My post was in response to pamarcole's post - I would agree that it had little to nothing to do with our discussion. However, again, it was in response to a different discussion within this thread.
Kirk was on 670 The Score here in Chicago today
- Thread starter LaQuintaHawkeye
- Start date
It was at this point I realized your entirely tongue in cheek. KF is very likely to get a statue of some form to honor him for being a great man or builder of the program. Not one person brought up winning percentage when there was a poll. It's not personal, its data.
To be honest, the type of offense you have will dictate the defense.
NFL style offense is more controlled, consistent. The power game seems to be strongest with this type.
Fast-paced offenses can score at a quick rate and put a team out of the game fairly quickly. However, the downside is the defense. It tends to be very vulnerable to power.
If I had a choice to be like Wisconsin or Oregon (under Kelly), I would choose Wisconsin. Teams that are fast paced can be put out of commission quickly if you establish the hardest hitting.
NFL style offense is more controlled, consistent. The power game seems to be strongest with this type.
Fast-paced offenses can score at a quick rate and put a team out of the game fairly quickly. However, the downside is the defense. It tends to be very vulnerable to power.
If I had a choice to be like Wisconsin or Oregon (under Kelly), I would choose Wisconsin. Teams that are fast paced can be put out of commission quickly if you establish the hardest hitting.
http://www.espn.com/college-football/statistics/teamratings/_/sort/projectedWPct
Perfect example. Georgia. They are not projected to win 11 games they are projected to win 10 but are Much Much more likely to finish with 11 than 9. Same thing. You dont round in stats.
Perfect example. Georgia. They are not projected to win 11 games they are projected to win 10 but are Much Much more likely to finish with 11 than 9. Same thing. You dont round in stats.
We use thirteen games because that’s how many games are played. Most of us expect to make a bowl game, it’s basically a given with our light pre-conference schedule.
If your going to include the bowls and try to say he is 8/13 you are only further from accurate. While 8/13ths is closer to his actual win percentage (.596) than either 7/12 or god forbid 8/12ths he is under .500 (7/15) in said bowl games. Again there is no rounding in stats. He is a "7 win a year coach" no matter how you slice it. If he wins 7.5 games a year and loses 5.1 (his entire body of work) and plays 12 regular season games how man wins does he have prior to the bowl game?
Last edited:
There's a little thing called the expectation value (out of probability theory) - and within that context you absolutely do not round down like you're suggesting. The expectation value of the random variable is also known as the mean (i.e. the average) and is simply the first moment of the distribution. If you want to speak about the likelihood of an event, you need information about the full distribution (which, in principle may require knowledge of an infinity of moments). It just so happens that some random variables are distributed in such a way that a closure exists wherein you can discern the full characteristic of the distribution from on the first 2 moments (this scenario is rather famously known as the central limit theorem).
In the context of speaking about college football, we cannot safely make any assumptions about things being normally distributed - so even the aforementioned scenario does not apply.
You may wish to address the fact that fundamentally there are no such things as a "part of a win" ... however, in order to engage in the sort of discussion you were trying to speak about, you cannot say anything more rigorously (like you were trying to) without more information about how spread out the distribution is about its mean. This information that tells us about the "spread" is related to the higher moments that I had mentioned before.
I hope our next coach does well right off the bat. If they only win three games their first year you'll be screaming for their head since they'll only be a three win coach.
Trying to find a long term average based off 1 data point would be gravely irresponsible, but it is still 1 point of data. We are talking about how many wins per season, we would only have 1 season.
Last edited:
What homer is talking about here is the curve. Most people think of a standard bell curve with an equal distribution however bell curves dont exist in the real world. If KF's data were a perfect bell curve he would have as many 10 win (regular season 12 games) as he does 4. Obviously he does not however he does have more 8s than 6 and what helps is alot is more 9s than 5's. Instead of a bell it would.look more like a tear drop since 8 is so much more likely than 6. He is still a 7 win a year average coach. P.s Homer you obviously have great knowledge of how stats work, why have you been sitting on your hands my friend? I figured the "you dont round to piss" and the "if you play 12 games how many have you actually won" took the "half a win" answer off the board but your point is probably more direct.
Last edited:
Coming up with a satisfying interpretation of stochastic and random processes and quantum mechanical processes is not without its own set of problems. Our conceptual explanations (written in a language like English) do not always match up well with the mathematical underpinnings of what is going on. For example, in order to supply a conceptual understanding of a single random variable - mathematicians often rely on applying ergodic arguments applied to an infinity of random variables - because it matches up with empirically derived distributions (which are conceptually more simple to think about). Unfortunately, the conceptual picture fails miserably if the random variables aren't independent and don't satisfy the requisite ergodicity requirements.
