Jonathan Sadow
1st String
Posts: 1,104
Joined: Jan 2006
Reputation: 27
I Root For: Strigids
Location:
|
RE: 2015 Football Performance Ratings
(09-16-2015 01:28 PM)mrbig Wrote: Jonathan, thanks again for doing these! I really enjoy seeing how these develop over the course of the season.
One question for you. Rice's Performance Rating against Wagner was initially 60.13, which increased to 62.39 after the loss to Texas. I know it is not unusual for Performance Ratings in individual games to shift as more data goes into the system. But with Wagner off last weekend and Rice losing badly to a UT team with a bad week 1 performance, why did Rice's Performance Rating for the Wagner game increase? Obviously, the margine-of-victory did not change, so that mens that either Wagner's power rating decreased (but why, since they did not play and have no other games played this season) or the value of Rice's home-field advantage changed (which seems odd, especially a 2+ point swing after one week).
Just curious.
You'd really have to ask Prof. Massey about that. From what I understand about how his system works, there are a couple of things that could be influencing these changes. The ratings at the beginning of the season are substantially influenced by the previous season's results, and as the present season is played out, that influence is diminished and finally disappears. It's possible that lowering the weighting from the previous season caused Wagner's rating to go up. Also note that the system is Bayesian, which means in essence that the ordering of results over time does have an influence on the value of the rating (unlike "classical" statistics, in which the order of the data has no effect on the statistical quantities derived from them).
Speaking of data, let's look at the data from Rice's most recent game. The 38-24 win over North Texas produced a performance rating of 54.95, which turns out to be the median result so far this season:
Wagner: 58.38
@Texas: 51.92
@North Texas: 54.95
That's actually the highest median value for the first three games of the season in the David Bailiff era, so at least he's got the guys playing consistently like an average Division I FBS team.
Here are the performance ratings for Conference USA teams this season:
School, Median, Mean, Standard Deviation
Louisiana Tech 61.86 61.74 7.24
Middle Tennessee 58.04 67.07 16.26
Western Kentucky 56.04 55.66 2.86
Rice 54.95 55.09 3.23
Southern Mississippi 52.20 51.31 1.66
Marshall 50.92 51.77 7.67
Florida International 44.71 47.48 7.63
Texas-San Antonio 43.18 38.55 23.07
Florida Atlantic 38.77 37.28 7.22
Charlotte 37.38 24.44 29.38
Texas-El Paso 36.82 30.44 11.70
Old Dominion 36.29 35.01 7.02
North Texas 32.62 32.62 6.12
There are lots of interesting tidbits to be gleaned from these data:
- In the Eastern Division, while Marshall is still a good team, it's not dominant like it was last season. Both Middle Tennessee and Western Kentucky look like they'll have a better claim to the divisional crown over the course of the season.
- For a team playing its first Division I FBS season, Charlotte performed creditably well in its first two games, victories over FBS bottom-feeder Georgia State and FCS member Presbyterian. However, the 49ers were totally destroyed by Middle Tennessee 73-14 in their C-USA opener, resulting in the lowest performance rating by a C-USA team (-9.13) since Alabama-Birmingham's rollover to Southern Miss in the 2013 finale (-19.50). Conversely, the Blue Raiders' win was the highest-performing game of the season so far for a C-USA team (85.84).
- Speaking of Southern Miss, the Golden Eagles appear to be recovering nicely from their three-year slump in which they won only four games (three of which were last season). USM's 56-50 win at Texas State, while hardly striking fear into the likes of Ohio State, was the type of game the Golden Eagles reliably lost in the previous three seasons. The win puts them at 2-1 on the season, with the other results being a creditable loss hosting Mississippi State and a big win against FCS bottom-feeder Austin Peay. Todd Monken's rebuilding plan appears to be working, and right now I'd say Southern Miss is the most improved (and perhaps underrated) team in C-USA.
- On the flip side, UTEP has to be the most disappointing. Sean Kugler got his Miner team to a bowl game last year, and even better things were expected this season. Instead, UTEP was not competitive against Arkansas and Texas Tech, at best mid-level Power 5 conference teams, then had to go into overtime to beat New Mexico State, perhaps the worst team in Division I FBS. Looking at the rest of the schedule, I don't see how the Miners are going to be bowl-eligible this season. They should beat Incarnate Word in their next game, but the only other possibilities for wins now look like Old Dominion, North Texas, and...
- ... UTSA, which is doing the same thing it did last season: have a competitive loss to a good Arizona team, then trend downward as the season progresses. Last week, it was a humiliating 69-14 loss to Oklahoma State, a team about as good as Arizona. It was never going to be easy for the Roadrunners, who have the toughest schedule by rating in C-USA this season, but given the trend over the last two seasons, you have to wonder what Larry Coker is doing over there.
Rice's opponent this week is Baylor, a consensus top 5 team. The Bears, obviously worried about the Owls, scheduled an off-week before the game and thus only have played two games this season, but both were five-touchdown victories over outmanned Southern Methodist and Lamar, respectively. In neither game, however, were the Bears truly dominant, giving up lots of points and winning only by massively outscoring the opposition. That'll likely happen against Rice as well, but maybe not to the degree it did in Baylor's first two games. The Owls are a better team than SMU and Lamar, and I think they'll keep it closer than those two did against the Bears. Given the wide disparity in the two teams ratings (Baylor's lesser performance of this season is about seven points better than Rice's best performance of the season), I see little chance of an Owl victory. However, for those who like to bet (legally, of course) on these things, given the fact that the spread is, like the margin in Baylor's first two games, about five touchdowns, putting some money down on the Owls might be profitable (keeping in mind that past performance is not entirely predictive of future results).
|
|