Tuesday, April 29, 2008

Kicking Luck


Football teams exercise much less control over what their opponents are able to do than most fans realize. From recovering fumbles to illegal shifts, random chance exerts a profound influence on the outcome of a football game. Perhaps one of the most random events in football is a field goal attempt by the opposition. Outside of a blocked kick, the defense is virtually unable to influence whether the football passes between the goal posts. With that in mind, which teams were the luckiest and least lucky in terms of opponent's field goal accuracy in 2007?

To determine which teams had the best and worst luck, we need to find what the out how often the 'average' opposing kicker succeeded in converting a field goal attempt. In 2007, kickers who kicked against Division IA (or FBS) foes made 1814 of 2530 kicks, for a field goal percentage of 71.7%. I for one am surprised by the seemingly high percentage of successful field goal attempts. It should also be noted that some Division IAA (or FCS) kickers are included in this list as it is based on opponents of Division IA teams. Anyway, to calculate what I have dubbed Kicking Luck Points (KLP), simply multiply the average field goal percentage by the number of field goals a teams' opponents attempted. Subtract this number from the field goals their opponents made and multiply by 3 (the point value of a field goal attempt). This resulting number is the number of points the team allowed above or below the national average. Confused? Here is the formula in action:

In 2007, opponents of the UTEP Miners made 16 of 23 field goal attempts in games against the Miners. That's a field goal percentage of 69.6%, just a bit below the national average. We would have expected UTEP's opponents to make about 16.49 field goals. Subtracting 16.49 from 16 leaves us with -.49. Multiplying this number by 3 produces -1.47. That means UTEP allowed about one and a half fewer points over the course of the whole season than would be expected if they had perfectly average luck in opponent field goal percentage. 1.47 points over the course of a 12-game season is pretty negligible. I used UTEP as an example because their luck was about average. Now, here are the 10 best KLP teams from 2007.Northwestern enjoyed the best luck in opponent field goal accuracy of any team in 2007. Their 12 opponents connected on only 4 of 12 field goal attempts, resulting in the team allowing almost 14 fewer points than expected if their opponents had average kicking accuracy. And now the other side of the coin, the 10 worst KLP teams from 2007.California allowed about 15 points more over the course of their 13-game schedule than they would have if their opponents had made field goals at an average clip. The difference between the team with the best luck and the team with the worst luck was a little more than 29 points. 29 points over a short season can be pretty significant, especially depending on when those points occur.

What does this mean for 2008? That's a good question. Northwestern and California aren't likely to see such divergent luck in their opponents' field goal accuracy, but to proclaim that all the teams that enjoyed good luck will decline and vice-versa is pure folly. In addition, I did not adjust the kicking accuracy for distance or account for weather variables. In the end, KLP is just a fun stat that may not prove particularly useful in predicting future performance.

Tuesday, April 22, 2008

MAC Look-Ahead: SDPI

One of my favorite set of posts from last offseason was the conference recap using SDPI. SDPI is a statistic I borrowed from Eddie Epstein that he used in his book, Dominance, to rank pro football's all-time greatest teams. SDPI stands for Standard Deviation Power Index and looks at how teams performed relative to the league average (or conference average in this case) and standard deviation in terms of points scored and allowed. The more standard deviations a team is above the mean, the better they are, and vice-versa. Here is the link to last year's MAC post. As you can see, SDPI was a useful tool in predicting some of the rise and fall among the MAC's teams. In the interest of providing an even better offseason analysis, I will now be conducting another SDPI, this time for yardage. It is calculated in the same manner as the SDPI for points, but will obviously be measured against the conference mean and standard deviation for yards. Think of it this way: Points are the end result and yards are the means to that end. Thus, looking at both sets of data, we can get an even better idea about which teams are likely to improve or regress in 2008.

First off, I'd like to extend my middle finger to the MAC and their juvenile scheduling practices in 2007. If you didn't know, the MAC added Temple last season, boosting their membership to 13 teams. That wasn't the problem. The problem was that they decided to let some teams play 7 conference games, and others play 8. Not even every team in each division played the same number of games. The MAC decided to user inter-division games to determine which teams played in the MAC Championship Game. Those intra-division games? According to the MAC, those are pretty much glorified exhibitions. But the real reason the schedule irked me was because of the performance analysis aspect. Since each team did not play the same number of conference games, the numbers have to be averaged together. Instead of calculating SDPI by total points and total yardage, it is done based on averages. Thankfully, the MAC has gone to a complete 8-game schedule for everyone in 2008.

If you want the meat of the article, skip this next paragraph as it just gives an example of how the SDPIis calculated. The mean points scored for all MAC teams in conference play (Championship Game excluded) was 28.17 per game. The mean points allowed for all MAC teams in conference play (Championship Game excluded) was 28.07 points per game. The difference is due to rounding. The standard deviation for points scored was 7.42 per game. The standard deviation for points allowed was 5.31 per game. Bowling Green scored 33.875 points per game in MAC play and allowed 27.375. Their offensive SDPI was 0.77 = ([33.875-28.17]/7.42). Their defensive SDPI was 0.13 = ([28.07-27.375]/5.31). Their total SDPI for points (SDPIP) was 0.90 which ranked 4th in the conference. The mean yardage for all MAC teams in conference play (Championship Game excluded) was 394.75 yards per game. The mean yardage against all MAC teams in conference play (Championship Game excluded) was 393.05 yards per game. The difference is due to rounding. The standard deviation for yardage for was 60.20 per game. The standard deviation for yards allowed was 42.70 per game. Bowling Green gained 407.38 yards per game in conference play and allowed 405.75 per game. Their offensive SDPI was 0.21 = ([407.38-394.75]/60.20). Their defensive SDPI was -0.30 = ([393.05-405.75]/42.70). Their total SDPI for yards (SDPIY) was -0.09 which ranked 7th in the conference.

