How Nebraska Would Have Done Elsewhere

[Mierin]

Statistics in the end don't mean near as much as people play them up to mean.

Yes they do. There are just a lot of variables in this.

 

[It'sNotAFakeID]

Statistics in the end don't mean near as much as people play them up to mean. Especially in the world of college football, where there is sooooo much parity among teams top to middle. Chances are, we would have fared much better because we would not have had to endure a shift in recruiting philosophy in terms of the types of players we recruited on the defensive side of the ball. We could have continued to march 6 or 7 DBs onto the field and been just fine. In the B1G, that's not true at all. Success in our new conference starts at the line of scrimmage, which is why we've seen such huge struggles from our defense in the first 2 years of the switch. This definitely gets underplayed a lot more than it should. There's no way of truly knowing how we would have fared in the Big 12.

Which is why they play the game.

I agree, and there are so many variables that can't be measured that have a visible effect on outcomes. Just a players mood on that given day can have an effect on the outcome. But that doesn't make statistics worthless or even of little value.

 

[HuskerShark]

Statistics in the end don't mean near as much as people play them up to mean. Especially in the world of college football, where there is sooooo much parity among teams top to middle. Chances are, we would have fared much better because we would not have had to endure a shift in recruiting philosophy in terms of the types of players we recruited on the defensive side of the ball. We could have continued to march 6 or 7 DBs onto the field and been just fine. In the B1G, that's not true at all. Success in our new conference starts at the line of scrimmage, which is why we've seen such huge struggles from our defense in the first 2 years of the switch. This definitely gets underplayed a lot more than it should. There's no way of truly knowing how we would have fared in the Big 12.

Which is why they play the game.

I agree, and there are so many variables that can't be measured that have a visible effect on outcomes. Just a players mood on that given day can have an effect on the outcome. But that doesn't make statistics worthless or even of little value.

I agree. Which is why I didn't say that.

 

[It'sNotAFakeID]

Statistics in the end don't mean near as much as people play them up to mean. Especially in the world of college football, where there is sooooo much parity among teams top to middle. Chances are, we would have fared much better because we would not have had to endure a shift in recruiting philosophy in terms of the types of players we recruited on the defensive side of the ball. We could have continued to march 6 or 7 DBs onto the field and been just fine. In the B1G, that's not true at all. Success in our new conference starts at the line of scrimmage, which is why we've seen such huge struggles from our defense in the first 2 years of the switch. This definitely gets underplayed a lot more than it should. There's no way of truly knowing how we would have fared in the Big 12.

Which is why they play the game.

I agree, and there are so many variables that can't be measured that have a visible effect on outcomes. Just a players mood on that given day can have an effect on the outcome. But that doesn't make statistics worthless or even of little value.

I agree. Which is why I didn't say that.

Okay, then the better question I should've asked you is, "how much value do you think statistics have?"

 

[tschu]

The mistake most people make when looking at statistics are A) viewing the stats as an absolute measure of a team's strength when in fact the stats are simply a messy variance-riddled representation of the team's strength and B) viewing predictions made by stats as absolutes instead of probabilities. (In biology terms, it's a little bit like trying to figure out an individual's genetic code by looking at its phenotype. Kinda.) But you can definitely glean a lot of useful things from stats.

The bad thing about football is that there are relatively few data points due to only playing 13 games. The same reason that football stats are so hard to deal with is also the same reason that sabermetrics works so well in baseball - you get 162 games and hundreds of clean data points from each player.

 

[HuskerShark]

Statistics in the end don't mean near as much as people play them up to mean. Especially in the world of college football, where there is sooooo much parity among teams top to middle. Chances are, we would have fared much better because we would not have had to endure a shift in recruiting philosophy in terms of the types of players we recruited on the defensive side of the ball. We could have continued to march 6 or 7 DBs onto the field and been just fine. In the B1G, that's not true at all. Success in our new conference starts at the line of scrimmage, which is why we've seen such huge struggles from our defense in the first 2 years of the switch. This definitely gets underplayed a lot more than it should. There's no way of truly knowing how we would have fared in the Big 12.

