GT should be ranked in the top 25

[bobongo]

It's correlated for years.

The object of the game is to win by scoring more points than you allow.

If two teams play the same opponent and team A wins by 3pts and team B wins by 21, most people would conclude that generally speaking Team B is probably better than Team A.

Your "logic" demands that we consider Team A and Team B equal and that we believe the inference that Team B is better to be garbage.

You are probably the only sports fan that thinks this way.

No, margin of victory counts, too, somewhat. I have noted that in previous posts. But you have to take this with a grain of salt. A team shouldn't be given undue credit for "running it up" vs. a team that goes conservative and rests its starters. This requires common sense and the computer that sits on the shoulders and not in the lap. But there is no transitive property in college football. Team A beats team B which beats team C which beats...team A. Thus it's the overall SOS and the overall record that must be considered. But it's refreshing that you concede that winning, which is scoring more points than your opponent, is the object of the game.

That's my whole point - if winning is the object of the game, why consider stats, which are not the object of the game?

 

[GTJake]

As for UGAG, I really think we are in their heads ... they are "scared to death" to lose three in-a-row at home to us and lose to a HS offense the week before playing the #1 team in the country for the SEC Championship.
I think CPJ may be their boogey-man ... I'm not sure I'm predicting a win, but I think we have a shot !

 

[AE 87]

No, margin of victory counts, too, somewhat. I have noted that in previous posts. But you have to take this with a grain of salt. A team shouldn't be given undue credit for "running it up" vs. a team that goes conservative and rests its starters. This requires common sense and the computer that sits on the shoulders and not in the lap. But there is no transitive property in college football. Team A beats team B which beats team C which beats...team A. Thus it's the overall SOS and the overall record that must be considered. But it's refreshing that you concede that winning, which is scoring more points than your opponent, is the object of the game.

That's my whole point - if winning is the object of the game, why consider stats, which are not the object of the game?

First of all, your last question reflects something that has come up previously in your posts which may lie at the root of the disagreement: you're arguing, in part, against "stats" in the abstract. Unfortunately, you don't seem to recognize that it is contradicting your first sentence as well as other things you've said. Margin of victory is a stat. Strength of schedule is a stat. You can't claim only wins matter and not stats without putting UCF in your top 4.

Second of all, what you don't seem to understand by simply glossing my points/drive vs. power 5 as just a "stat" is that this stat is actually addressing the object of the game in a direct way, scoring more points than your opponents. When you break that down to its more basic level, the object of the game is for your offense to score more points than your defense allows. So, while there might be strange games where run-back scores and safeties etc affect the outcome, for the most part, the team whose offense scores more points/drive than their defense allows in any given game will win. Like, I said, you apparently just haven't understood the points/drive stat which is based on scoring more than your opponent because if you had you would not have labeled me agreeing that scoring more points is the object of the game as a concession. It's the basis of the stat!

So, what my stat says is that over the course of a season, as teams play more Pwr5 competition, the strength of opposition offenses and defenses begins to level out. For example, while we could score fairly easily against LOLvl whose D isn't very good, we struggled against CU and d'oh U whose D's are really good. The measure of the offense against a spectrum of power5 defenses begins to give an accurate measure of how well the offense scores. Similarly, the measure of the defense against a spectrum of power5 offenses begins to give an accurate measure of how well it keeps the other team from scoring.

In other words, the reason why my points/drive rankings match the poll rankings so well is that they are based on the same principle as the w-l record, scoring more points than you allow. By only using power5 opponents and requiring at least 3, it gives a rough accounting for strength of schedule. By using points/drive, it gives a rough accounting for margin of victory.
,
Finally, your reference to there being no transitive property is beside the point.

 

[JacketFromUGA]

As for UGAG, I really think we are in their heads ... they are "scared to death" to lose three in-a-row at home to us and lose to a HS offense the week before playing the #1 team in the country for the SEC Championship.
I think CPJ may be their boogey-man ... I'm not sure I'm predicting a win, but I think we have a shot !

For their players? I doubt it.

For their coaches? Maybe.

For their fans? Absolutely not.

 

[bobongo]

First of all, your last question reflects something that has come up previously in your posts which may lie at the root of the disagreement: you're arguing, in part, against "stats" in the abstract. Unfortunately, you don't seem to recognize that it is contradicting your first sentence as well as other things you've said. Margin of victory is a stat. Strength of schedule is a stat. You can't claim only wins matter and not stats without putting UCF in your top 4.

