Hello and welcome back. The technical glitch was resolved (or at least enough for today) so it’s time to bring in this week’s post. For those who haven’t figured this out, I’m not known for taking it easy. I never have been one to sit still well. A couple of years ago, I went to a college homecoming (Go Blue Hose #246) and recognized someone I hadn’t seen in over 30 years (from his Facebook page). It was obvious from the look on his face that he was trying to place me. Finally, he said, “oh…you’re the guy with all the energy.” Although I’m getting older (that’s a good thing), I find that while the energy level is still pretty high, I do crash during stressful times a bit quicker. I started a new job last month – it’s exciting and I really like it – it does take extra energy and a little bit of starting new stress. I’m a Master Black Belt and a Lean Sensei in Lean Six Sigma – so I’m all about doing it right the first time – just can’t stand it when things don’t work as planned or designed. I’m in a band, Project Detour (Project Detour Band on Facebook, Twitter, and Instagram) and, shameless plug for those in the Atlanta and nearby environs, make your reservations now at Houck’s to see us November 9. If you go, I’ll be the guy jumping and dancing around and generally acting like an idiot – it’s a lot of fun! In the band, I’m known as G. Just G. Anyway, all of this to say I apologize for taking the easy way out last night. By the time I found out my hard drive was toast and got the confirmation from a tech guy, my energy stores were sapped. Helping my daughter with some homework (ridiculous amounts given by her teachers – especially one of them who clearly doesn’t actually teach anything but rather assigns the kids to figure it out for themselves – is a sign of the opposite of energy – not a fan of lazy people – sorry for this side rant) was about all I had the oomph to do.
That said, a friend was talking to me about naming the various routine posts. I have one (normally) on Tuesday that could be straight opinion or could be light analysis but is designed to be more fun. This is where your questions or suggestions for research/analysis might appear. Then on Thursday, I talk about the most important games for the weekend along with some predictions. Early Sunday afternoon, I post the latest and greatest GCR with minimal analysis. Lastly, Sunday night I post deeper analysis. I’ve come up with some potential names of these posts – let me know what you think.
| Tuesday night/Midweek | G’s Exploration |
| Thursday night | G’s Expectation |
| GCR Ratings Post | G’s Evaluation |
| GCR Analysis | G’s Explanation |
I’m open to any changes you may have. Speaking of that and shameless plugs, keep sending in questions/comments/requests/complaints, etc., so that I can use those thoughts to help make these blogs more meaningful.
Now to this week’s Exploration: I mentioned in a previous blog that I rarely talk about the AP or any other poll. Today is an exception. This is a new analysis for the GCR, but as we approach the midway point of the season (some teams have played 6, most 5), I thought it might be a good test. I took this week’s AP poll (I looked at the USA Today poll as well, but did not include it for simplicity purposes), and compared that ranking to the GCR. The first thing I noticed is that, this week, there were no matches, but there were 7 that were within 2. Whew! There also were 10 that were more than 10 spots apart (note: this includes Texas Tech that the GCR has #25 but received no votes in either the AP or the USA Today polls). I’ve listed those teams in ALL CAPS. How could this be? First look at the table:
The first thing that sticks out is the highly disparate SOS. That accounts for some of the differences. There are 2 top 10 SOS on the list: one of them is Auburn which has pundits talking about how they’ve played the toughest so far and all that. The GCR has them 10th – see below for the top 10 so that you can see why. The other is Texas Tech which is getting no love whatsoever from the pollsters but have ridden that 8th place SOS to a top 25 slot. The other columns are what I’d like to talk about. First the Offensive Efficiency Rank: The OER is a measure of how good a team’s offense is. Now, in the GCR, I don’t track 3rd down conversions, turnovers or yardage gained (these and others are really good stats to know and could be indicative), because the team that gets 300 yards first does not win the game. All of the myriad stats out there are really cool, especially for Lean Six Sigma number geeks, but at the end of the game, only one statistic matters – points. The entire engine of the GCR is based on points. So, if a team scores 35 points in a game is that good or bad? Of course, the answer is it depends. For the GCR rankings themselves, the next question is “did they win or lose and by how much?” (there are other questions too but that’s for another blog). For OER, the question is “how many points does that opponent normally give up?” So, if they team that your team scored 35 on normally gives up 17 per game (including the game against you), that’s really good, but if they normally give up 50, your offense is not quite up to snuff. Now each opponent has different points allowed, so we have to convert all of those games to a single average: the OER. Also, as a reminder, the GCR assumes each game played happened at the same time, so these numbers are always changing. For example, when you played team x and scored 35, their average opponent scored 17 against them. But 3 games later that were blowouts and their average moved to 36 per game and you were just average. So sorry. This is the GCR and that means I have a ranking for that measure – of course you do, G, of course, you do. The last column is essentially the same thing but on defense (DER) – and, again, with a ranking. 1 is the best offense (or defense), 256 is the worst. These columns give an idea whether a team is offensively focused or would rather fight it out in a low-scoring game. Look at the top 18 teams – all of them have a top 25 offense, defense, or both. After that, it’s hit or miss. For the offensive side, 3 teams stick out as being close to mediocre: Iowa, Arizona State, and Cincinnati. One of these games, look for them just to get outscored by a “lesser” opponent. An example is Iowa/Michigan last week – the Hawkeyes gave up 10 and lost. On the flip side, SMU, Minnesota and App State are just waiting to lose a shootout game. All are undefeated now, but SMU had to come back down 21 in the 4th to win last week against Tulsa.