To refresh your memory, here are the 2007 MAC Standings.
Now here are the 2007 SDPI standings.

See why the MAC scheduling sucks? Miami of Ohio finished 5-2, technically a half game behind Bowling Green, yet still won the division thanks to a better record against divisional teams.

Best Offense: Central Michigan 1.98 (SDPIP) and 1.94 (SDPIY)
The offense didn't miss a beat wit the departure of Brian Kelly.

Worst Offense: Temple -1.44 (SDPIP) and -1.78 (SDPIY)
Held below 10 points 3 times in conference play.

Best Defense: Miami 1.79 (SDPIP) and 2.29 (SDPIY)
The Redhawks were outstanding in the regular season, allowing only 295 yards and 18.6 points per game to their MAC opponents. Central Michigan proceeded to shred them for 452 yards and 35 points in the MAC Championship Game.

Worst Defense: Toledo -1.87 (SDPIP) and -1.35 (SDPIY)
The second-best offense in the conference was wasted by a defense that couldn't stop anyone.

Schedule strength is not evaluated thanks to the MAC's decision to have some teams play 7 conference games, while others played 8, and to only count division games in the standings.

Looking ahead to next season, the prohibitive favorite should be...

East
Look to the Bottom of the Standings
In 2005, Ohio went 3-5 and finished in 4th place in the MAC East. Kent State was 0-8 in MAC play in 2005 ad dead last in the MAC East. In 2006, those teams finished 1st and 2nd respectively in the MAC East with a combined conference record of 12-4, with Ohio winning the division at 7-1. In 2006, Bowling Green finished 3-5 and tied for 3rd in the MAC East. Miami finished 2-6 and alone in 5th place. In 2007, both schools combined to finish 11-4 in MAC play and atop the standings. Look for a similar phenomenon to occur in 2008. Ohio (4-4) and Kent State (1-7) are two teams to look out for in 2008.

West
Central Michigan
What Brian Kelly hath begun, let not Butch Jones tear asunder. Jones did an admirable job replacing Kelly, leading the Chippewas to an 8-6 record, another MAC title, and a near upset of Purdue in the Motor City Bowl. The Chips once again dominated their MAC opponents, going 6-1 in the conference (7-1 if we count the MAC Championship Game) after a 7-1 season in 2006. However, there were some rough patches outside the league. Kansas beat them by 45, Purdue beat them by 23 in the first meeting, Clemson hung 70 on them, and Division IAA North Dakota State beat them by 30. The Chips have a host of positive indicators as they begin their quest to become the first 3-peat MAC Champions since Marshall completed a 4-year run in 2000. For one, the coaching staff is intact. Despite interviewing for the vacant West Virginia job, Butch Jones is back in the fold to begin his second season as head coach. Having a 3rd head coach in as many seasons would have made the 3-peat quest much more difficult. But perhaps most importantly, Dan LeFevour is back for his junior season. The poor man's Tim Tebow (+3600 yards and 27 TD's through the air and +1100 yards and 19 TD's on the ground) is back along with 8 other starters to lead an offense that should once again be the best in the conference. The defense which was rather porous in 2007 (9th in the MAC in yards allowed in conference play) returns 8 starters and merely needs to be competent to give the Chips a chance to win every week (against MAC foes). The Chips also get two of the better contenders in the MAC West (Western Michigan and Ball State) at home. However, aside from Eastern Michigan and Northern Illinois, any one of 4 teams (the aforementioned trio plus Toledo) could realistically be expected to walk away with the MAC West title.

The team(s) that will improve are...