Second of all, what you don't seem to understand by simply glossing my points/drive vs. power 5 as just a "stat" is that this stat is actually addressing the object of the game in a direct way, scoring more points than your opponents. When you break that down to its more basic level, the object of the game is for your offense to score more points than your defense allows. So, while there might be strange games where run-back scores and safeties etc affect the outcome, for the most part, the team whose offense scores more points/drive than their defense allows in any given game will win. Like, I said, you apparently just haven't understood the points/drive stat which is based on scoring more than your opponent because if you had you would not have labeled me agreeing that scoring more points is the object of the game as a concession. It's the basis of the stat!

So, what my stat says is that over the course of a season, as teams play more Pwr5 competition, the strength of opposition offenses and defenses begins to level out. For example, while we could score fairly easily against LOLvl whose D isn't very good, we struggled against CU and d'oh U whose D's are really good. The measure of the offense against a spectrum of power5 defenses begins to give an accurate measure of how well the offense scores. Similarly, the measure of the defense against a spectrum of power5 offenses begins to give an accurate measure of how well it keeps the other team from scoring.

In other words, the reason why my points/drive rankings match the poll rankings so well is that they are based on the same principle as the w-l record, scoring more points than you allow. By only using power5 opponents and requiring at least 3, it gives a rough accounting for strength of schedule. By using points/drive, it gives a rough accounting for margin of victory.
,
Finally, your reference to there being no transitive property is beside the point.

Yes, w/l is a stat in a broad sense. But it is the very object of the game. MOV is a stat directly related to w/l, because outscoring your opponent is an object of the game. but should be weighed by a number of factors not quantifiable. But these are primary stats related to the object of the game, which is winning. SOS is necessary, so you're not comparing apples to oranges. The object of the game is not to run up secondary stats. And why some secondary stats and not others? Who decides that? I'll tell you who - people who have a preconceived notion about how the rankings should stand, and then put the template of stats over the teams that arrange them in that order. So the "objectivity" of those rankings is based on a subjective preconception.

And speaking of apples and oranges, what about this PPDvP5? What P5? Clemson, or Louisville? Alabama, or Arkansas? It's just ground up into one, big garbage stat.

I get your drift but I just don't agree. But I enjoy reading your posts. Thanks for responding.

 

[AE 87]

Yes, w/l is a stat in a broad sense. But it is the very object of the game. MOV is a stat directly related to w/l, because outscoring your opponent is an object of the game. but should be weighed by a number of factors not quantifiable. But these are primary stats related to the object of the game, which is winning. SOS is necessary, so you're not comparing apples to oranges. The object of the game is not to run up secondary stats. And why some secondary stats and not others? Who decides that? I'll tell you who - people who have a preconceived notion about how the rankings should stand, and then put the template of stats over the teams that arrange them in that order. So the "objectivity" of those rankings is based on a subjective preconception.

And what about this PPDvP5? What P5? Clemson, or Louisville? Alabama, or Arkansas? It's just ground up into one, big garbage stat.

I get your drift but I just don't agree. But I enjoy reading your posts. Thanks for responding.

I'm really not sure that you do get my drift. Your whole first paragraph is defending stats related to the "object of the game" but then reject points/drive scored and allowed, which are also directly related to the object of the game, as my previous post tried to explain. Yet, you still make a case against what you call secondary stats without addressing what I said about points/drive.

So, while I appreciate that you seem to have a philosophical objection to secondary stats. Let me say again, points/drive is not a secondary stat. It is a raw stat (7pts times the number of offensive Touchdowns plus 3pts times the number of field goals all divided by the number of drives). It's calculated for every game against Pwr5 competition, so it eliminates running up the score on FCS or Gp5 teams. It recognizes that when GT's offense scores 27 points in 8 drives, it is more efficient than another offense scoring 27 pts in 11 drives. Similarly, when GT's defense allows 21 points in 8 drives it is less efficient than another defense allowing 21 points in 11 drives.

Just so you know, that last observation gets to the origin of using points/drive to rank teams. I began by wanting quick and simple way to compare our offense to other offenses given that we often had fewer drives/game. The standard raw stat is points/game which doesn't work to compare Oklahoma and GT. Once I did it for the offense, I did it for the defense. I restricted it to opponents from BCS AQ and later Power5 conferences because the difference was also significant. Teams that average 3 points/drive playing 9 Grp 5 opponents are not as good as teams averaging 3pts/drive against 9 power5 opponents.