Which is why they play the game.

I agree, and there are so many variables that can't be measured that have a visible effect on outcomes. Just a players mood on that given day can have an effect on the outcome. But that doesn't make statistics worthless or even of little value.

I agree. Which is why I didn't say that.

Okay, then the better question I should've asked you is, "how much value do you think statistics have?"

I think they give a good indication of a team's performance, as long as you account for the defenses/offenses that they go up against. A good example would be this: Say hypothetically that Oklahoma and Nebraska both have the exact same offensive statistics for every measurable category: rushing, passing, turnovers, etc. Which would be more impressive? No doubt, NU's. The defenses in the B1G are WAY better than the Big 12 IMO.

My view on statistics is that I take them at face value, but the eye test is much more important to me, as well as at what point in the games the team is getting its production, and against what opponents, at what time in the season, on and on and on. There are soooo many more important things than the statistics themselves. Stats are a result. The more important thing is how you came about those results. And I think that applies to life as well, not just sports.

 

[It'sNotAFakeID]

The mistake most people make when looking at statistics are A) viewing the stats as an absolute measure of a team's strength when in fact the stats are simply a messy variance-riddled representation of the team's strength and B) viewing predictions made by stats as absolutes instead of probabilities. (In biology terms, it's a little bit like trying to figure out an individual's genetic code by looking at its phenotype. Kinda.) But you can definitely glean a lot of useful things from stats.

The bad thing about football is that there are relatively few data points due to only playing 13 games. The same reason that football stats are so hard to deal with is also the same reason that sabermetrics works so well in baseball - you get 162 games and hundreds of clean data points from each player.

A power problem. I hate power problems. And the only ways you can get more data points is 1) breaking it down to a game by game basis, which for a cluster analysis is ridiculous because then you're going to get OU in cluster 3 in some games, cluster 4 in some games, and cluster 1 in other games. Which one cluster do you choose to put them in? The one they appeared most in? Okay, then why not just use the season averages? A lot less data entry for the same results. The way you can get more data points is by looking back a couple of years but that also poses a problem: the 2010 Nebraska team wasn't the same team as the 2011 Nebraska team.

All I'm trying to do is to take a statistical snapshot of the year in an attempt to answer the question about how Nebraska would have done in other conferences. The model isn't perfect (i.e. I had to remove Kansas from the same cluster as Oklahoma, Nebraska, and Texas because it was dragging those teams down behind Oklahoma State and Baylor. And while Kansas State was dragged down by Iowa State and TCU, all the teams were similar enough such that it was safer to assume that Kansas State just flat out overachieved in 2012. I tried to balance the difference between the offensive/defensive attacks faced by Nebraska and the offensive/defensive attacks faced by teams of other conferences by including strength of schedule, which indirectly should measure the offensive and defensive strength of opponents.

 

[skersfan]

My guess is that a statistics person from Oklahoma, or Oregon, or Boston College would have a completely different idea about it. Too much going on, rain, snow, wind, team sick, tv delays for one team the other does not have them. Good refs, poor refs. when you play a rated team and who follows that team. SEC bully or ackron. I think we play the games for a reason.

Now if you can make predictions off of these statistic for next year, and they prove to be correct, then it is interesting.

Hindsight I guess is perfect, but doubt our porous Defense changed much in any league. If they could score, we were screwed.

 

[It'sNotAFakeID]

I think it's funny how I post that we would've finished about 7th in the SEC and about 4th in the ACC, and I don't get a whole lot of negative responses about "the statistics don't mean blah blah blah." I post that we would've finished about 2nd in the Big XII and all of a sudden that same statistical test, wrought about by the same process isn't as correct as it was when we were finishing 7th and 4th.

