back in a run-based offense. The defense also returns 7 starters so the system likes them to hold their place. SDPI, the statistic on which the majority of the predictor model is based, also saw the Flashes as much better then their 3-5 league mark last season.
Temple--The Owls were 4-4 last season, but the league's worst offense loses its senior quarterback. An average defense returns 9 starters, so improvement is likely on that end. Owls must also face 2 road games against the West.
Buffalo--System is extremely wary of Bulls great turnover margin (+10 in league play last year) and their great luck recovering fumbles. Bulls also lose senior quarterback Drew Willy. The system does like the defense to improve with 8 starters back.
Bowling Green--The manic-depressive projection. On the one hand, the Falcons lost their 4 league games by a combined 19 points and boasted the second-best point differential last season. On the other hand, the strength of last year's team, the defense (2nd in the conference in yards allowed last season) returns only 3 starters. Add a new coach plus 3 games versus the West, and the result is a losing record.
How good is Jimmy Clausen? Clausen, you may remember, was a much balleyhooed recruit who was thrown to the wolves his freshman season playing behind a young and bad Notre Dame offensive line. The results were not pretty. Clausen averaged just a shade over 5 yards per pass as a freshman with 7 touchdown passes and 6 interceptions as the Irish suffered through a miserable season wherein they finished 3-9 and averaged only 242 yards per game (dead last in the nation). Calusen rebounded as a sophomore, playing with a more experienced team, throwing 25 touchdowns and 17 interceptions, while ranking a respectable 43rd in the nation in quarterback rating. The team improved as well, going 7-6 and ranking as a middle of the road offense, averaging 355 yards per game (65th in the nation). So is Clausen primed for even more improvement as a junior? The answer, unfortunately, is likely 'Yes'.
To answer this question, I devised a rough estimate of a similarity score. Similarity scores were invented by Bill James, noted baseball statistician, to estimate a player's future performance based on how similar players have performed in the past. The similarity score formula I used is a little rough (and will likely be tweaked in the offseason), but is listed below.
100-[(difference in pass attempts/10)+(difference in completion %)+(difference in yards per pass*2)+(difference in touchdown passes/3)+(difference in interceptions/3)+(difference in QB rating/10)+(difference in rush attempts/10)+(difference in rush yards/50)+(difference in rushing touchdowns/3)
A higher score means a player is more similar, and vice-versa. What I wanted to do with this metric is calculate similarity scores for Clausen's freshman and sophomore seasons. To keep the measure more accurate, I only compared Clausen's freshman season to other freshman seasons, and likewise for his sophomore campaign. It makes no sense to compare Clausen's freshman season to other players' senior or junior seasons. Clausen started almost immediately as a freshman, posting relatively abysmal numbers. However, he has three more years to grow and progress. A junior or senior quarterback who posts similar numbers has no time to progress, and offers no insight into how Calusen's career will take shape. With that being said, here are the 5 most similar freshman quarterbacks to Jimmy Clausen ranked by similarity scores (I only looked at quarterbacks from the 2005-2007 seasons).
1. Drew Willy, Buffalo, 2005, 85.1
2. Rusty Smith, FAU, 2006, 83.5
3. Josh Freeman, Kansas State, 2006, 82.6
4. Matthew Stafford, Georgia, 2006, 79.7
5. Aaron Opelt, Toledo, 2006, 78.9
Some of these guys are pretty famous. Stafford and Freeman were selected in the 1st round of the NFL Draft in April. However, each player was pretty bad in his freshman season. None of the 5 quarterbacks posted a passer rating over 114. No quarterback threw more than 7 touchdown passes, and only Opelt had more touchdowns than interceptions (6-5). But amazingly, these guys all improved significantly as sophomores. Just like Clausen. Here's their cumulative average stat line as sophomores compared to Clausen's sophomore campaign (the final 3 columns are rushing numbers).This similarity score seems to do a pretty good job of identifying players who are truly similar, and also of projecting what's to come. So what can we expect from Clausen as a junior? Here are the 5 most similar sophomore quarterbacks to Clausen (once again 2005-2007 only).
1. Rusty Smith, FAU, 2007, 84.0
2. Josh Freeman, Kansas State, 2007, 83.8
3. Curtis Painter, Purdue, 2006, 79.1
4. Matthew Stafford, Georgia, 2007, 77.4
5. Rudy Carpenter, Arizona State, 2006, 75.3
This list is very similar to the freshman list with two new names thrown in. Here's how those 5 quarterbacks performed collectively as juniors.Once again, we have significant improvement. The rushing numbers are a skewed quite a bit by Josh Freeman. After running for -40 yards and 4 touchdowns as a freshman, Freeman rushed for 400 yards and 14 touchdowns as a sophomore. Besides the rushing numbers, I would say this is a solid projection for Clausen's 2009 stat line. It looks like he is destined to improve even more as junior. Remember this when the media starts hyping Clausen as a darkhorse Heisman candidate in late-October.