I then took the difference, the offensive points/drive vs pwr5 minus the defensive, to get a team ranking. The logic, again, was that while if you played only 1 power5 team, it would make a big difference whether it was CU or LOLvl, the more pwr 5 teams you played the more your average opposition becomes the average-power5-opponent common to all teams. As a result, the PPD difference becomes a measure of the margin of victory of your team against this average-power5 common opponent. Teams who beat this generic power5 opponent by the highest margin, have the highest difference between their offensive and defensive ppdvp5, would be the better teams.

So, again, my theory was that offenses who score the most points/drive versus BCSAQ/Pwr5 opponents were the better offenses and that defenses which allowed the fewest points/drive verse BCSAQ/Pwr5 opponents were the better defenses. I calculated the team ranking by differencing them as a way of checking the theory. I took the correlation between my ranking and the polls as validation that points/drive versus power 5 is a legitimate way of ranking offenses and defenses individually as well as teams as a whole.

 

[MikeJackets1967]

As for UGAG, I really think we are in their heads ... they are "scared to death" to lose three in-a-row at home to us and lose to a HS offense the week before playing the #1 team in the country for the SEC Championship.
I think CPJ may be their boogey-man ... I'm not sure I'm predicting a win, but I think we have a shot !

There is plenty of room for GT to be in UGAG's head since they're brainless

 

[ibeattetris]

I'm really not sure that you do get my drift. Your whole first paragraph is defending stats related to the "object of the game" but then reject points/drive scored and allowed, which are also directly related to the object of the game, as my previous post tried to explain. Yet, you still make a case against what you call secondary stats without addressing what I said about points/drive.

So, while I appreciate that you seem to have a philosophical objection to secondary stats. Let me say again, points/drive is not a secondary stat. It is a raw stat (7pts times the number of offensive Touchdowns plus 3pts times the number of field goals all divided by the number of drives). It's calculated for every game against Pwr5 competition, so it eliminates running up the score on FCS or Gp5 teams. It recognizes that when GT's offense scores 27 points in 8 drives, it is more efficient than another offense scoring 27 pts in 11 drives. Similarly, when GT's defense allows 21 points in 8 drives it is less efficient than another defense allowing 21 points in 11 drives.

Just so you know, that last observation gets to the origin of using points/drive to rank teams. I began by wanting quick and simple way to compare our offense to other offenses given that we often had fewer drives/game. The standard raw stat is points/game which doesn't work to compare Oklahoma and GT. Once I did it for the offense, I did it for the defense. I restricted it to opponents from BCS AQ and later Power5 conferences because the difference was also significant. Teams that average 3 points/drive playing 9 Grp 5 opponents are not as good as teams averaging 3pts/drive against 9 power5 opponents.

I then took the difference, the offensive points/drive vs pwr5 minus the defensive, to get a team ranking. The logic, again, was that while if you played only 1 power5 team, it would make a big difference whether it was CU or LOLvl, the more pwr 5 teams you played the more your average opposition becomes the average-power5-opponent common to all teams. As a result, the PPD difference becomes a measure of the margin of victory of your team against this average-power5 common opponent. Teams who beat this generic power5 opponent by the highest margin, have the highest difference between their offensive and defensive ppdvp5, would be the better teams.

So, again, my theory was that offenses who score the most points/drive versus BCSAQ/Pwr5 opponents were the better offenses and that defenses which allowed the fewest points/drive verse BCSAQ/Pwr5 opponents were the better defenses. I calculated the team ranking by differencing them as a way of checking the theory. I took the correlation between my ranking and the polls as validation that points/drive versus power 5 is a legitimate way of ranking offenses and defenses individually as well as teams as a whole.

Basically this. I've never mentioned using yards, turnovers, sack rate, whatever. Specifically points per play and points per drive. Points are the literal win condition of football, and normalizing them via play or drive ensures you are comparing them equally across offensive types. Teams are seeking to maximize offensive points and minimize defensive points so measuring how effectively they do so is not a secondary measure.

Running up the score is pretty easy to extinguish by removing plays and drive that occur once the game score reaches a certain threshold. Plenty of advanced stat sites do this.

 

[bobongo]

I'm really not sure that you do get my drift. Your whole first paragraph is defending stats related to the "object of the game" but then reject points/drive scored and allowed, which are also directly related to the object of the game, as my previous post tried to explain. Yet, you still make a case against what you call secondary stats without addressing what I said about points/drive.

By that extended logic, any stat is "related" to the object of the game. But no, the object of the game is to win, not compile points per drive. In addition, maybe a drive is used principally to drain the clock, and not just at the end of the game and not just to score points. So it isn't necessarily related to the object of the game, is it?

I'm done.