Kent State, Northern Illinois, Ohio, and Western Michigan
Despite their 3-9 finish, solitary MAC win, and 7-game losing streak to end the season, Kent State has a real chance to gain bowl-eligibility and perhaps even get to the MAC Championship Game in 2008. The Golden Flashes, who had tough luck in just about every category in 2007, should see their fortunes improve in 2008. They were 1-4 in one-score games and had a turnover margin of -3 in conference play (8th in the MAC). Plus the starting quarterback, Julian Edelman, missed the last 4 games. Edelman should begin the 2008 in good health, and if he can remain that way, the Golden Flashes should rise in the standings. Like their conference brethren in Kent, Ohio, the Huskies of Northern Illinois managed only one conference win in 2007. That number should double, likely triple, and perhaps quadruple in 2008. Longtime head coach Joe Novak has departed and Jerry Kill has stepped in to take his place. Kill had a very successful run in Carbondale at Southern Illinois, where he made 5 postseason appearances in 7 seasons (5 straight to close his career). At Southern Illinois, his specialty was offense. The Salukis averaged at least 34 points per game every season save his first. Northern Illinois finished 9th in the MAC in yards per game in conference play, but they were even worse defensively finishing 12th in yards allowed per game. The deplorable defense gets all 11 starters back and should improve by default, while the offense should improve under Kill with the starting quarterback, running back, and top 5 receivers returning. The Huskies were also a tough luck 2-5 in one-score games (1-4 in conference play) and had a MAC-worst -10 turnover margin in conference play. Both of those facets of the team should also see a market correction and leave the Huskies as one of the nation's most improved teams in 2008. After a sensational 2006 campaign that saw them win the MAC East, the Ohio Bobcats fell back to the pack and finished in a tie for 4th place in 2007. A lot of things went right for the Bobcats in 2006. They were 3-0 in one-score games and finished 9-5 despite being outgained on the season. Based on how well they played, their 9-5 record was a bit of a mirage. The universe adjusted in 2007, and despite keeping their play at nearly the same level, the team stumbled to a 6-6 record (still good by recent Ohio standards). Though the defense regressed from one of the MAC's best to mid-level, the passing game experienced a significant uptick. After averaging only 5.5 yards per pass attempt in 2006, the Bobcats raised that number by almost 2 yards to 7.4 in 2007. While the player responsible for most of those passes, Brad Bower, is gone, his replacement, Theo Scott, received significant playing time in 5 of the team's 12 games last season so inserting him into the lineup this year won't be akin to throwing him to the wolves. On the whole, the offense gets 8 starters back, although the unit does lose running back Kalvin McRae. He of the 3 consecutive 1000-yard seasons. The defense gets 7 starters back and should improve upon last year's version. The Bobcats also had a turnover margin in MAC play of -5. Only 3 teams (Akron, Western Michigan, and Northern Illinois) were worse. Their turnover margin should be better this year as turnovers are highly variable. In the wide-open MAC East, Ohio has a decent shot of returning to the MAC Championship Game. Big things were expected of the Western Michigan Broncos in 2007. A lot of publications had them taking the MAC West. Unfortunately, a tough non-conference schedule (including top-10 teams in West Virginia and Missouri), a poor record in one-score games (1-3), and a poor turnover margin (-6 in MAC play) caused them to finish below .500 in conference play. The most gut-wrenching close loss can be viewed (in all it's grainy detail) below.
Going into 2008, the Broncos should challenge Central Michigan not only for the MAC West title, but also for the best offense in the conference. Quarterback Tim Hiller missed all of the 2006 season with an injury and struggled somewhat upon his return to the team in 2007. However, he closed with a bang (see below), and with his two-top receivers (Jamarko Simmons and tight end Branden Ledbetter) and starting running back (Brandon West) returning, the Broncos should be able to move the ball quite efficiently. The defense loses only one contributor in safety Anthony Gebhart, and should again be one of the league's best. The only fly in the proverbial ointment is that the Broncos must travel to Mt. Pleasant to take on the Chippewas of Central Michigan. Otherwise the Broncos would be the favorites to take the MAC West in 2008.

The team(s) that will decline are...

Ball State, Buffalo, Eastern Michigan, and Temple
2007 was a special season for the Ball State Cardinals. The prolific offense, led by quarterback Nate Davis, guided the team to 7 wins, their first bowl game since 1996, and a near epic upset of Nebraska in Lincoln. That game was the first real indication that the Husker defense had serious issues. For 2008, Davis and the other 10 starters return on offense, so they should crank out the points. The defense also gets 8 starters back so it appears Ball State should in for a big-time season in 2008. Were it not for the Cardinals' tremendous turnover margin in 2007, I'd be inclined to agree. Ball State was an astounding +12 in MAC games last season. They accumulated that tremendous margin in only 7 conference games, meaning the Cardinals averaged nearly two fewer turnovers per game than their conference opponents. How did they do this? Nate Davis simply did not throw interceptions. As a team, the Cardinals threw 6 on the year (Davis accounted for all of them). Only 3 teams in the NCAA threw fewer. Fresno State, Air Force, and Navy each threw 5. However, Navy attempted only 136 passes on the year, Air Force 214, and Fresno State 343. Ball State put the ball in the air 483 times. Davis is a great quarterback, but expecting him to repeat that extremely low interception percentage is folly. In 2006, Davis threw 8 interceptions in 245 attempts. That's a great ratio (3.3 per 100 passes), but much lower than his other-worldly 2007 numbers (1.3 per 100 passes). Despite the returning personnel, Ball State will not enjoy such an extreme turnover margin in 2008, and will not be able to win the extremely tough MAC West. Another team that enjoyed a special season in 2007 was the Buffalo Bulls. Since returning to Division IA football in 1999, the Bulls had won 12 games prior to the 2007 season (never winning more than 3 in any one season), while losing 79. The Bulls proceeded to go 5-7 and actually finished with a winning record (5-3) in the MAC under second-year head coach Turner Gill. Instead of acting as a stepping stone season, it appears the Bulls will fall back in 2008, in the hopes of taking another step forward in 2009. Why? For starters, while the team won more than they lost in MAC play, they were not technically a 'good' team. The Bulls were outgained by about 31 yards per game in conference play. Those are not numbers befitting a team that wins more than half their games. The Bulls were greatly aided by their turnover margin of +4 in conference play (3rd best in the MAC). With 10 starters back on offense and 8 back on defense, the Bulls will probably improve performance-wise, but they won't post a winning conference record. Eastern Michigan quadrupled their win total from 2006, posting a 4-8 record and finishing in a tie for 3rd in the MAC West. The Eagles closed the year with a huge road upset of eventual champion Central Michigan. The Eagles, like the Buffalo Bulls, were consistently outgained (on average by about 41 yards per game in MAC play), but thanks to their phenomenal turnover margin of +11 (2nd in MAC play), they were able to tread water. In 2008, that turnover margin should normalize and the Eagles will return to their rightful place at the bottom of the MAC West. Yet another perennial also-ran achieved moderate success in 2007, as the Temple Owls won 4 conference games despite an offense more inept than the cops in Hazard County. Temple averaged only 288 yards per game in conference play, the worst performance by a wide margin. The second worst team (Akron) averaged 346 yards per game. The Owls made up for it by playing very good defense (344 yards allowed per game in conference play--2nd best), winning the close ones (3-1 in one-score games), and winning the turnover battle (+2 in conference play). 3 of the Owls 4 conference wins were by a combined 12 points. In the one close game the Owls lost (22-17 to Connecticut), they should have won if not for a blown call that was also blown upon review.
While the Owls return a host of personnel on offense and defense, that appears to be a common refrain for every team in the MAC. Because of their good fortune last season, and because of the strength of the rest of the conference, Temple will have trouble matching last season's 4 MAC wins.