There are a lot of variables which go into predicting an outcome, some of which just can't be measured. All I'm doing is taking those variables which can be measured and making a summary of the findings. A person from Oklahoma, Oregon, Florida, or Kentucky can't cluster the teams differently; with those stats the program is going to spit out those clusters every time. The only thing they can deviate from me with is if they wanted to use more or less variables. If they wanted to not include Nebraska, a team from a different conference, then just exclude Nebraska from all my analyses, I'm betting the clusters remain the same.

Stats are stats, they are not subjected to bias on their own. The bias comes from the researcher and the only bias I've shown is by putting Nebraska ahead of Georgia Tech in the ACC, Mississippi State and Vanderbilt in the SEC, and Texas in the Big XII.

 

[VA Husker Fan]

I don't have too much of an issue with your conclusions. Maybe a spot here or there, and I don't really think we'd have beaten KSU either. My issue is with how you started off:

People on this board, when they look at our performances on the field ask the question, whether aloud or in their heads, "How would we do if we were in the SEC, the Big XII, the Pac 12, and to a much lesser extent, the ACC?"

The answers to those questions, at least for the 2012 questions can be found here as I'll be looking at how Nebraska's performance in the 2012 season would have fared in the SEC, Pac 12, Big XII, and the ACC, the legitimate BCS conferences.

How did I get here: Using a multitude of variables, I created a database with all of the data from Nebraska and from the SEC, Pac 12, Big XII, and ACC. In order to create a fair comparison, those stats were standardized as if they were measured using the same scale. Then, through the use of a cluster analysis, four distinct clusters were identified, and the means on those standardized values (as well as the unstandardized wins value) were compared.

So all that being said, here's how Nebraska would have fared in the Southeastern Conference, the best conference in the nation.

It comes off as you having the absolute answers. These are just projections or predictions based on statistics. This isn't how they would have done. It's how they might have done, based on the statistic data you included. It's probably better then Lee Corso waving his pencil around, but there's no guarantee.

The bias isn't just in interpreting the results. It also comes in with which data you include and how much weight you give it. I'm not saying you rigged the input to favor Nebraska, but you certainly could've played around with the variables until you got the results you wanted. Also, the program assumes all the data is valid, with no weighting (as far as I can tell) to weed out less meaningful garbage stats that might come when the game is in hand and the subs are in.

There are just too many human factors to say "this is it".

It is, however, an interesting in-depth statistical perspective on last season. You didn't try to make any projections onto 2013, and no one should, because the personnel changes too much from year to year.

 

[JJ Husker]

This is a great thread BBBXII. Great idea and very interesting. I never had the pleasure (many would say displeasure) of taking statistics classes. From my limited knowledge on the subject I think you've done a damn fine job, have been fair in your method, and made it more than abundantly clear that the results are not absolutes but rather statistical probabilites. I'll be +1ing your posts for quite awhile because....damn. Good job.

BTW- I prefer the written interpretation. It makes me feel less stupid that way.

 

[NUance]

Great way to look at and evaluate Husker football BBBXII. +1 to you. As an aside, one thing that stats don't consider very well our strategy changes for each game we play. Take the Penn State game from last season, for example. Leading up to that game our coaches saw something on film that they thought they could exploit. Might've been the way the Natty Lions keyed on us, or the way they played Taylor, or something about the individual matchups. I really don't know what it was they saw exactly. But during the game we ran the toss sweep 22 times. That's probably two or three times as much as we normally run off tackle. And it worked! I'm not sure how much, if any, all those toss sweeps affected the aggregate running stats for that game. Probably some, I'd guess. The point is, we tweak our strategy for each game. On both sides of the ball. That said, I suppose statistics are the best way to evaluate our strengths and weaknesses, and tendencies, against other teams.

Also, nicely done with the graphics in this thread, sir. :thumbs :

 

[It'sNotAFakeID]

I don't have too much of an issue with your conclusions. Maybe a spot here or there, and I don't really think we'd have beaten KSU either. My issue is with how you started off:

People on this board, when they look at our performances on the field ask the question, whether aloud or in their heads, "How would we do if we were in the SEC, the Big XII, the Pac 12, and to a much lesser extent, the ACC?"

