Monday, September 08, 2008

Thoughts on Rankings, and My Poll

I really hate ranking teams this time of year.

We simply don’t know enough about any teams at this point to make credible rankings. The preseason BlogPoll contained such turkeys at Clemson at 8th, West Virginia at 9th, and Virginia Tech at 14th, all which look ridiculous now.

I still have yet to have anyone articulate a coherent and cogent philosophy of what exactly ranking should mean – a “unifying theory” if you will. Some seem to believe it is about an ordering of the “best” teams, some that it is a semi-predictive process, and others still that it is a rankings of merit. Many seem to treat it is an undefined and amorphous combination of all three.

Obviously all current systems have their shortcomings.

The most irritating use of polling to me is that of the predictive, which seems inevitably to lead to the “So-and-so was ranked wrong” discussions. In my view rankings shouldn’t be some quasi-odds/lines maker process, and have no relevance as to which team might beat another in the future. According to my line of thought, not only were the rankings of Clemson, West Virginia and Virginia Tech in the original BlogPoll “wrong”, but in fact all the preseason rankings are wrong, and inherently so, because they are based on information that is – by and large – old and/or speculative.

Personally I prefer “resume” ranking, while likewise acknowledging its inherent limitations. Brian of Mgoblog has pointed out (somewhere) that ranking teams according to the “quality” of their wins is as inconsistent as ranking based on preseason speculation of strength where you are, in fact, speculating on opponent’s quality, a quality that is equally unknown. This is certainly true early in a season.

However as we have seen, later in a season the unknowns become known and traditional and resume ranking systems seem to converge, as fewer teams of obvious quality emerge. And, eventually

No the differences between methods is most obvious (and perhaps least useful) now, early in the season, where a resume ranker would have ECU number one, while a traditionalist would probably go USC.

After an effort at some hybrid ranking based on my preseason poll, I’m going to return to last year’s system of resume ranking. I am aware this is going to cause crazy “deltas” in my poll, but nothing else to me feels genuine. My philosophy will rank teams according to the following –

Top Tier Teams – the high quality and elite teams in BCS conferences. Examples include USC, Oklahoma, Ohio State and Georgia.

Mid Tier Teams – Most of the remaining BCS conference teams, and some of the better non-BCS conference teams, like Utah.

Low Tier Teams – The worst of the BCS teams (like Duke), and the rest of the non-BCS teams.

Does this have an element of subjectivity? Hell yes, but so did ranking West Virginia high preseason.

Here is how I will apply the tiering -

Wins over top tier teams will be the highest achievement in my polling When two teams have wins over comparable teams, the strength of the win will be compared. However in all cases a win over a quality team will be superior to –

Wins over mid tier teams – highly ranked teams are expected to beat these teams comfortably. Wins over similar teams will likewise be compared. But in all cases again, these wins are superior to -

Wins over lower tier teams -I fully expect ranked teams to beat lower tier teams without mercy. A lower tier team should never be competitive at any time in a game.

I’m not sure an example of the first has occurred this year, but the winner of the OSU-USC game will qualify.

In the second case, both of ECU’s wins fall firmly here (if not better). Florida’s win over Miami is also in this category.

The third case is most of the other games won by top teams so far. Ohio State failed to provide the quality beating over Ohio expected of this category.

After wins are compared, losses will be considered on a similar basis. A loss to a top tier team will be a “good” loss, while a loss to a lower tier team will virtually eliminate the loser from any serious ranking.

Obviously the status of teams will (and should) change during the season. An immediate example of this is the status of Virginia, who is closer to the third category than second after their pitiful over Richmond this past weekend.