Sunday, April 13, 2008

Sun Belt Look-Ahead: SDPI

One of my favorite set of posts from last offseason was the conference recap using SDPI. SDPI is a statistic I borrowed from Eddie Epstein that he used in his book, Dominance, to rank pro football's all-time greatest teams. SDPI stands for Standard Deviation Power Index and looks at how teams performed relative to the league average (or conference average in this case) and standard deviation in terms of points scored and allowed. The more standard deviations a team is above the mean, the better they are, and vice-versa. Here is the link to last year's Sun Belt post. As you can see, SDPI was a useful tool in predicting some of the rise and fall among the Sun Belt's teams. In the interest of providing an even better offseason analysis, I will now be conducting another SDPI, this time for yardage. It is calculated in the same manner as the SDPI for points, but will obviously be measured against the conference mean and standard deviation for yards. Think of it this way: Points are the end result and yards are the means to that end. Thus, looking at both sets of data, we can get an even better idea about which teams are likely to improve or regress in 2008.

If you want the meat of the article, skip this next paragraph as it just gives an example of how the SDPI is calculated. The mean points scored and allowed for all Sun Belt teams in conference play was 195.88 points. The standard deviation for points scored was 41.39. The standard deviation for points allowed was 48.83. Louisiana Monroe scored 176 points in Sun Belt play and allowed 166. Their offensive SDPI was -0.48 = ([176-195.88]/41.39). Their defensive SDPI was 0.61 = ([195.88-166]/48.83). Their total SDPI for points (SDPIP) was 0.13 which ranked 4th in the conference. The mean yardage for and against for all Sun Belt teams in conference play was 2749.13 yards. The standard deviation for yardage for was 401.89. The standard deviation for yards allowed was 310.12. Louisiana Monroe gained 2911 yards in conference play and allowed 2825. Their offensive SDPI was 0.40 = ([2911-2749.13]/401.89). Their defensive SDPI was -0.24 = ([2749.13-2825]/310.12). Their total SDPI for yards (SDPIY) was 0.16 which ranked 4th in the conference.

To refresh your memory, here are the 2007 Sun Belt Standings.

Now here are the 2007 SDPI standings.
Once again, a team that statistically dominated the league lost out on a Sun Belt conference title due to a close loss in the final game. In 2006, Middle Tennessee State had outscored it's conference brethren by almost 19 points per game before losing by a single point to Troy in the regular season finale. Troy and Middle Tennessee shared the Sun Belt title, but Troy had the tiebreaker and garnered the New Orleans Bowl berth. Thanks to a dearth of bowl-eligible BCS member teams, Middle Tennessee was invited to play in the Motor City Bowl. The karmic justice of college football played out in 2007, as Troy dominated their first 6 Sun Belt foes, winning by almost 26 points per game and even shocking Oklahoma State of the Big 12. But alas, in the season finale the Trojans fell to Florida Atlantic 38-32 and were forced to share the Sun Belt title. Florida Atlantic earned the New Orleans Bowl bid thanks to their head-to-head win over the Trojans, and Troy was denied an invite to postseason play despite an 8-4 record and a Sun Belt co-championship.

Best Offense: Troy 1.43 (SDPIP) and 1.39 (SDPIY)
The Trojans did not score fewer than 24 points or gain fewer than 368 yards against Sun Belt opposition. As a point of reference, Louisiana Monroe (the 5th most proficient scoring offense in Sun Belt play) averaged a shade over 25 points per game. North Texas (the 5th most proficient total offense in Sun Belt play) averaged 383 yards per game.

Worst Offense: Florida International -1.13 (SDPIP) and -1.47 (SDPIY)
It;s hard to not improve after going 0-12. The Panthers averaged only 21.2 points per game in conference play, but that was nearly 3 times what they averaged in those same games in 2006 (7.7 points per game).

Best Defense: Troy 1.84 (SDPIP) and 1.39 (SDPIY)
Another clean sweep for the Trojans. If not for their in-conference turnover margin of -6 (tied for last with North Texas), the Trojans would have been even more dominant.

Worst Defense: North Texas -1.13 (SDPIP), Louisiana Lafayette -1.30 (SDPIY)
Florida International avoided the cellar. Hoo-rah.

Looking ahead to next season, the prohibitive favorite should be...

