The Baffler's been upgrading their online presence from "none" to "some" over the past couple of months and posting a lot of stuff from their print archives. The Northwestern decision made Hinton's old article timely again.
I'd like to see how strong the correlation is between bursts of profanity and changes in the win probability graph. The fanbase with the strongest correlation could be said to be the most intelligently profane in sports.
As long as Vin Scully is still working, Shulman's not the best play-by-play guy in sports, but he may be the best at working together with analysts. I brought up Buck Martinez because he was way better with Shulman than he's ever been working with anyone else.
Shulman's probably the best option for working with Vitale, but I think he deserves a new partner so the debate can be "Shulman: best play-by-play guy?" instead of "Vitale: tolerable or insufferable?"
I'm not a fan of Vitale because Dan Shulman has been one of the best play-by-play men in the business going back to his time calling Blue Jays games with Buck Martinez. But it's been ages since I've been able to enjoy him do his thing, because he's paired with someone who has absolutely no ability to take it down a notch.
Tlon, Uqbar, Orbus Tertius (Level 10) is at the top of the scale. It's for "otherworldly" awesome moments such as your favorite moment from the fourth quarter of Notre Dame UTL.
Gospel According to Mark is at the bottom of the scale for complete and utter disasters. Example: @Nebraska after Denard went down.
The order of the stories in the Borges-O-Meter is not correlated with my opinion of their quality as stories, The Library of Babel is much better than Level 3 as a story, but that's where it fits in best on the B-O-M.
I am glad mgoblog is still a place where dumbass mistakes are quickly corrected, even if they are my own. (However, I failed in correcting usage, not grammar.)
I'll cut the other guys some slack for blaming the Nebraska loss on the refs in the heat of the moment. Not the best attitude, but understandable. But blaming four losses on the refs? That's worthy of special disdain.
A column about the "junk wins" from CFB's early days could have been interesting, but Forde was too busy thinking his lame jokes were the wittiest japes of the season. "Something called Physicians & Surgeons of Illinois" is now known as the University of Illinois at Chicago. Case Institute of Technology (now part of Case Western Reserve) was a prestigious school - it was where the Michelson-Morley experiment was performed and, as Yeoman noted, a solid football program.
Worst of all to me, Forde abbreviates the California Institute of Technology as "Cal Tech." It is Caltech and always has been. He should get an editor and maybe do some research next time.
To be fair, isn't part of the reason he changed from Oklahoma to Indiana because Kevin Wilson went from being Oklahoma's OC to Indiana's HC? Sounds like a reasonable enough reason for that first switch, at least.
George Washington and Eastern Illinois. I guess if you really don't like Tony Romo that'll help you get your hate on. Regardless of the opponents, go Connie!
I did a Wilcoxon signed-rank test on Michigan's 8 Big Ten games to estimate the probability that the Kenpom predictions differ significantly from the final scores of the games. The set of differences between the Kenpom predictions and actual final scores is {-7,-1,-10,-25,+3,+20,-5,-6}. Based on those differences, I get a critical value of S=9, which corresponds to a probability of about 20% that the median score margin predicted by Kenpom is not the same as the actual median score margin. This is not statistically significant, so I won't reject my null hypothesis that Kenpom is doing an OK job.
Based on most of the numbers in that set of differences being negative, it wouldn't be unreasonable to wonder if Kenpom is underrating Michigan with respect to the rest of Big Ten, but there's not enough data to prove of disprove it.
The Oklahoma band must not think Boomer Sooner is catchy, seeing as they play it every damn time Oklahoma makes a good play. Do they think we'll forget it if we don't hear it 100 times a game?
If Dantonio didn't want to listen to what Hecklinski had to say, he could have taken out his phone and played some Angry Birds. If he wanted to be a dick, he could have "forgotten" to turn the sound off before he started playing.
If he'd done that, at least he'd have been an amusing dick.
No, kind sir, I am not a woman not was I outside the Sheraton in New Orleans. I do however appreciate you making mgoblog a community where we do not reflexively assume everyone is male.
If you look at our blog, you can also find designs for Martin, Van Bergen, and Heininger.