 

[AE 87]

By that extended logic, any stat is "related" to the object of the game. But no, the object of the game is to win, not compile points per drive. In addition, maybe a drive is used principally to drain the clock, and not just at the end of the game and not just to score points. So it isn't necessarily related to the object of the game, is it?

I'm done.

I think you may be the only person I've encountered who thinks that football coaches have a primary plan for winning games that does not boil down to their offense scoring more than their defense allows.

You saying that scoring more points than you allow is no more related to winning than having more passing yards suggests that you don't understand how football works or have some immature inability to admit you were wrong.

 

[bobongo]

I think you may be the only person I've encountered who thinks that football coaches have a primary plan for winning games that does not boil down to their offense scoring more than their defense allows.

You saying that scoring more points than you allow is no more related to winning than having more passing yards suggests that you don't understand how football works or have some immature inability to admit you were wrong.

I never said anything of the kind.

It's like this: Look at the scoreboard. What do you see? You see the teams that are playing (SOS), you see which team is on top (w/l), and you see how much they are on top (margin of victory). Simply put, that's all you need to know to draw a conclusion on who should be ranked where, because that's all that matters. And although I do appreciate your posts, I do not appreciate your insults. Good day.

 

[AlphaBuzz]

I never said anything of the kind.

It's like this: Look at the scoreboard. What do you see? You see the teams that are playing (SOS), you see which team is on top (w/l), and you see how much they are on top (margin of victory). Simply put, that's all you need to know to draw a conclusion on who should be ranked where, because that's all that matters. And although I do appreciate your posts, I do not appreciate your insults. Good day.

It isn't that complex. How do you rate Michigan against Alabama or Clemson if none of the 3 have played any of the other? This PPD systhem gives you a metric to compare the teams that haven't played each other. By your metric of "who scored more points", Clemson is better than Bama because they scored more on Loserville and we don't know where to put UCF because nobody has a shared opponent with them.

Final score is the ultimate arbiter, but PPD gives you a way to compare teams that haven't played each other.

 

[GTNavyNuke]

Just to introduce a concept we also beat to death concerning rankings:

The rankings y'all are trying to perfect are for a season average. But teams get better and worse during the season. Shouldn't more recent performance be more heavily weighted? And bowl bids more reflect the second half of the season?

 

[AE 87]

I never said anything of the kind.

It's like this: Look at the scoreboard. What do you see? You see the teams that are playing (SOS), you see which team is on top (w/l), and you see how much they are on top (margin of victory). Simply put, that's all you need to know to draw a conclusion on who should be ranked where, because that's all that matters. And although I do appreciate your posts, I do not appreciate your insults. Good day.

OK, for some reason you are just trolling. Not cool.

 

[ibeattetris]

Just to introduce a concept we also beat to death concerning rankings:

The rankings y'all are trying to perfect are for a season average. But teams get better and worse during the season. Shouldn't more recent performance be more heavily weighted? And bowl bids more reflect the second half of the season?

I think it would be a valid approach when comparing teams of similar record and strength, but only as part of the equation. If you lost your first three and then win your next nine, you didn't necessarily get better, those three games just might have been the toughest. If you lose your last three, maybe it was just because those were against the #1,2, and 3 teams. Context certainly matters when it comes to bowl game selections, but I am not sure I agree it should out weigh overall performance. You can't control who you play every year out of conference, so it seems unfair to penalize a team for playing a big OOC game early in the year (which it partially would if a big win early counts less later). If you were to take a rolling 2-3game average performance and compare it against you overall performance and use it as a "momentum" indicator, I think it could be useful.

Ranking are incredibly hard to do, and there are many ways to skin the cat. That is actually why I said earlier I prefer rankings that aggregate multiple other rankings. I feel like if you have 100 different people (with a mixture of computer and human) coming up with their own rankings, and you average them all together, you'll get a fairly good estimation of the current rankings.

 

[smathis30]

First of all, your last question reflects something that has come up previously in your posts which may lie at the root of the disagreement: you're arguing, in part, against "stats" in the abstract. Unfortunately, you don't seem to recognize that it is contradicting your first sentence as well as other things you've said. Margin of victory is a stat. Strength of schedule is a stat. You can't claim only wins matter and not stats without putting UCF in your top 4.

Second of all, what you don't seem to understand by simply glossing my points/drive vs. power 5 as just a "stat" is that this stat is actually addressing the object of the game in a direct way, scoring more points than your opponents. When you break that down to its more basic level, the object of the game is for your offense to score more points than your defense allows. So, while there might be strange games where run-back scores and safeties etc affect the outcome, for the most part, the team whose offense scores more points/drive than their defense allows in any given game will win. Like, I said, you apparently just haven't understood the points/drive stat which is based on scoring more than your opponent because if you had you would not have labeled me agreeing that scoring more points is the object of the game as a concession. It's the basis of the stat!