At my core, I'm a quantum field theorist. You can view classical probability theory as being a special case of quantum field theory wherein you treat all your (probability) density operators as being commutative. One of the primary complications that the "quantum" side of things affords is that your density operators are inherently non-commutative (as are your algebra of observables).
So, after 4 seasons were you ready to dump Ferentz since he had a full 4 seasons and was only a 5 win coach?
Just because something has an average doesnt mean there are not trends in the average. If your question is "did you want to dump him after an 11win year when we were all able to see progress" the obvious answer is no. If kirk would have gone 4 wins, 6 wins, 5 wins, 5 wins...to get to an average of 5. Yes there would have been alot of people 2nd guessing the hire. The level of heel digging by the "not 7-5" crowd is amazing.
More relevant question? Hes yet to coach a game. Likely has the full duration of his contract unless he does worse than the coach before him, which I doubt but you never know. Id argue that its more relevant to talk about KF as he may have 5 years or less depending on what he wants to do. I dont see a scenario where hes fired baring consecutive terrible seasons, which I doubt but you never know.
But to your question, Ill bite. I have no clue. But someone will end up coaching after Frost if thats in 5 years or 25 years so Im not worried about it.
True, but no-one can seem to get that through your head
I realize that Frost hasn't coached a game in Lincoln, but your missing the point of my inquiry. Most of your Neb posters come here and regal us with stories of the havoc that Frost will wreak upon the BIG west. Same crap we've heard numerous times before, ONLY NOW their really serious, because this is the prodigal son. NOW you've really got your guy! OK, good enough, but if he doesn't win some division titles at least, what then? From outside appearances the guy looks the part, but he's never coached a single game as a power five head coach, so we'll see........
If you can show me how Kirk averages 8 wins a year, using a traditional schedule, while not removing any of the data, and not segmenting the data, while not rounding, I'm all ears. In other words, I've shown my math every day possible, now it's your turn. You guys can't figure out how to get to .615 (8/13) a number that isnt even relevant, let alone an actual 8 win a year coach 8/12 (.666) Remember, no rounding, no removing any data. Go for it. I have been wrong before.
Last edited:
Ps. I LOVE THAT THIS THREAD STAYS AT OR NEAR THE TOP OF THE LIST. Anyone with a working knowledge of statistics and no bias is going to really get who is intelligent and who isnt.
Iowa12, if you want to criticize Homer’s posts, you are in the wrong place.
His most recent post was discussing statistics and the variables that make it difficult, which admittedly isn’t a thrilling topic, but his other posts are thorough, thought provoking and all about Iowa football.
We will see what kind of poster you will become, but so far ... not a good start.
His most recent post was discussing statistics and the variables that make it difficult, which admittedly isn’t a thrilling topic, but his other posts are thorough, thought provoking and all about Iowa football.
We will see what kind of poster you will become, but so far ... not a good start.
So, what I'm hearing, is that you acknowledge the trend upwards and that the first few seasons were not indicative of his work at that point. So...that was the case in '02 but not now? How many 1 win seasons has he had since then? How many 3 win seasons? Are you really going to argue the case that they should be considered relevant when evaluating his body of work? If so now, why not then?
The data needs to be included in order to show reliable data. If you want to say looking at the most recent 5 years is more valid than his entire body ofnwoek you might have a stance but your sacrificing reliability for validity ar the point. Having all the data included makes the information more reliable if you want to say it's worth the reliability for what you feel is more validity. Your case study would be crushed by anyone with a working knowledge of stats.
Second this. If there's anyone on this board that deserves complete respect, it's Homer.
If someone thinks it's invalid to throw out fliers that have a clear and obvious reason for being fliers and that reason has zero impact on the remaining data, it not only makes sense to throw it out, it's misleading not to. But carry on. I've said my piece. I'm more than comfortable with my position and I know that no amount of logic, regardless of how sound, will affect your position. You're squarely in "I've made up my mind, don't confuse me with the facts" mode, so, again, carry on.
I got your point and thats great, but Im not doing that. What happens happens. If he has success great! If he doesnt, hopefully the next coach will. My expectations are realistic. Im fully cognizant that the landscape has changed dramatically and until all schools are forced to play by the same rules it will always be lopsided. I want Nebraska to be competitive, I want Iowa to be competitive. I want the West to be competitive and get some due respect. Id love to see a B1G West team in the playoffs. Obviously would be thrilled to see Nebraska there, but wouldnt be jealous or bent out of shape if it was Iowa or Wisconsin or any other West school. Frost may get us there and he may not. Im not sure even the biggest prognosticator of Nebraskas demise however would argue that Frost will bring more to the table than Riley did in terms of fielding a decent team. But again, who knows and what happens happens.
I missed this earlier and must say I am thourghly enjoying your contributions. If I may attempt to put this in lay mans terms, and please correct me if I'm wrong.