This did make me wonder who the various best teams were in the three categories; SOS, OER, and DER. That led to a table to show this and it’s really interesting. Auburn fans (and ESPN, et al) are really pushing their SOS this year. I have them 10th because they have played one OF the most difficult. But, let’s take emotion out of this and see what it says:
Kent State with 3 top 25 away games has the lead. Interestingly the top 3 are 0-8 against top 25 teams and 7-0 against everybody else. Also interesting is that there are only 2 home games in the 19 top 25 games for this crew. One new stat is the last column – the Opp IND Rec is the wins and losses of a team’s opponents if that team’s games are removed. To continue picking on Auburn (not really because their schedule is still the 10th most difficult – compare this to most of the top 25 in the first table), the GCR has them as 5th in the Power 5 behind Tennessee, Northwestern, Texas Tech, and Purdue (of note, those teams are a combined 6-14 compared to Auburn’s 5-1). This will change as the season continues as the Power 5 schools have a much higher probability of increasing their SOS as conferences games are played.
I also looked at OER – who has the best offense in the land. Here’s what I discovered (for fun I put in the 3 most likely NOT to be offensive schools too):
Iowa State and Auburn’s SOS are showing here: they are the only top 10 OER team’s scoring fewer than 40 points per game. The actual OER can be read as saying a team scored x% of their average opponent’s average opponent. That sounds weird, I know, so Ohio State scores 187% of it’s opponents’ average opponent’s defensive points per game. So, if they played a school that gave up 20 per game, the Buckeyes may score 37 or so (there is another number that has to go into that so it’s not that simple – another topic for another blog). But it does show scale. With the exception of Iowa State, all of the top 10 are ranked in the top 25.
Likewise, there is a DER ranking and here is the top 10 (plus 3 of the most porous schools):
For the DER, higher is better as well, so Wisconsin’s opponent’s score 4 times as many points against others as they do against Wisconsin. A couple of interesting teams on this. First San Diego St which is in the, somewhat, high-scoring Mountain West division, are the opposite. Their OER Rank is 223. They’ve done well because they have a stingy defense. Hapless Massachusetts, a Group of 5 Independent, is really not solid on defense – worst than ANY Division 1 team. Their opponent’s score just over half as many points against others as they do against the Minutemen.
In a future blog, I’ll talk about how to use the OER and DER to predict the scores of games. It’s clearly not an exact science, but it’s fun. I’m over 2300 words in this one already, so send comments/questions/queries. Please subscribe, if you haven’t, and share with others who may be interested,
Thanks, G or r or whatever I go by today.
Love G’s Es! That was a cool idea!
Thank you very much!!!!!
Thinking back to this post… Many of your followers will compare your rankings to the other main popular rankings. Following the rankings as the season progresses is very interesting, but as we know, more data makes good analysis easier and better. Therefore, the best rankings would be those at the end of the season after all the games are played, including the playoffs and bowls. Your rankings will become a much more accurate predictor of who would beat who if everyone played each other on the day after the college football season (including bowls and playoffs) ended. That’s the point of any rankings system I think. Of course, the AP and the coaches’ poll would also say their rankings become more accurate, too, although they are based on different somewhat more subjective (some would say more detailed analysis, such as position by position) data.
There is also the question of recency. Some teams improve more than others over the course of a season. I know you intentionally don’t take that into account. For example, does the AP discount early losses more than late losses, because that might be a better predictor of who could beat who at the end of the season?
This is all to say, I wonder how the GCR compares to the AP for the last couple seasons for the end of season rankings. I’m sure you’ve looked at this, and say to yourself “That team is way over-rated or under-rated (based on my unbiased computer).”
Once again, Carl, you make me think. I think any subjective poll is suspect to bias. I tried hard to remove any bias from the GCR. But sometimes, the pure unemotional calculations show what “feels” wrong. Not sure what the right answer is. Each year, I add something new based on feedback or nagging questions. I do believe recency is a real component, but haven’t figured out how to calc it fairly yet. It diminishes the early season games against lower level teams and maximizes the week-before-rivalry-week ones. Part of me wants to say x team played an FCS team and that’s it, but part of me wants to take “momentum”, whatever that means, into account. On the opposite end of the spectrum, an early game between future top 10 teams (regardless of their rank at the time) could have a significant impact on the rankings and would be deemed less important by the recency metric. A measure for momentum or recency would somewhat account for personnel change that is inevitable during the season. I’ll continue to work on it and am always open to suggestions.