Florida Atlantic
Rusty Smith, the best quarterback in the Sun Belt, is back and though it will be tough to improve on his outstanding 2007 season, a little decline shouldn't hurt the Owls bottom line. 9 other starters return to what should be the league's top offense, including senior running back Charles Pierre, and every receiver who caught at least 5 passes. The defense has 8 starters back and should remain one of the better units in the Sun Belt. Still, there are a few causes for concern. For starters, the Owls have only 4 true home games, and their non-conference schedule is dotted with trips to Texas, Michigan State, and Minnesota (who will surely be out for revenge). While those games obviously won't count in the Sun Belt standings, it's vital the Owls do not lose confidence if they open the season 1-3 before playing their first conference game. Fortune also smiled on the Owls last season as their turnover margin in conference play (+10) ranked second behind Middle Tennessee State, and they were 4-1 in one-score games (3-1 in Sun Belt play). The vagaries of turnover margin and close games could dent the Owls title hopes this season. While the Owls are hardly a sure thing in the ever-changing Sun Belt, they are the best bet at becoming the first repeat champions (Troy lost the tiebreaker to Florida Atlantic last season) since North Texas won 4 Sun Belt titles in a row from 2001-2004.

The team(s) that will improve are...

Arkansas State
Yardage wise, the Red Wolves (formerly the Indians) were the third best team in the Sun Belt last year, behind only the juggernaut at Troy and the surging Owls of Florida Atlantic. But turnovers, as is often the case, did them in. Their turnover margin of -3 in Sun Belt does not seem terrible at first, but it ranked ahead of only Troy and North Texas (both at -6) in the Sun Belt. As a result they managed only 3 conference wins and finished tied for 5th with Louisiana Lafayette. Things could change quickly in 2008, as the Red Wolves have one of the league's biggest stars. With the departure of Calvin Dawson of Louisiana Monroe, the best running back in the Sun Belt now resides in Jonesboro, Arkansas. Entering his junior season for the newly christened Red Wolves, Reggie Arnold has racked up over 2100 yards on the ground. His backfield mate, junior quarterback Corey Leonard, also returns as do 4 of the top 6 pass-catchers. The biggest problem for the offense, despite the return of many skill position players will be the offensive line that loses 4 starters. Of course, that may not be a bad thing as the line allowed more sacks (39) than all but 8 teams last season. Defensively, the Red Wolves bring back 5 starters from the front 7, but lose all 4 starters from the secondary. Still, last season's second best Sun Belt defense (based on yards allowed), should remain competent. The biggest reason to expect big things from the Red Wolves is the schedule. While Arkansas State has only 3 conference home games, one of those games is against Florida Atlantic. A win there would give the Red Wolves a realistic shot at taking the conference crown, especially with two of their four conference road games coming against North Texas and Florida International, arguably the league's weakest teams. If the ball bounces their way, and if the offensive line jells, Arkansas State could well find themselves in New Orleans in late December. And another thing...

No matter your feelings on 'political correctness' or 'insensitivity', this...
is much cooler than this...

The team(s) that will decline are...

Middle Tennessee State
On the surface, 2007 looked like a pretty good rebuilding year for the Blue Raiders. One season after sharing the Sun Belt title with Troy, the Blue Raiders rebounded from an 0-4 start to finish 5-7, and in a tie for 3rd in the Sun Belt with a 4-3. But the roots beneath that veneer of success may not be as strong as they appear. Although the Blue Raiders scored 207 points in conference play (3rd in the league), they were not very proficient moving the ball on a down-to-down basis. Only one team in the Sun Belt (Florida International) gained fewer yards in conference play that Middle Tennessee State. The offense that you're likely to hear a lot about in some preseason magazines, featuring two exciting quarterbacks in Joe Craddock and Dwight Dasher, was actually well below average. To be fair, Dasher and Craddock were very efficient, as the Blue Raiders posted a combined pass efficiency rating of 136.42 (27th in the nation). However, when you consistently gain fewer yards than 6 of your 7 conference mates, it doesn't take much for the offense to begin to sputter. Also working against the Blue Raiders is the fact that their fantastic turnover margin (+11 in Sun Belt play) is unlikely to continue. If the turnovers don't come as often, the deficiencies of the offense will show up on the scoreboard and not just in the box score.

Step Up
The Sun Belt has been in existence for 7 seasons. It's league members have participated in 9 bowl games. Sun Belt teams have gone 3-6 in those 9 games. Those certainly aren't SEC numbers to say the least. However, there has been improvement. From 2001-2004, Sun Belt teams were 1-4 in bowl games, losing by an average of nearly 13 points per game. More recently, from 2005-2007, Sun Belt teams have managed a 2-2 mark, winning by an average of 3 points per game. Here's how the Sun Belt compares with the MAC in bowl games over those two separate time frames. So has the MAC justified their bowl tie-ins or does the Sun Belt deserve to snatch one up?

Saturday, April 05, 2008

C-USA Look-Ahead: SDPI

One of my favorite set of posts from last offseason was the conference recap using SDPI. SDPI is a statistic I borrowed from Eddie Epstein that he used in his book, Dominance, to rank pro football's all-time greatest teams. SDPI stands for Standard Deviation Power Index and looks at how teams performed relative to the league average (or conference average in this case) and standard deviation in terms of points scored and allowed. The more standard deviations a team is above the mean, the better they are, and vice-versa. Here is the link to last year's Conference USA post. As you can see, SDPI was a useful tool in predicting some of the rise and fall among the Conference USA's teams. In the interest of providing an even better offseason analysis, I will now be conducting another SDPI, this time for yardage. It is calculated in the same manner as the SDPI for points, but will obviously be measured against the conference mean and standard deviation for yards. Think of it this way: Points are the end result and yards are the means to that end. Thus, looking at both sets of data, we can get an even better idea about which teams are likely to improve or regress in 2008.

