If you don't rank by resume, aren't you just pleasuring yourself?
Here's Andy Staples of SI's explanation for not ranking Michigan this week.
I didn't rank Michigan this week because I still don't know how good the Wolverines are. I know quarterback Denard Robinson is one of the best players in the country, but a win against Notre Dame doesn't reveal anything because we don't know if Notre Dame is any good, either.
Uh huh. Well, how do we know if Alabama or Ohio State are any good either? All they did was win their first to games. Is it because Miami and Penn State are ranked? Well, how do we know Miami and Penn State are any good? All they did was win ONE of their first two games.
Now, this is a serious question. If you were a press person with a ranking, and you did not rank on resume, aren't you just stroking your ego? "I know more than reality, and baby, it feels good"?
I hereby declare a new rule. BCS conference teams are worth 3 points. Non AQ FBS teams are worth 2 points. FCS teams are worth 1 point. After points are calculated, ties will be broken based on the following criteria. Every touchdown scored is worth a point. Every touchdown given up loses a point, and every 100 yards gained or given up is worth half a point up or down, respectively.
There's your rankings. Prove to me that's not pretty much the only sane way to rank teams this early in the year. Later in the year, you can mix in Opponent Wins, etcetera.
September 16th, 2010 at 12:53 PM ^
The title of this thread can be taken in many different directions.
September 16th, 2010 at 12:54 PM ^
how it was intended.
September 16th, 2010 at 12:56 PM ^
There is a fundamental difference between Michigan this year and OSU and Alabama this year. Namely, Michigan wasn't very good last season and there are obvious questions about how much Michigan improved this past off season. OSU and Alabama are coming off big seasons, so there's an underlying assumption that they'll be as good as last year until they show they've regressed somehow.
Basically, that's an unrealistic comparison because Michigan has to prove they're good this year (understandably), but OSU and Alabama, so far at least, has to prove that they're not worse than they were last year.
I completely agree that resumes and strength of schedule should be an important factor. But it's not like Notre Dame and UConn are OSU and Alabama. There are obvious questions about how good they are, as well.
September 16th, 2010 at 1:05 PM ^
Noone cares untill bowl week.
September 16th, 2010 at 6:22 PM ^
Shut up already, nobody wants to read your blathering b.s.
September 16th, 2010 at 6:23 PM ^
Your point regarding Bama and OSU is taken, what about the other 23 teams?
September 16th, 2010 at 12:57 PM ^
I'm realize I'm being a hypocrite by saying this, but rankings at this point shouldn't be discussed or debated much. As BHGP and others have argued, the concept of resume is silly this early in the season, because you're just bumping your arbitrary assumptions of team quality one degree of seperation down.
Polls are "pleasuring yourself" either way. Resume ranking is using a glove to disguise whats going on...but just barely.
September 16th, 2010 at 1:05 PM ^
The only assumptions are that:
A: BCS teams are better than non BCS
B: FBS teams are better than FCS.
Probably not true, but at least not a totally baseless, "Well, this is what I think because I'm awesome."
September 16th, 2010 at 1:23 PM ^
is flawed. Washington St <Alabama. Your formula would disagree.
September 16th, 2010 at 1:23 PM ^
It doesn't matter what system you use to organize your lack of valid performance information on the present teams, your "guess/vote" is still a guess.
So rankings are always just a tool to stimulate arguing and bragging, which is a great way to gether eyeballs to your website (or in olden times to sell papers)
At the end of the day all you have is a top ten list.
Wins and Losses are the only true measurable results,
and next year when we have a championship game there will be a clear and singular Big Ten Champion.
September 16th, 2010 at 1:34 PM ^
There are other reasonable assumptions that can be made regarding team quality. Its not baseless to go on talent level, incoming recruits, returning players, and history either. These things correlate pretty strongly. You assume BCS>not and someone else assumes Ohio State>Indiana. I'm not sure which is more correct.
September 16th, 2010 at 12:58 PM ^
Whether they rank us or not, I personally think most of the MSM can go pleasure themselves.
September 16th, 2010 at 1:30 PM ^
Most of them don't have any alternative
September 16th, 2010 at 1:15 PM ^
I'd care more if this wasn't week 2 of the season. I don't blame anyone for being a bit skeptical about UM after the past few years, and it is UM's job to win them back over.