Based on the above, the following is what I see after week 2, with my impressions of the quality of wins posted

1. East Carolina – wins over at least 2 mid-tier (if not top tier) teams puts them here.
2. Florida – 1 mid, 1 low. Miami is better than any other mid-tier win of the teams below here.
3. Alabama – 1 mid, 1 low. How good is Clemson? We don’t know yet, but I think UM is better.
4. UCLA – 1 mid. Win over Fulmer’s crew gets them this notch.
5. Georgia Tech – 1 mid, 1 low. Win over BC reasonably impressive.
6. Fresno State- 1 mid. Impressive win at Rutgers, but how good are the Scarlet Knights?
7. Oklahoma – 1 mid, 1 low. Thrashing of Cincy lands them here.
8. Vanderbilt – 1 mid, 1 low. Win over Spurrier gets them 8th.
9. Penn State – 1 mid, 1 low. Is Oregon State mid? Well, they whipped them.
10. Missouri – 1 mid, 1 low. Gave up too many yards to Illinois, but won.
11.Cal – 1 mid, 1 low. Michigan State mid-tier? Better than Louisville I think.
12. Kentucky – 1 mid, 1 low. Win over Louisville.
13. Arizona State – 1 mid, 1 low. Counting Stanford as mid-tier (barely).
14. Wake Forest – 1 mid, 1 low. Same with Ole Miss, barely mid-tier.
15. USC – 1 low. Virginia simply isn’t very good, but it was on the road.
16. Arizona – 2 low. Proper whipping of two low tiers.
17. Texas – 2 low. Proper beatings.
18. Wisconsin – 2 low. Proper beatings both.
19. Georgia – 2 low. Beat the low-tiers, but has given 39 points?
20. Oregon – 2 low. Beat them well
21. Kansas – 2 low, decent beatdowns.
22. Auburn – 2 low. Where is the offense here?
23. LSU – 1 low. Need more evidence.
24. Texas Tech – 2 low, beatdowns.
25. Ohio State – 2 low (with bad effort against Ohio)

Other 2 win teams didn’t sufficiently impress. Yet.

I’m sure this is going to bring me much grief from the BlogPollers, but people need to recognize THIS IS STILL VERY EARLY. Many teams have played only a single game. This poll will change dramatically until about the 5th week.

And, once again, I am requesting comments from readers. Please know my methodology is not going to change, but I welcome comments on how I ranked the teams subject to my system.

10 comments:

Henry Louis Gomez said...

Well you'll certainly generate a lot of traffic to the blog. Should I turn the comments off?

Anonymous said...

"We simply don’t know enough about any teams at this point to make credible rankings."

Mergz, you had it right to start with.

I'm not starting the poll at my blog until after the games of week six are in the book. There's just no point in trying earlier, IMO.

Mergz said...

Henry - Nah, leave them on. We can scrub any that are too nasty.

Bluto - Were it up to me I wouldn't rank this early either. However the BlogPoll is an unforgiving master, and it requires results. I don't start my own Power Rankings until about week 6th either. It is based on data that, if used now, would result in ridiculous rankings. Kind of like all the other polls.

Unknown said...

Yeah I was going to say... is this purely to gain traffic? Vandy at 8?

Mergz said...

First of all, I don't really care about traffic.

As for Vandy, they beat South Carolina. By my reckoning only the teams above them have defeated better opponents THUS FAR (key).

I don't think Vandy the 8th best team in the nation, but they have the 8th best record right now.

This isn't all that different from what the SMQ (or whatever his moniker is) does. I haven't seen his poll this week yet, but I bet he ranks Vandy too.

Anonymous said...

Great rankings. Clearly stated reasonings, and the guts to make unpopular picks. My only qualm is the "1 mid > 2 low" philosophy. Fresno had a good win...but they've only played once. I'd be heistent to put them higher than a team that showed me two wins over possibly worse competition.

Anonymous said...

Thank you, I'm glad you went back to this system of voting since I believe as we get toward the end of the season this is the only system that should be used. Otherwise what does playing on the field matter, its what we think the teams can do, and that changes very little as games go on.

Jams said...

I like the system and the order of the rankings, but I think it worth noting that you misnamed your number one team. They're East Carolina, not Eastern.

That is all.

Mergz said...

dethwing- Unfortunately at this point several teams have only 1 game (USC, for instance). I'll say it again - I hate ranking at this point, it makes me feel "dirty". But to particpate in the BlogPoll, I have to.

James - Noted and fixed. Thanks.

Anonymous said...

Wow Rutgers layed a huge egg. That really de-values Fresno's win. However, with Wisconsin on tap it may not matter.