If you want the meat of the article, skip this next paragraph as it just gives an example of how the SDPI is calculated. The mean points scored and allowed for all Conference USA teams in conference play (Championship Game excluded) was 267.75 points. The standard deviation for points scored was 52.79. The standard deviation for points allowed was 54.96. Rice scored 308 points in Conference USA play and allowed 340. Their offensive SDPI was 0.76 = ([308-267.75]/52.79). Their defensive SDPI was -1.31 = ([267.75-340]/54.96). Their total SDPI for points (SDPIP) was -0.55 which ranked 8th in the conference. The mean yardage for and against for all Conference USA teams in conference play (Championship Game excluded) was 3619.25 yards. The standard deviation for yardage for was 437.64. The standard deviation for yards allowed was 531.93. Rice gained 3702 yards in conference play and allowed 4176. Their offensive SDPI was 0.19 = ([3702-3619.25]/437.64). Their defensive SDPI was -1.05 = ([3619.25-4176]/531.93). Their total SDPI for yards (SDPIY) was -0.86 which ranked 9th in the conference.

To refresh your memory, here are the 2007 Conference USA Standings.
Now here are the 2007 SDPI standings.
It's pretty clear that in 2007, the UAB Blazers were in a class of ineptitude all by themselves. In SDPIP they were more than one standard deviation below the second worst team and in SDPIY they were more than two standard deviations below the second worst team.

Best Offense: UCF 1.54 (SDPIP), Tulsa 1.89 (SDPIY)
Tulsa gained the most yards in Conference USA play, but UCF scored the most points. Why?
1. UCF scored 4 non-offensive touchdowns (punt, kickoff, fumble, or interception returns) in conference play while Tulsa scored but 1.
2. UCF made 12 of 13 field goal attempts in conference play while Tulsa made only 4 of 6 attempts.
3. UCF was +5 in turnover margin in conference play while Tulsa was dead-last at -9.

Worst Offense: UAB -2.08 (SDPIP) and -2.06 (SDPIY)
Neil Callaway's maiden voyage was a long one. The Blazers were a very feast or famine offensive team. In 4 of their conference games, they scored 32 points and gained 1102 yards; never scoring more than 10 points nor gaining more than 338 yards. In their other 4 conference games, they scored 126 points and gained 1614 yards; never scoring fewer than 26 points nor gaining fewer than 343 yards .

Best Defense: Southern Miss 1.54 (SDPIP), UCF 1.62 (SDPIY)
Southern Miss was very close to netting the SDPI clean sweep as the league's best defense last season. They allowed 7 more yards than UCF in Conference USA play.

Worst Defense: UTEP -1.55 (SDPIP) and -1.25 (SDPIY)
When you're worse than Rice and SMU, there are problems.

Hardest Schedule (based on cumulative SDPI of opponents): UAB 5.91 (SDPIP) and 6.14 (SDPIY)
The Blazers played in the tougher division and in their intradivision games they drew the two best teams from the West (Tulsa and Houston).

Easiest Schedule: Houston-4.45 (SDPIP) anf -5.55 (SDPIY)
The Cougars played in the the easier division and avoided the best team from the East (UCF), while drawing the worst (UAB).

Looking ahead to next season, the prohibitive favorite should be...

East
Southern Miss
Things are set up quite nicely for Larry Fedora as he takes over for Jeff Bower. The Golden Eagles were easily the second best team in the East last season despite their 4th place finish. They were markedly better based on yards and points than two of the three teams that finished ahead of them (East Carolina and Memphis). They do lose two signal callers to graduation, Jeremy Young and Stephen Reaves, but their team pass efficiency rating of 114.90 ranked 88th in the nation and ahead of only two teams in Conference USA (Tulane and UAB), so the loss may not be as pronounced. Perhaps more importantly than the quarterbacks they lose, are the offensive skill players that return. Junior running back Damion Fletcher is back after a 1500 yard season. Seven of the top eight pass-catchers are back, including sophomore receiver Torris Magee and senior tight end Shawn Nelson, the top two pass-catchers from 2007. On the other side of the ball, the Golden Eagles have some major personnel losses on the defensive line with Matthew Chatelain and Martavius Prince both departing (combined for 14 sacks). However, linebacker Gerald McRath, arguably the team's best defensive player, returns. The real reason to like the Golden Eagles in the East though, is the attrition at the other alleged contenders. Defending champ UCF loses their stud running back Kevin Smith and his 2500+ yards as well as starting quarterback Kyle Israel and two stars off the offensive line, center Kyle Smith and tackle Josh Sitton. Their best defensive linemen, Leger Douzable and his 7.5 sacks, is also gone so the Golden Knights should fall back to the pack. Memphis loses its starting quarterback and running back as well. Plus the Tigers were actually outscored in conference play despite their 6-2 record. 5 of those 6 wins were by 3 points or fewer. It's likely the Tigers will not be as fortunate in close games in 2008. As for East Carolina? We'll get to them in a moment.

West
Houston
Like Larry Fedora at Southern Miss, another rookie coach, Kevin Sumlin, likely has the best team in the division. According to the yardage version SDPI, Houston was actually the best team in Conference USA. They were also very balanced. They gained the second most yards and allowed the third fewest points. Unfortunately, most of those points allowed came in a 56-7 beatdown at the hands of Tulsa. Things will probably be a little different in that game this season primarily because it is in Houston instead of Tulsa, but also because while Tulsa loses their senior quarterback Paul Smith, the Cougars bring back two competent quarterbacks in sophomore Case Keenum and junior Blake Joseph. Do-everything running back Anthony Aldridge is gone as are the top two receivers, but the defense should remain stout while the offense works out its kinks. 8 starters return to one of the best units in Conference USA, led by defensive end Phillip Hunt and his 10.5 sacks. Look for the 2008 Conference USA Championsip Game to be a rematch of the 2006 contest won by Houston over Southern Miss 34-20.