September 16th, 2010 at 1:18 PM ^
if you don't have any priors.
September 16th, 2010 at 1:24 PM ^
I don't really have a problem with this. If we keep winning, the rankings will take care of themselves. Maybe I'm being overly generous because I generally think Staples is one of the more reasonable members of the sports MSM.
September 16th, 2010 at 1:35 PM ^
We can tell that this is a different team than the one that started last year 4-0. but these guys have to study more than one team and it's not as clear to them. We've had 2 wins over quality opponents but neither was ranked. They have moved us up from 70ish to start the year. All we have to do is keep winning.
September 16th, 2010 at 1:37 PM ^
This topic is more of an Ohio State thing:
(Yes, it is SFW)
September 16th, 2010 at 2:08 PM ^
I'm not complaining about the fact we're not ranked. I'm complaining about the rationale for not ranking us. He did not say, "Well, they haven't been very good the last couple of years." He said, "We don't know yet because we don't know how good the teams they've beaten are."
We don't know that about ANY teams yet. It's impossible to know two weeks into the season. Impossible. Otherwise, M wouldn't have lost to App State in 07 and subsequently beaten Florida the same year.
If this were an OSU thing, I'd be bitching, "We're not ranked! This sucks!" But I'm not. I'm questioning the reasoning.
Also, I hate you forever now. Do NOT call me a Buckeye. Seriously.
September 16th, 2010 at 3:23 PM ^
We're ranked in both polls (#20 in the AP, #22 in the Coaches').
September 16th, 2010 at 1:42 PM ^
When you create a resume, in real life, you don't just list your current job. You list the stuff you've done for the last few years, your education, your skills, and references who will say how good you are. Employers weigh all of this information and "rank" prospective employees that way.
To use a "resume" of two games to rank teams would be like looking at one item on a resume for each applicant. Michigan's two wins are probably more impressive than Florida's so far, but there is significant evidence that Florida is better--their performance last year, the track record of their coaches, their returning players, and their recruiting rankings, just to name a few.
September 16th, 2010 at 3:18 PM ^
Have you seen Florida play this year?...
I think your example is hurting your argument rather than helping it.
September 16th, 2010 at 1:49 PM ^
Who is Andy Staples? My point exactly.
September 16th, 2010 at 1:53 PM ^
Hey, Dr. Saturday has Michigan No. 1 in his blogpoll, so strict resume voting doesn't really work either.
Even at the end of the season, resume voting isn't perfect. For any team, you're going on, at most, 13-14 games against vastly different schedules. So you're going to have some games that a team lost which it would win 99 times out of 100.
I don't bother with the blog poll early in the season because the resume voters screw it up by applying an end-of-season formula. It would be like someone saying a flipped coin will always land heads because it landed heads in two trials. If you want to predict results, you need to use projection, especially when you have as little data as any length of a college football season presents you. There is no way to determine a completly fair NCAA football champion with a schedule of over 100 teams, of which each will have played an inherently uneven schedule.
I don' t think polls are useless, even if they are necessarily imprecise. I think they are as good as the data put behind them. If part of that data is past-season performance of returning players, and scouting, and depth projection based on recruiting over the last five years, that is more data. Polls should be based on both performance and expectation. If I were Brian, I wouldn't count the resume voters before Week 8; they -- and in this I unfortunately include the otherwise respect-worthy Hinton -- are purposefully leaving out important data that would make the poll more useful toward its ultimate purpose (projecting future performance) over a misguided adherence to a false pretense.
As stupid as Staples's poll is for ignoring the two games we have seen so far (Denard's play and the positives from the front 7 certainly give Michigan the ability to beat most or any other ranked team), it's still better than strict resume voting, which even the best such pollers (e.g. Hinton) are basically rewarding teams for scheduling non-cupcakes before the same cupcakes everyone else faces.
Honestly, I don't think Notre Dame nor UConn will finish in the Top 25 this year. I think Michigan should be ranked between 15 and 25, which concurs with the resume and what we know outside of the football field. Anything beyond that is overweighting one factor or the other.
September 16th, 2010 at 2:08 PM ^
through week 8 is fine, as long as you're willing to throw any non-resume voters after it.
At some point historical projections have to be removed from the equation when determing the best team for this season.
September 16th, 2010 at 2:39 PM ^
No, not really. Projections just need to diminish as each resumé gives us more data.