I don't think you can have a completely "fair" system when 120+ teams play 12-14 games each, but I feel a necessary requirement for whatever system is used is that it not have obviously unfair elements. A necessary requirement for the system not being "unfair" is that it not depend on what half-informed coaches or Harris poll voters had for breakfast the morning they sent their votes in.
I'd rather the system be fair than that it be stacked in our favor. It's unsportsmanlike otherwise.
On a cruder note, I want a system where there's never even the tiniest sliver of truth when Sparty complains the system shafted them because they're Sparty. A fair system forces them to own their failures and stop making excuses.
Go find the stuff he wrote for the Boston Phoenix before ESPN hired him. He was pretty good! He's more or less a hipster sportswriter. He stopped being cool when he got big.
Weasels and wolverines are both members of the family Mustilidae. That's as close as Brady Hoke gets to being a weasel.
As for GGERG Easterbrook, his ability to pass judgment on situations where he has no understanding is unsurpassed. In academia, it's perfectly acceptable for a professor to move from one school to another if the new school offers them better facitilies, resources, or reputation. If we're going to claim that football players are "student-athletes," then football coaches are "professor-athletes," and the ethics of them changing jobs should be the same as those for the professors. Moving on up in the academic world is a normal and completely ethical thing to do - the same should be true in the college football world. The difference between professors and coaches is that students can follow a professor to a new school and players can't follow a coach, but I think the issue there is the NCAA's lack of ethics, not the coach's.
Even someone like Todd Graham isn't a weasel for changing jobs frequently. His inappropriate conduct was not addressing his team face-to-face like an adult. The only coaches I can think of who may have acted unethically in taking college coaching jobs are Saban and Petrino, because they left NFL jobs, where the collegiate standards of behavior may not apply.
It's actually a compliment to VT to call them Tech as if no explanation was needed. The school that's earned the right to be called just Tech is Georgia Tech (more tradition, better research and academics - not that VT is terrible in any of these things.)
I've been told that they hate to be referred to as "VaTech" though.
I was going to put "only on a Michigan blog" as a tag on the post, but I decided to put in a Jonathan Lethem reference instead. Now that is something that would only happen on a Michigan blog.
Correcting an error: I wrote before the 2010 game, not the 2009 one. Still, we had only lost to them twice in a row, and I thought had a good chance of winning that one. :(
I have been waiting two loooong years to update this quiz, which I originally wrote in the leadup to the 2009 MSU game. At that time MSU had beaten U-M precisely once in a row. To give MSU their props, I am impressed that it did take them almost three full seasons between blowing leads in the 4th quarter.
Peter King is an alumnus of the actual Ohio University and the reason why I don't care if they find it annoying that we call the school in Columbus "Ohio."
Sorry if I sounded patronizing. I think that removing the outliers requires a good justification and I just don't see it in this case. I think the numbers look pretty bad for Michigan even with the long runs included. Sacks don't count in my analysis because they're passing plays not running plays.
If I were to pull outliers, I'd pull the highest and lowest designed run for each player, just to be symmetric. In this case, I'd just say, looking at all the data says Vincent Smith had a bad day, and pulling out the high & low runs says he had a really bad day.
I got a little carried away. I was in full on XKCD "Something is wrong on the Internet" mode.
They are different sure, but that's why you shouldn't assume taking the mean explains everything. You don't need to remove the outlier to understand why you'd rather have the second guy. My points are:
1) Don't remove outliers if they don't diminish your point. You don't need to remove outliers to show that Edwin Baker had a better day than Vincent Smith; and
2) Good procedure (which is different from commonly used procedure) is that you don't assume inconvenient data is an outlier. If it's an outlier, you explain why it doesn't belong. For example, in a data set I work with, we remove one subject as an outlier because the experimenter reported that he/she didn't obey the conditions under which the study was being conducted (he/she took some drug that wasn't allowed, I think). In that case, I've explain why it's not part of the distribution. That 91-yard run is just as much part of the distribution as all the others, as football is a game of a lot of short plays and the occasional long play. I used Barry Sanders as my example because that's how I remember him: lots of runs of 0,1, or 2 yards, and then occassionally busting out a 15+ yard one. Sanders's 5 YPC for his career hides a distribution that I think was somewhat less than optimal for an RB.