In the real world it is "impossible" to label a coach a "8-4" or "7-5" etc coach because the team they are playing against changes game to game and year to year. This is equally true no matter what label you are trying to put in the coach. If we truly want to know if KF was a "7-5" it would be best if he was able to always play against the exact same team on the exact same day while KF got to change his variables.
Ps. Homer I would love to look at some other data sets with you. My background is a M.A.C. (counseling) and with that I had to take some higher level statistics and conduct several studies while running data analysis on my findings. You sir have that beat by a mile. I would love to hear your input on variables I probably have never considered.
Never once in the history of science has a study had more reliability due to removing data. It's amazing how I responded "if you can show me your math i have been wrong before" and your response was "you only believe what you want to believe." Its more likely you are incapable of using the information and providing "proof". You just did the message board equivalent of "I'm rubber your glue and I'm taking my ball and going home".
As long as you're comfortable standing pretty much alone on this...carry on.
If my options are being correct or being with the masses I'll always choose correct. I may go against the trend of the boards every now and then but I tend to.provide alot of data to support my stance.
Hold up, are you punking me? Your way to close to center to be a Nebraska fan?????
No, Im old enough to have seen the best of times (90's run) and unfortunately the worst of times. I was fully aware of what was happening in the glory days. I.e. lightning in a bottle. I took it all in knowing full well I may never see it again. Frost is exciting for the obvious reasons and Im cautiously optimistic he will do well. But Im not about to put the cart before the horse. He has a big hill to climb to even get back to what Pelini was doing. Im hopeful he gets there and if fans are thinking anything other than some ups and downs next year they need to wake up.
Take a more advanced statistics course - and then you'd find the content in my words. My apologies for being lazy ... I didn't feel like translating the mathematics to lay-speak. I suppose that you wouldn't appreciate the relationship between the Radon-Nikodym derivative and the probability density either, eh? Your loss ... it's awfully cool stuff.
<----- googled radon-nikodym derivative. I'm a little rusty but it is cool stuff.
https://en.m.wikipedia.org/wiki/Radon–Nikodym_theorem
Not going to lie Homer, that one is above my education level.
https://en.m.wikipedia.org/wiki/Radon–Nikodym_theorem
Not going to lie Homer, that one is above my education level.
Depending on the resolution of your data-set and/or your ability to acquire more data (and integrate it in a seamless fashion into your overall data-set) ... typically the resolution doesn't let you get reliable measurements past the first few moments. That's why people tend to be so lazy and teat everything as if it were normally distributed. However, this is a pretty big mistake - because, if you're investigating something that is susceptible to rare-events ... then such things are characterized by having heavy tailed distributions. When you coarse-grain descriptions of such things - you often run into examples of Fisher-Tippett distributions (like the Gumbel distribution).
BTW - thanks for translating some of my statistics posts into something more digestible.
At this point I'm not interested in explaining the numbers anymore to the peanut gallery if you dont get it you dont get it. Homer is over here dropping Pearl's of wisdom and I want to take advantage of that opportunity. If you cannot figure out how numbers work, that there is no such thing as rounding in statistics, what reliability and validity are, what a good sample size is, or the most basic fact that we play 12 regular season games so you should literally point and laugh at anyone who tries to say "8/13" that's between you and your God. You might notice there is a guy over here talking in a language most of us have only seen on a white board while watching big bang theory who's only objection to what I'm posting is "it's hard to put a diffenative label on something in a world where the opposition or "control" changes. (That's true whether you label him 7-5 or 8-4)
Last edited:
There IS such thing as good data versus bad data. If data is collected in an exceedingly careful fashion and it is meticulously annotated ... then YES, it would be a mistake to disregard such data out of hand. In fact, many interesting results in science have resulted from coming up with new models to accounted for such "interesting" (and seemingly anomalous) data points.
On the flip side, if data is taken in a relatively more haphazard fashion and is poorly annotated - then such data can actually lead to incorrect conclusions and screw up results. For example, in micro-biology, when doing careful phylogenetic analysis .... skilled bioinformaticians usually have to exclude Craig Venter's big data dumps because they're so poorly annotated. If you're unfamiliar with Venter - he's the same guy who wanted to patent the human genome (thankfully public entities "beat him" ... and were able to block his attempts). However, later in life, he rode around the ocean shotgun sequencing the crap out of samples from the ocean. However, there are all sorts of problem with his data ... like the quality of filters he was using (it's likely that his data also includes genomes of giant viruses - but you wouldn't know it from his data). He was of the opinion that the more data he could get - the better. Unfortunately, so much of his data was acquired in a sloppy fashion and wasn't properly annotated ... that it's still not yet clear if his data dumps did more harm than good.