The team(s) that will improve are...

Marshall, SMU, and Tulane
It's been rough going in Huntington since Bob Pruett called it quits after the 2004 season. His replacement, Mark Snyder, has won only 12 games in 3 seasons, bottoming out at 3-9 in 2007. However, 2008 should be the year the proverbial worm turns. The Thundering Herd were a middle of the pack Conference USA team according to both measures of SDPI, yet finished only 3-5 thanks to a poor record in one-score games (1-3). The Thundering Herd were also -4 in turnover margin in conference games (8th in the league) so a market correction in both categories could help the Herd climb in the standings. The Herd do lose their starting quarterback, Bernard Morris, but the top 3 running backs and 4 of the top 5 pass-catchers return to aid the development of the new quarterback. Defense though is where the Herd should be much improved. Defensive end Albert McClellan had 11.5 sacks as a sophomore in 2006. He went down in the 2007 preseason with a knee injury and missed the entire season. In his absence, the Herd sacked opposing quarterbacks only 13 times in 2007 (tied for 111th in the nation). That lack of pressure also resulted in a lack of forced turnovers. The Herd forced only 7 turnovers in 2007 (dead last in the nation). The return of McClellan, even if he is not 100%, will mean more sacks, more turnovers, and a better team. The non-conference slate is tough, with Illinois State standing as the only likely win, but the Herd are a good darkhorse pick to win the East. For SMU, there is nowhere to go, but up. June Jones is now the coach, so what can we expect from the Pony Express in his first year in Dallas? Here's what he did at Hawaii in 1999.As you can see, the Hawaii offense improved substantially under Jones. However, it should be noted that any offense that averages only 12 and half points per game should improve the following season. So Jones gets a lot of the credit, but the simple nature of the world--piss-poor and great performances are extremely hard to sustain over the course of time, also gets a little. Unlike Hawaii in 1998, SMU's problems in 2007 were not the fault of the offense. The Mustangs were 7th in yards gained in Conference USA play (just a little below average). In fact, the team averaged 28.3 points per game over the course of the season. The culprit was the defense, turnover margin, and bad luck in close games. The Mustangs allowed 507 yards per game in Conference USA play. Amazingly, that number was better than 3 other teams (Rice, UAB, and UTEP), but it was clearly the team's weakest link. In conference play, the Mustangs turnover margin was -6, which ranked 10th in the conference. But the main reason the Mustangs failed to win a single conference game was their bad luck. 5 of their 8 conference losses were by 7 points or fewer, and the combined margin in those 5 games was only 20 points. With any kind of luck, the Mustangs could have eked out 2 or 3 conference wins. While that is certainly not an impressive number, it serves notice that SMU was not nearly as bad as their winless conference record would seem to indicate. Quarterback Justin Willis returns in 2008, looking to improve upon his somewhat disappointing sophomore season. With Jones' tutelage, there's a good chance he will. The situation in Dallas is not nearly as dire as it was nearly a decade ago in Honolulu, but a bowl game may be asking a bit too much. For starters, Texas Tech, TCU, and Navy dot the non-conference slate, so coupled with the likely win over Texas State, the best the ponies can hope for is a 2-2 record outside the league. In the West, it's possible SMU could line up third behind the big dogs Houston and Tulsa, but an improving Tulane team, and dangerous squads at Rice and UTEP could curb that line of thinking. 5-7 seems about right for SMU, primarily because Jones' biggest strength--offense, just needs some fine-tuning, not a complete makeover. The play of the defense will determine how high SMU climbs in the conference standings. And speaking of that improving Tulane team, they may be the third team from the West that goes bowling. How can Tulane improve when they lose 2000 yard rusher Matt Forte? The answer, everybody else on offense is back (almost). Quarterback Anthony Scelfo cut his teeth as a sophomore and looks to improve as his top four receivers and his entire offensive line are back. While the running attack may dip without Forte, the passing game should improve and the running game should be steady. In fact, despite Forte's excellence last season, the Green Wave only finished 8th in yards gained in conference play. The strength of the team last season, besides the rushing attack, was the defense. That unit does lose a pair of stout linemen, Antonio Harris and Avery Williams, but 7 other starters return. Plus Bob Toledo, while he may not be Bear Bryant, is a fine coach, and with a little luck the Green Wave will be bowling in 2008.

The team(s) that will decline are...