That doesn't require high level stats knowledge. It just requires thinking about what sort of distribution you'd expect and want from YPC.
OK, I'm willing to admit my sentence was poorly constructed. I did not mean to imply that removing outliers is universally poor procedure. But since you're going to be an ass about it and accuse me of being an ignoramus, it's time for me pull my argument from authority card.
Confession time: I didn't take Stats 101 at Michigan. I took Stats 620, 621, and 625 at Michigan. Is that enough for you? Google my user name, then find and Google my real name. Look up the name of the workshop I helped run last month. Look up the numerous papers I have written on stochastic processes. Then keep on telling me I know nothing about statistics.
"If you run 2, 3, 3, 2, 5, 3, 56, 3, 3, etc...then 56 is an outlier that skews the distribution." Sure it skews the distribution, but that doesn't necessarily mean it's not important. Believe it or not, sometimes distributions are naturally skewed!!! That 56-yard run is the difference between Barry Sanders and Kevin Jones and that is really damn important when fairly evaluating running backs. You do not remove outliers blindly assuming a Gaussian distribution, which is what you do if you have taken Stats 101 and never actually considered the deeper reasons as to why you perform statistical procedures and when you should perform them.
Removing the longest run in order to make your statistical comparison is massaging your data and is the football equivalent of a baseball announcer saying "Brandon Inge is 3 for 11 against southpaws on day games on Thursdays." It's cherry-picking. The fact that it's OK in Stats 101 doesn't mean it applies here.
Stats 101 is where you assume everything is normally distributed and thus learn a lot of things that are wrong wrong wrong wrong wrong when applied to any other statisical distribution and almost every real world application. I hope you weren't insinuating I don't have a familiarity with Stats 101.
Removing only the big runs is deliberately distorting in this case because the distribution of yards per carry is not Gaussian and probably better estimated by an exponential or a power law. Big runs are not outliers caused by measurement noise: the measurement noise can be at most +/- 1 yard! They are predictable events that are part of a running back's entire performance. Taleb would call them the black swans.
If you're going to massage your data in order to prove your point, you need a better reason than "That's what they do in Stats 101." Smith's numbers are still not good compared to Baker's without any need to resorting to statistical trickeration. Robinson's numbers weren't up to his usual standards, but the math used to get this meaningless YPC stat is nonsense.
You can't pull out outliers on only one side, especially when the distributions are heavily skewed to have a lot of small runs and an occasional large run. Pulling out the long runs is cherrypicking and a meaningless way to analyze the statistics.
Also, you didn't remove Denard's sacks when calculating his rushing yardage. Without those he's at 5.4 YPC. You also have to remove Gardner's sacks when analyzing the team rushing performance.
Recent Comments
The Baffler's been upgrading their online presence from "none" to "some" over the past couple of months and posting a lot of stuff from their print archives. The Northwestern decision made Hinton's old article timely again.
We reviewed "Billionaire Ball" at HSR when it came out in print in 2012:
http://hooverstreetrag.blogspot.com/2012/11/essay-review-matt-hinton-bi…
Is there data to look at "Fire Borges" to "Fire Funk" ratio? I noticed a significant uptick in the latter sentiment recently.
I added to the description that Beutjer later transferred to Illinois. Also, it should accept "Hawaii" without the 'okina.
I'd like to see how strong the correlation is between bursts of profanity and changes in the win probability graph. The fanbase with the strongest correlation could be said to be the most intelligently profane in sports.
This performance against Ohio or in the Rose Bowl would earn a 10 on the Borges-O-Meter. It only goes up to 9 for regional rivals.
As long as Vin Scully is still working, Shulman's not the best play-by-play guy in sports, but he may be the best at working together with analysts. I brought up Buck Martinez because he was way better with Shulman than he's ever been working with anyone else.
Shulman's probably the best option for working with Vitale, but I think he deserves a new partner so the debate can be "Shulman: best play-by-play guy?" instead of "Vitale: tolerable or insufferable?"
I'm not a fan of Vitale because Dan Shulman has been one of the best play-by-play men in the business going back to his time calling Blue Jays games with Buck Martinez. But it's been ages since I've been able to enjoy him do his thing, because he's paired with someone who has absolutely no ability to take it down a notch.
