It’s Bill Connelly season. #BillCzn should probably be a thing, I think.
Connelly, most notably known for his S&P+ rankings as seen on Football Outsiders and SB Nation, finally got his preview series started Monday with the first team preview published (hello, Bowling Green!) and the projected college football rankings for the 2019 season.
By the rankings, Western Michigan is the top-projected MAC team at 75th nationally (-0.5 S&P+ proj.) with Northern Illinois right on its heels at 76, then Toledo and Ohio come in next at 78 and 82, respectively — the four teams you probably would’ve expected to see here.
Here’s what Connelly’s math shows, being tunnel-visioned with the MAC (side note: I’ll link each of Connelly’s team preview in this table as they come out):
|MAC Rk||Team||Proj. S&P+ Rk||Proj. Off. S&P+ Rk||Proj. Def. S&P+ Rk||Recruiting impact||Returning production||Weighted 5-yr|
|1||WMU||75 (-0.5)||39 (33.2)||100 (33.7)||84||78||66|
|2||NIU||76 (-0.6)||120 (18.9)||19 (18.9)||101||73||81|
|3||Toledo||78 (-1.3)||33 (33.8)||107 (35)||71||88||45|
|4||Ohio||82 (-1.9)||34 (33.7)||109 (35.6)||99||82||82|
|5||Miami||93 (-4.9)||95 (25.1)||82 (30)||96||91||99|
|6||EMU||96 (-6.4)||106 (23.2)||79 (29.7)||125||95||112|
|7||Buffalo||97 (-7)||88 (26.7)||99 (33.7)||112||97||109|
|8||Ball State||110 (-12.9)||105 (23.5)||111 (36.4)||116||108||115|
|9||Kent State||111 (-13.7)||99 (24.6)||115 (38.3)||105||110||124|
|10||CMU||122 (18.5)||128 (13.7)||93 (32.2)||108||122||93|
|11||BGSU||123 (-19.4)||85 (27.5)||130 (46.9)||113||124||90|
|12||Akron||124 (-19.9)||127 (14.4)||101 (34.3)||122||125||103|
Considering that there are 130 FBS teams and the MAC’s best team comes in at 75 on the list, this is not what the MAC wants to see — even if they are just preseason rankings. Putting WMU at 75 means that they’re 10 positions lower than the midway point of 65, and is the 16th-highest rated Group of 5 team on the list. The 15 teams ranked higher than WMU: 24. Boise State, 26. Memphis, 27. UCF, 31. Appalachian State, 42. Utah State, 44. Cincinnati, 50. BYU, 51. Fresno State, 54. San Diego State, 66. Temple, 69. Troy, 70. Arkansas State, 71. USF, 73. Houston and 74. Southern Mississippi.
As Connelly noted in his piece, in bold font: “The ranking… is not where they rank in returning production but where they would rank after the projected changes are applied to last year’s S&P+ averages.”
Another note: Connelly did change up his math on these projections a little bit and went back to streamline his new S&P+ algorithm to reflect all of the other years of the statistic, 2005 through current day. Whatever the teams’ rankings were before might be a little bit different, but not too much.
After the ratings are determined, I project previous games based on those ratings, and I track each conference’s average performance versus projection. For the top conference, I found that by the end of the season it was aiming low by two or three points per game per team. For the bottom conference, it was the reverse.
Shifting each team’s rating based on this conference average, and increasing the weight of said adjustment as the season progresses, basically improves against-the-spread performance by about 1 percentage point per season and cuts the average absolute error by somewhere between 0.2 and 0.3 points per game. (SB Nation)
For a given season, shifting each conference team in this manner can mean that the best conference in FBS ends up with quite a few teams near the top. For the 2008 season (as presented in the example above), that means a heavy Big 12 presence. And for 2018, it means that the SEC dominated the ratings as much as any conference ever has — as much as SEC fans like to think their league dominates every year. (SB Nation)
In his post that explains the edited algorithm, Connelly points out that Fresno State finished ninth in its year-end S&P+ rating last year, but the new math says it’s 16th instead. UCF dropped from eight to 18. Appalachian State was 11th, now 29th. But the funnest of them all is way up top. Even after Clemson beat the pants off of Alabama in the national title, the Tigers are third in the year-end rankings below Bama and, naturally, Georgia.
To take that a step further in the MAC’s direction, here’s another table for you. I’ve taken Connelly’s 2018 projected rankings (doesn’t include the new math), the 2018 year-end rankings before the new math, the 2018 year-end rankings after the new math and the new projections:
|Team||2018 proj.||2018 finish (before)||2018 finish (after), change||2019 proj.|
|Ball State||117||117||119 (-2)||110|
|Kent State||127||120||124 (-4)||111|
- The projected conference averages has the MAC ninth place. From top to bottom: 1. SEC, 2. Big Ten, 3. Big 12, 4. Pac-12, 5. ACC, 6. AAC, 7. Mountain West, 8. Sun Belt, 9. MAC, 10. C-USA.
- Connelly freaking nailed it with Ball State and Akron, didn’t he? Respectively, the teams were projected to finish 117 and 119 before the season, and they both did! Until they didn’t. But even then, they both barely slid down to 119 and 120.
- EMU’s projected finishes before both 2018 and 2019 seasons (with two sorta-different algorithms) are both 96. EMU’s year-end finish for last year dropped from 62 to 86.
- The only team to have its year-end S&P+ ranking to have a move in the positive direction was Bowling Green, from 127 to 126. Not much to brag about there.
- Buffalo had a 10-win season which was fun and impressive and whatnot, but S&P+ must’ve taken into account that Buffalo’s 10 wins came over teams like Delaware State, Rutgers and the four lowest-ranked MAC teams — CMU, Akron, Kent State and BG, which is how the Bulls finished at 82 on the new S&P+ math last year. The 24-point swing is the widest of any team in the MAC. The Bulls have the lowest returning production percentage in the MAC (barely not the worst team nationally), which is how the team’s projected to finish at No. 97 this year.
- The three highest-rated projected MAC offenses by the metric: 33. Toledo, 34. Ohio and 39. WMU. All three teams return their starting QBs, but they’re all a little bit different after that. Toledo’s got to replace its three awesome wideouts, WMU’s got an identity crisis (the team’s better off going more up-tempo on offense, but tends to not do that), and Ohio has Nathan Rourke but A.J. Ouellette and Papi White are both graduated.
- Even without two-time All-American Sutton Smith and Josh Corcoran as NIU’s defensive ends again, the Huskies have the MAC’s best-projected defense (no surprise there) at No. 19 (that’s higher than I would’ve thought). Just goes to show how established NIU is in terms of grooming its defenses & recruiting well enough to fill most voids (see: Antonio Jones-Davis filling in for Juwaun Johnson after he transferred to TCU last year). EMU’s the second-best projected defense in the league at 79, facing similar problems with departing players (Maxx Crosby, Jeremiah Davis, Kyle Rachwal, Justin Moody, etc.) but doesn’t have the establishment that NIU has to influence the rankings otherwise. The next best MAC defenses: No. 82 Miami, No. 93 CMU, No. 99 Buffalo and No. 100 WMU.