East Carolina and Memphis
What's not to love about East Carolina? They pulled off one of the biggest shockers of the bowl season, stunning heavily favored Boise State on a last-second field goal. That win marked the second straight bowl appearance for the Pirates under Skip Holtz. The offseason coaching carousel also avoided Greenville and Skip Holtz is around for at least one more season. So why won't we see the Pirates in the Conference USA Championship Game? For starters, despite their 6-2 record in Conference USA, the Pirates were actually outgained by league foes. The offense that gained 476 yards in the Hawaii Bowl against a pretty good Boise State defense was actually well below average. The Pirates gained more yards in conference play than UAB. That's it. Every other squad in the conference gained more yards. And the hero of that Hawaii Bowl win, running back Chris Johnson? He's gone. So one of the league's worst offenses loses its best player. That's not the recipe for a conference or division title. Plus, East Carolina was on the right side of a lot of good fortune in 2007. They were 4-1 in one-score games (2-1 in conference play) and their turnover margin of +11 in conference play led the league. Both of those facets of the team are likely to regress in 2008. And finally, the schedule, particularly the non-conference slate, is very tough. An ACC trio of Virginia Tech, NC State, and Virginia dot the schedule as well as a game against West Virginia. Best case, the Pirates wind up 1-3 against those teams. That would require at least 5 conference wins to even get to bowl eligibility. Don't be surprised if the Pirates wind up at home for the holidays. The Memphis Tigers could also find themselves at home come late December. Despite being outscored by 23 points over the course of the conference season, the Tigers finished 6-2 in league play. That record was greatly aided by a 5-2 record in one-score games (5-0 in conference play) and a turnover margin of +6 in conference games (2nd in Conference USA). Quarterback Martin Hankins and running back Joseph Doss are gone so the offense will need some serious retooling if it is to be anywhere near as productive as it was last season. With the offensive personnel losses, in order for the Tigers to avoid a losing season, the defense will have to take a major step forward. The defense allowed 442 yards per game last season, which ranked 100th in the nation. 8 starters are back on that unit, but while limited improvement would be expected, a miracle is unlikely. The Tigers will likely not be as fortunate in pulling out close games or consistently winning the turnover margin, so expect a 2nd bowl-less season in 3 years for Memphis.

What's up with UTEP's late-season swoon?
Since coming to the west Texas town of El Paso in 2004, Mike Price has tried to do his best Don Haskins impression and turn around the downtrodden fortunes of the UTEP football program. He's done a fine job in 4 seasons of work, compiling a 25-23 record with two bowl appearances, at a school where the previous coaching regime won only 14 games (and lost 34) in 4 seasons. But the Miners also seem to come apart at the end of the season. In 2004, in their final season in the WAC, UTEP rebounded from a 1-2 start with the defeats coming by a combined 48 points, to win 7 straight games. They stumbled and lost their final two games, the first to a bad Tulsa squad (4-8) in the regular season finale, and the last to a decent Colorado team (8-5) in the Houston Bowl. In 2005, the Miners looked to improve upon their surprise 2004 campaign. After 9 games, they stood 8-1 (5-1 in their first season in Conference USA), and were staking out a place in the innagural Conference USA Championship Game. Alas, they lost their final two regular season games to two teams that both finished 5-6 (UAB and SMU). Tulsa took their spot in the title game, and the Miners headed to Mobile and the GMAC Bowl where they were throttled by Toledo 45-13. The Miners regrouped and began 2006 with a 4-2 record, with their only losses coming to Texas Tech (in a 38-35 squeaker) and New Mexico. They proceeded to lose 5 of their last 6 games and finished 5-7 to miss out on the postseason for the first time under Mike Price. 2007 was eerily similar. The Miners began the year 4-2, with their only losses coming to Texas Tech and New Mexico State. But the Miners one-upped themselves by losing their last 6 games to finish 4-8 and ahead of only winless SMU in the Conference USA's West Division. Words can give us a good idea of how things happened, but as they say a picture is worth a thousand words, and I am no Shakespeare. Below is a summary of UTEP's record under Mike Price split by months.
As you can see, the Miners tend to fall off substantially in their last few regular season games and on into the postseason. Why is this? Are the Miners poorly conditioned? Does the rugged El Paso climate wear out even the heartiest of men? Let's delve a little deeper.
Close games here are defined as one-score games (those decided by 8 points or fewer). This split helps explain some of the difference in UTEP's record. If we normalize the close game record to .500, for both sets of games the Aug-Oct and Nov-Jan differences are not as pronounced. UTEP's Aug-Oct 'luck' adjusted record drops to 17-14 and the Nov-Jan 'luck' adjusted record rises just a shade to 6-11. That's still a pretty significant difference, but not nearly as pronounced as the actual record. Before we assume it's the conditioning accounting for this split, let's look at one more thing.
Yardage differential can give us a good idea about how well UTEP played in each time period. The amazing thing is, in 2004 and 2005 when UTEP closed with a 4-5 record in November and January, they were playing nearly just as well (2004) and better (2005) than they were when they were a combined 12-3. The difference is even more pronounced in 2006. While UTEP stood 4-4 after October, the Miner's were actually being outgained by 42 yards per game. Over the final 4 games, during which time they went 1-3, the Miners outgained their opponents by 12 yards per game. While 2007 doesn't seem to fit in with the other 3 years, it does to some extent. The Miners were truly much worse in Nov-Jan as their entire team went to hell and they were outgained by 131 yards per game. However, the Miners were busy being outgained by 58 yards per game from Aug-Oct and yet still managed a 4-4 record, which is quite a feat for a team being outplayed so consistently. A glance at the totals can give us an idea at what is actually happening. The Miners from Nov-Jan are not substantially worse than the Miners from Aug-Oct. The offense is a bit better and the defense is a bit worse, but the net result is the same--the Miners have been outgained by about 5 yards per game over the course of 31 games from Aug-Oct in the Mike Price era and by about 5 yards per game over the course of 17 games from Nov-Jan. Thanks to random chance, the Miners have played over their heads from Aug-Oct and a bit under from Nov-Jan. We would expect a team that is outgained on average by about 5 yards per game to finish with a winning percentage around .500, maybe a little above and maybe a little below. The Miners' winning percentage under Mike Price? .521.