Tlon, Uqbar, Orbus Tertius (Level 10) is at the top of the scale. It's for "otherworldly" awesome moments such as your favorite moment from the fourth quarter of Notre Dame UTL.
Gospel According to Mark is at the bottom of the scale for complete and utter disasters. Example: @Nebraska after Denard went down.
The order of the stories in the Borges-O-Meter is not correlated with my opinion of their quality as stories, The Library of Babel is much better than Level 3 as a story, but that's where it fits in best on the B-O-M.
I am glad mgoblog is still a place where dumbass mistakes are quickly corrected, even if they are my own. (However, I failed in correcting usage, not grammar.)
Damn, I screwed up my idioms. I am 100% in error.
But wasn't calling me stupid more fun than rehashing RR?
Is correcting a stranger's word choice on the Internet a better use of time than arguing over RR again? You make the call!
I suggest the "Mississauga rattler."
http://en.wikipedia.org/wiki/Massassauga_rattlesnake
I'll cut the other guys some slack for blaming the Nebraska loss on the refs in the heat of the moment. Not the best attitude, but understandable. But blaming four losses on the refs? That's worthy of special disdain.
I had to update it twice today. Once for the game, then again for the postgame tweeting.
A column about the "junk wins" from CFB's early days could have been interesting, but Forde was too busy thinking his lame jokes were the wittiest japes of the season. "Something called Physicians & Surgeons of Illinois" is now known as the University of Illinois at Chicago. Case Institute of Technology (now part of Case Western Reserve) was a prestigious school - it was where the Michelson-Morley experiment was performed and, as Yeoman noted, a solid football program.
Worst of all to me, Forde abbreviates the California Institute of Technology as "Cal Tech." It is Caltech and always has been. He should get an editor and maybe do some research next time.
I'm planning on doing the opposite. I don't plan on cheering for them, but I'm going to call them "Ohio State" until Sunday.
To be fair, isn't part of the reason he changed from Oklahoma to Indiana because Kevin Wilson went from being Oklahoma's OC to Indiana's HC? Sounds like a reasonable enough reason for that first switch, at least.
George Washington and Eastern Illinois. I guess if you really don't like Tony Romo that'll help you get your hate on. Regardless of the opponents, go Connie!
http://www.jeopardy.com/minisites/collegechamps-s28/matchups/
The lesson here is that if you don't do your stats homework on MGoBlog, five people are going to jump in and do it for you.
I did a Wilcoxon signed-rank test on Michigan's 8 Big Ten games to estimate the probability that the Kenpom predictions differ significantly from the final scores of the games. The set of differences between the Kenpom predictions and actual final scores is {-7,-1,-10,-25,+3,+20,-5,-6}. Based on those differences, I get a critical value of S=9, which corresponds to a probability of about 20% that the median score margin predicted by Kenpom is not the same as the actual median score margin. This is not statistically significant, so I won't reject my null hypothesis that Kenpom is doing an OK job.
Based on most of the numbers in that set of differences being negative, it wouldn't be unreasonable to wonder if Kenpom is underrating Michigan with respect to the rest of Big Ten, but there's not enough data to prove of disprove it.
The Oklahoma band must not think Boomer Sooner is catchy, seeing as they play it every damn time Oklahoma makes a good play. Do they think we'll forget it if we don't hear it 100 times a game?
If Dantonio didn't want to listen to what Hecklinski had to say, he could have taken out his phone and played some Angry Birds. If he wanted to be a dick, he could have "forgotten" to turn the sound off before he started playing.
If he'd done that, at least he'd have been an amusing dick.
I'm glad this is Justin Taylor's response to the Napkin.
Also, I'm posting links to copyrighted material. Take that SOPA!
No, kind sir, I am not a woman not was I outside the Sheraton in New Orleans. I do however appreciate you making mgoblog a community where we do not reflexively assume everyone is male.
If you look at our blog, you can also find designs for Martin, Van Bergen, and Heininger.
While it is a random toss, I would hope that 'Tree would be especially skilled in catching the T-shirts.
I don't think you can have a completely "fair" system when 120+ teams play 12-14 games each, but I feel a necessary requirement for whatever system is used is that it not have obviously unfair elements. A necessary requirement for the system not being "unfair" is that it not depend on what half-informed coaches or Harris poll voters had for breakfast the morning they sent their votes in.