Let's go back to the coin analogy. After one flip of the coin, our resumé says heads always lands up, but our projections are still at 50-50. After two flips that give us heads both times, we now have a slightly stronger resumé for Heads dominance, but our projections are probably still more accurate, right?
Our projections are not based in thin air. They are based on the fact that we perceive very little weight difference or shape difference on either side of the coin that should favor Heads over Tails.
After 1,000,000 tosses, if we have found that HEADS will come up 75 percent of the time, obviously our projection is not as valuable anymore, and our statistical data is pretty conclusive that Tails will usually lose to Heads.
Make sense? As the season progresses, we won't get anywhere near a million trials, but we will have progressively more and more direct trials to accompany the circumstantial evidence. Michigan will win or lose the remaining games on its schedule, and the manner in which that was accomplished will determine our final rankings.
There's nobody in the blogpoll who will keep his rankings entirely based on pre-season projections through the end of the season, therefore there would be zero non-resumé polls to throw out. However, since resumé voters are willfully ignoring evidence -- even if that evidence is circumstantial -- they would not be good jurors. However, by the end of the season, they are much better jurors, since they're voting on the same smoking gun that the rest are.
September 16th, 2010 at 3:14 PM ^
While theres no powerpoll voters that "keep his rankings entirely based on pre-season projections through the end of the season". Theres also no resume voters that will entirely remove subjective powerrankings either by week 6, probably even earlier. "therefore there would be zero non-resumé polls to throw out" also applies to resume-rankers prior to week 8. [To my knowledge no voters replicate or use a pure algorithm, otherwise I'm wrong.]
With only 12 or so trials, and most of them so heavily weighted as to be unworthy of inclusion, you never get anything close to a quality sample size. Just because your experiment is concluded doesn't mean you have sufficient data. If you follow your line of reason, its reasonable to keep your preseason opinions through an entire season because of a lack of data to counter it. I feel confident (based on my own subjective opinions and supported by gambling lines) that USC was probably the best team in each of the last 5 or so years, regardless of going undefeated or having a 2 loss record. The evidence on the field (close losses in tough situations) hardly swayed my subjective opinion - USC was still going to beat Boise or Florida or Texas or anyone else that they played head-to-head more often than they'd lose.
At some point you have to discard that subjective opinion and make a decision that results matter, not opinons. Otherwise, its a popularity contest with a few bits of evidence included (or excluded) arbitrarily.
Personally, I'd like to see each voter start with a quality computer poll (or an index of them) thats based entirely on results. Any deviations from the computer poll should be justified with evidence that the computer fails to quantify.
September 16th, 2010 at 4:30 PM ^
If your point is that polls are subjective, I agree.
If your point is that there's no way to get a perfect ranking, I absolutely agree.
But I completely disagree with the notion of giving up because perfection is not attainable. The goal is to get the best possible poll, using all available data.
The best resumé voter I can name is Matt Hinton. Yet he has Michigan at No. 1 this week, because his poll willfully ignores a large amount of circumstantial evidence that we are not the best team out there. He readily admits that his poll will change next week simply by virtue of Michigan playing U-Mass. That Michigan will play U-Mass next week is already known, as is every other team's schedule.
BHGP, another resumé voter, tends to do things just as crazy, but with even less regard for the quality of wins (Hinton at least uses the BCS>Mid-Major>FCS preconception). So when you say "Theres also no resume voters that will entirely remove subjective powerrankings either by week 6, probably even earlier," that's true, but they have removed enough subjective information so as to have results that are neither supportable, predictive, nor in any way useful except to say "hey, at least Michigan played two BCS schools."
Ultimately, these polls are entertainment. Because standings hardly matter anymore, since the game became national instead of regional, what we want as fans is for our polls to reflect the strength of our teams, and for winning and losing during the season to reward or punish accordingly. The best way to do that, in my opinion, is to predict the outcome of remaining games. This means starting with a poll that is entirely based on circumstantial evidence, and replacing each prediction with the result as each week progresses.
The final poll should most closely reflect the strength of the season that was played by each team. The first poll should most closely reflect the predicted strength of each team. The second week's poll should take into account the games played, and the projection for each future game. If the manner of winning or losing each game, and the strength of opponents, is added to the equation, what we end up with is a poll that each week rewards winning, rewards beating better teams, and ultimately resembles the value of each team through the season.