The answers to those questions, at least for the 2012 questions can be found here as I'll be looking at how Nebraska's performance in the 2012 season would have fared in the SEC, Pac 12, Big XII, and the ACC, the legitimate BCS conferences.

How did I get here: Using a multitude of variables, I created a database with all of the data from Nebraska and from the SEC, Pac 12, Big XII, and ACC. In order to create a fair comparison, those stats were standardized as if they were measured using the same scale. Then, through the use of a cluster analysis, four distinct clusters were identified, and the means on those standardized values (as well as the unstandardized wins value) were compared.

So all that being said, here's how Nebraska would have fared in the Southeastern Conference, the best conference in the nation.

It comes off as you having the absolute answers. These are just projections or predictions based on statistics. This isn't how they would have done. It's how they might have done, based on the statistic data you included. It's probably better then Lee Corso waving his pencil around, but there's no guarantee.

The bias isn't just in interpreting the results. It also comes in with which data you include and how much weight you give it. I'm not saying you rigged the input to favor Nebraska, but you certainly could've played around with the variables until you got the results you wanted. Also, the program assumes all the data is valid, with no weighting (as far as I can tell) to weed out less meaningful garbage stats that might come when the game is in hand and the subs are in.

There are just too many human factors to say "this is it".

It is, however, an interesting in-depth statistical perspective on last season. You didn't try to make any projections onto 2013, and no one should, because the personnel changes too much from year to year.

Yeah, my word choice wasn't exactly the best. Instead of answer, it should have been answer based off these stats; instead of would have fared, it should have been could have fared; and instead of would have done it should have been could have done.

I agree, that bias is introduced into stats by the researcher, who can tinker around with the variables to produce the picture he wants to paint. But I assure you that that isn't the case here. I know better than that; I'm not trying to deceive anyone, because there really is no purpose in me saying that Nebraska would have been the best team in every conference--I'm not getting any additional benefit by doing this research other than passing time until the season begins.

When I looked at the variables, I wanted variables that showed strategy (pass attempt and rush attempt), strategy faced (pass attempt allowed and rush attempt allowed), offensive execution (pass completion, passing yards, yards per completion, and completion to touchdown %, rushing yards, yards per carry, rush attempts to touchdown %, and turnovers committed), and defensive execution (the same offensive execution variables with 'allowed' or 'forced' after them).

The model doesn't consider garbage time, but I'm not sure how much that would alter the teams' averages.

NUance, that's why I included the game-by-game, cluster vs cluster post. Because while each team is different, each team does share similarities with other teams within the conference. So looking game-by-game at each cluster vs cluster matchup really allows you to see which cluster each cluster struggled as a whole against. For example, OU and UT struggled against Oklahoma State and Baylor, but had no problem with Texas Tech. So it's fair that, if Nebraska were in the Big XII, they too would probably struggle against Okie Light and Baylor, but probably have no problem with Texas Tech.

 

[NUance]

NUance, that's why I included the game-by-game, cluster vs cluster post. Because while each team is different, each team does share similarities with other teams within the conference. So looking game-by-game at each cluster vs cluster matchup really allows you to see which cluster each cluster struggled as a whole against. For example, OU and UT struggled against Oklahoma State and Baylor, but had no problem with Texas Tech. So it's fair that, if Nebraska were in the Big XII, they too would struggle against Okie Light and Baylor, but have no problem with Texas Tech.

Yeah, that's the best way of looking at it. Most likely to produce an accurate comparison. :thumbs :

 

[HuskerShark]

I think it's funny how I post that we would've finished about 7th in the SEC and about 4th in the ACC, and I don't get a whole lot of negative responses about "the statistics don't mean blah blah blah." I post that we would've finished about 2nd in the Big XII and all of a sudden that same statistical test, wrought about by the same process isn't as correct as it was when we were finishing 7th and 4th.

I hadn't read a word of this thread until you posted your status the other day. It doesn't matter what the situation is, that's always been my view of stats. :dunno