I know it's gauche to link to your own blog post, so I apologize for doing so, but I did write something last month about what I think would be a fairer way to rank teams that would be too long to repost here: http://hooverstreetrag.blogspot.com/2011/12/everyone-else-has-solution-…
Once we rank teams fairly, we can then argue over whether it's least unfair to have 2, 4, 6, 8, 12, or 16 teams in the "playoffs."
I'd rather the system be fair than that it be stacked in our favor. It's unsportsmanlike otherwise.
On a cruder note, I want a system where there's never even the tiniest sliver of truth when Sparty complains the system shafted them because they're Sparty. A fair system forces them to own their failures and stop making excuses.
None of them actually matter now that the season's over.
Go find the stuff he wrote for the Boston Phoenix before ESPN hired him. He was pretty good! He's more or less a hipster sportswriter. He stopped being cool when he got big.
Weasels and wolverines are both members of the family Mustilidae. That's as close as Brady Hoke gets to being a weasel.
As for GGERG Easterbrook, his ability to pass judgment on situations where he has no understanding is unsurpassed. In academia, it's perfectly acceptable for a professor to move from one school to another if the new school offers them better facitilies, resources, or reputation. If we're going to claim that football players are "student-athletes," then football coaches are "professor-athletes," and the ethics of them changing jobs should be the same as those for the professors. Moving on up in the academic world is a normal and completely ethical thing to do - the same should be true in the college football world. The difference between professors and coaches is that students can follow a professor to a new school and players can't follow a coach, but I think the issue there is the NCAA's lack of ethics, not the coach's.
Even someone like Todd Graham isn't a weasel for changing jobs frequently. His inappropriate conduct was not addressing his team face-to-face like an adult. The only coaches I can think of who may have acted unethically in taking college coaching jobs are Saban and Petrino, because they left NFL jobs, where the collegiate standards of behavior may not apply.
If you'd gotten your $75 to go the Big Ten Championship game, you'd have more than half the money you need to not attend the Sugar Bowl!
stiffarmtrophy.com is the go-to place for Heisman predictions, and they say RG3 with 70% of the max points. Luck #2, Richardson #3.
It's actually a compliment to VT to call them Tech as if no explanation was needed. The school that's earned the right to be called just Tech is Georgia Tech (more tradition, better research and academics - not that VT is terrible in any of these things.)
I've been told that they hate to be referred to as "VaTech" though.
I was going to put "only on a Michigan blog" as a tag on the post, but I decided to put in a Jonathan Lethem reference instead. Now that is something that would only happen on a Michigan blog.
Correcting an error: I wrote before the 2010 game, not the 2009 one. Still, we had only lost to them twice in a row, and I thought had a good chance of winning that one. :(
I have been waiting two loooong years to update this quiz, which I originally wrote in the leadup to the 2009 MSU game. At that time MSU had beaten U-M precisely once in a row. To give MSU their props, I am impressed that it did take them almost three full seasons between blowing leads in the 4th quarter.
If you don't like gloating over Sparty until we beat them next fall, play this quiz instead: http://www.sporcle.com/games/schnoxl/michigan_opponents and name every college Michigan has ever played in football.
Also read my stuff at the Hoover Street Rag: http://hooverstreetrag.blogspot.com/2011/12/i-wish-there-was-some-word-…
I really want Michigan vs. Baylor. Denard Robinson vs. Robert Griffin III. I hope this matchup isn't too awesome to exist.
Hey man, just post the link to Sporcle and let people play the quiz and appreciate the effort I went into updating it seconds after the game ended!
Anybody else notice the Mandelbrot set at around 23 seconds?
edit: Apparently I should read the other comments, first...
The best part is the puck flying through the Mandelbröt set at 0:23.
Peter King is an alumnus of the actual Ohio University and the reason why I don't care if they find it annoying that we call the school in Columbus "Ohio."
Sorry if I sounded patronizing. I think that removing the outliers requires a good justification and I just don't see it in this case. I think the numbers look pretty bad for Michigan even with the long runs included. Sacks don't count in my analysis because they're passing plays not running plays.