If you think Michigan is the 2nd best team in the land, and Ohio State is the first, and Ohio State beats Michigan in Columbus by 3 points after a questionable call, and no other team has had as good of a season as Michigan, then yes, Michigan should stay ranked No. 2. If undefeated USC loses to undefeated Texas in a neutral-site bowl game, and you think USC should be ranked above Texas because USC has Reggie Bush, that is your perogative as a voter, but I would point out it's a stupid one since I can show you a game where Texas beat USC on a neutral field.
There's no such thing as a perfect poll. But we should always strive to create the best poll we possibly can. Resumé voters are not striving to make the best poll they can, since they ignore data.
Completely off-topic (or at least not having anything to do with our argument): If Team A had three solid BCS wins and lost to Team B; and Team B had two solid wins plus a cupcake, and had a loss to a decent team, which of these should be ranked ahead? I ask because 1) I don't really know what I'd do, honestly, and this has been bugging me for over a decade; and 2) because a strict resumé convention would say Team A was stronger, even though they lost to Team B and had an identical record, and though I think that's technically right, I just hate the idea of ranking a team directly below the team they beat, and this happens all the time!
September 16th, 2010 at 6:14 PM ^
Using what you call "all available data" is pretty subjective in itself. Should I include the data that Michigan won many national titles prior to 1930? Fight song popularity? Regional bias? SEC speed? NFL draft hitory? etc. What is supportable is up for debate. To me, anything outside of results is hard to support because its inclusion is highly subjective. The notion that you can effectively capture, filter, weigh, and process ALL available data seems improbable at best, more likely -- impossible. Better to focus on the important data, the data that you know how to process, and focus on refining your analysis of that.
Your use of predictive elements, seems dubious to me. Obviously that approach has a place before the season starts, but once games are played I don't see a point in it. The Vegas bookies do predictive better than anyone and I believe a poll to that effect is already published, so why replicate it with less thought and insight? I also don't see the point of including games that have yet to be played, as you imply Hinton and others should. Most people take the poll to be a snapshot in time, not a predictor of who will be ranked where by the end of the season. This is where some notion of criteria would be beneficial.
As for what is useful...well, we'll just have to agree that regardless of the criteria or methods it makes for enjoyable discussion.
-------------------
As to your hypothetical conondrum, I'd focus on addressing the biggest weakness in the sport - cupcake matchups that bore fans. To lessen the attractiveness of these matchups they need to be rewarded less - something most people agree on even if they don't support it in practice (rankings). To take it the next step, losses against quality opponents need to be punished less. I'd take the extreme viewpoint that, until scheduling difficulty is improved dramatically across the college football landscape, losses should be ignored in favor of the resume of wins. Who cares who you lost to?--Show me a list of who you beat! I subscribe to the view that PSU shouldn't be punished for getting whooped by Alabama - everyone expected that to be the case. But Alabama deserves credit for beating a quality opponent (assuming they prove themselves to be one) instead of cupcakes.
In your example, I'd rank the team with more impressive wins higher, regardless of head-to-head outcome, particularly since there are many flukes. If VaTech goes on to a 2 loss season and James Madison losses all its games, you would rank VaTech higher. As for "strict resume", I'm not sure such a thing exists exactly. Someone like Hinton, who also looks at losses, wouldn't necessarily ignore the head-to-head outcome as you imply. He looks at losses as well and still includes subjective weighting. He's not a computer.
September 16th, 2010 at 3:27 PM ^
Interesting question: why should predictions based on talent and experience matter? I mean, there's no correlation whatsoever between previous performance and this year's performance. Makes me wonder why there's so much fuss about this "Ingram" guy when we have no idea if he's any good at football, since he hasn't done anything this season.
I suppose I'm making my point in kind of a dickish way, but you see what I mean here.
September 16th, 2010 at 10:23 PM ^
I am fine with voters who rank the teams based on how good they think a team is during the season, just as long as the final poll is strictly resume.
I always reflect on the '97 season. Michigan played a more difficult schedule than Nebraska that year, but a lot of the voters felt that Nebraska was the better team. I always found that rationale to be devoid of merit. If you are going to award somebody a championship, it should be awarded on the basis of what they have accomplished, not on the basis of what they are hypothetically capable of doing in the future.