So, what my stat says is that over the course of a season, as teams play more Pwr5 competition, the strength of opposition offenses and defenses begins to level out. For example, while we could score fairly easily against LOLvl whose D isn't very good, we struggled against CU and d'oh U whose D's are really good. The measure of the offense against a spectrum of power5 defenses begins to give an accurate measure of how well the offense scores. Similarly, the measure of the defense against a spectrum of power5 offenses begins to give an accurate measure of how well it keeps the other team from scoring.

In other words, the reason why my points/drive rankings match the poll rankings so well is that they are based on the same principle as the w-l record, scoring more points than you allow. By only using power5 opponents and requiring at least 3, it gives a rough accounting for strength of schedule. By using points/drive, it gives a rough accounting for margin of victory.
,
Finally, your reference to there being no transitive property is beside the point.

Issue is that almost every poll has the same top 4 and top 9. It's the rest of the rankings where controversy comes in across the board as it comes all grey. Whole hearted disagree about SOS evening out. gT had the second easiest P5 schedule for their offense heading into Miami and was middlrmof the road for their Defense SOS. Your implication is that all conferences are equal and I find it impossible to believe you'll find any logical ranking that doesn't go SEC-Big12-big10-ACC-PAC12 based off this year. I get filtering out G5 teams helps eliminate issues with SOS but teams like Fresno, UCF and Cincy can reasonably put up a fight against most P5 teams

 

[AE 87]

Issue is that almost every poll has the same top 4 and top 9. It's the rest of the rankings where controversy comes in across the board as it comes all grey. Whole hearted disagree about SOS evening out. gT had the second easiest P5 schedule for their offense heading into Miami and was middlrmof the road for their Defense SOS. Your implication is that all conferences are equal and I find it impossible to believe you'll find any logical ranking that doesn't go SEC-Big12-big10-ACC-PAC12 based off this year. I get filtering out G5 teams helps eliminate issues with SOS but teams like Fresno, UCF and Cincy can reasonably put up a fight against most P5 teams

Thanks for the response.

OK, let's start with the basics: we agree that there is no perfect ranking. Of course there's going to be controversy about ranking, and that's part of the fun. In a game, you can crown the winner. In a tournament, you can crown a champion. However, when you look to rank which teams are better or worse than other teams, then it's not so cut and dried.

So, my posts about using my points/drive stat were not to argue that it's perfect. No ranking is perfect. They were not to argue that better approaches couldn't be made with more sophisticated approaches to strength of schedule etc. I was simply arguing that this was a fairly straight forward way of ranking teams which proves not to be "garbage" by its correlation with the polls.

With that being said, you raise an interesting point with respect to strength of schedule. My response is yes and no. Obviously, I'm using a raw stat and not a sophisticated calculation of individual strength of schedule. I don't see any easy way to use it to calculate conference strength. That's the fun of debating rankings.

Let's use an example. This week, the CFP committee ranked 2-loss LSU ahead of 1-loss WashSt, WVU, and tOSU. Neither the AP nor the Coaches did that. Both of LSU's two losses were to SEC teams: one was Bama but the other was to 3-loss Florida. Now, if you just kind of wave the SEC is the best conference flag as a way of blessing this, that's one way, but does that make it better or worse?

Now, what am I claiming with my stat? As I've said, I'm not claiming its perfect, and I agree that the strength of schedule argument works against it as much as it works against any other approach to ranking. I mean, we could scream at the TV, "YOU LIE!!!," every time they speak about the #4 offense or we can just allow that they are using an imperfect measure. I think points/drive is more accurate than total points to rank offenses, but it doesn't eliminate the problem arising from strength of schedule. How much of the scoring in the Big 12 arises from them having great offenses and how much from bad defenses? Again, that's the fun of debating.

Still, I AM NOT SAYING exactly that all conferences are equal. I AM SAYING that the Power5 conferences are comparable enough for the stat to still be meaningful. Here is a table of results after last week's games:

So, take it for what it's worth.

 

[MikeJackets1967]

Athlon in their Top 130 has Georgia Tech ranked #50

 

[RamblinWreck92]

Athlon in their Top 130 has Georgia Tech ranked #50

sounds about right. Until we beat someone ranked higher than #54, kinda hard to think otherwise.

 

[AE 87]

sounds about right. Until we beat someone ranked higher than #54, kinda hard to think otherwise.

Not if, you know, you're a fan.