If I were to pull outliers, I'd pull the highest and lowest designed run for each player, just to be symmetric. In this case, I'd just say, looking at all the data says Vincent Smith had a bad day, and pulling out the high & low runs says he had a really bad day.
I got a little carried away. I was in full on XKCD "Something is wrong on the Internet" mode.
They are different sure, but that's why you shouldn't assume taking the mean explains everything. You don't need to remove the outlier to understand why you'd rather have the second guy. My points are:
1) Don't remove outliers if they don't diminish your point. You don't need to remove outliers to show that Edwin Baker had a better day than Vincent Smith; and
2) Good procedure (which is different from commonly used procedure) is that you don't assume inconvenient data is an outlier. If it's an outlier, you explain why it doesn't belong. For example, in a data set I work with, we remove one subject as an outlier because the experimenter reported that he/she didn't obey the conditions under which the study was being conducted (he/she took some drug that wasn't allowed, I think). In that case, I've explain why it's not part of the distribution. That 91-yard run is just as much part of the distribution as all the others, as football is a game of a lot of short plays and the occasional long play. I used Barry Sanders as my example because that's how I remember him: lots of runs of 0,1, or 2 yards, and then occassionally busting out a 15+ yard one. Sanders's 5 YPC for his career hides a distribution that I think was somewhat less than optimal for an RB.
That doesn't require high level stats knowledge. It just requires thinking about what sort of distribution you'd expect and want from YPC.
OK, I'm willing to admit my sentence was poorly constructed. I did not mean to imply that removing outliers is universally poor procedure. But since you're going to be an ass about it and accuse me of being an ignoramus, it's time for me pull my argument from authority card.
Confession time: I didn't take Stats 101 at Michigan. I took Stats 620, 621, and 625 at Michigan. Is that enough for you? Google my user name, then find and Google my real name. Look up the name of the workshop I helped run last month. Look up the numerous papers I have written on stochastic processes. Then keep on telling me I know nothing about statistics.
"If you run 2, 3, 3, 2, 5, 3, 56, 3, 3, etc...then 56 is an outlier that skews the distribution." Sure it skews the distribution, but that doesn't necessarily mean it's not important. Believe it or not, sometimes distributions are naturally skewed!!! That 56-yard run is the difference between Barry Sanders and Kevin Jones and that is really damn important when fairly evaluating running backs. You do not remove outliers blindly assuming a Gaussian distribution, which is what you do if you have taken Stats 101 and never actually considered the deeper reasons as to why you perform statistical procedures and when you should perform them.
Removing the longest run in order to make your statistical comparison is massaging your data and is the football equivalent of a baseball announcer saying "Brandon Inge is 3 for 11 against southpaws on day games on Thursdays." It's cherry-picking. The fact that it's OK in Stats 101 doesn't mean it applies here.
Sorry, double post.
Stats 101 is where you assume everything is normally distributed and thus learn a lot of things that are wrong wrong wrong wrong wrong when applied to any other statisical distribution and almost every real world application. I hope you weren't insinuating I don't have a familiarity with Stats 101.
Removing only the big runs is deliberately distorting in this case because the distribution of yards per carry is not Gaussian and probably better estimated by an exponential or a power law. Big runs are not outliers caused by measurement noise: the measurement noise can be at most +/- 1 yard! They are predictable events that are part of a running back's entire performance. Taleb would call them the black swans.
If you're going to massage your data in order to prove your point, you need a better reason than "That's what they do in Stats 101." Smith's numbers are still not good compared to Baker's without any need to resorting to statistical trickeration. Robinson's numbers weren't up to his usual standards, but the math used to get this meaningless YPC stat is nonsense.
Not until Brian stops inadvertently trolling my HSR posts by calling Community "a bad hour-long NBC drama."
You can't pull out outliers on only one side, especially when the distributions are heavily skewed to have a lot of small runs and an occasional large run. Pulling out the long runs is cherrypicking and a meaningless way to analyze the statistics.
Also, you didn't remove Denard's sacks when calculating his rushing yardage. Without those he's at 5.4 YPC. You also have to remove Gardner's sacks when analyzing the team rushing performance.
There are 12 silhouettes total.