|06/09/2018 - 2:14pm||Collapsible threads are nice…||
Collapsible threads are nice, but even once you've collapsed them, they still take up a ton of space.
|06/09/2018 - 2:10pm||re: the banner
re: the banner
I'm pretty sure blind trekkie is Xavier Simpson and Tron is Quinn Hughes. I personally love the banner.
|05/25/2018 - 9:56am||just curious:||
why are the metrics meaningless or misleading?
It's been fun to watch, keep the updates coming.
|05/15/2018 - 4:12pm||To be clear: I was not||
To be clear: I was not implying Lebron is clean. I was only pointing out that mgrowold was not making strong arguments.
|05/15/2018 - 3:59pm||isn't rookie Lebron an 18||
isn't rookie Lebron an 18 year old kid?
I don't generally disagree with the premise that monitoring is behind doping, and that it may be rampant throughout all professional sports, but comparing 1990 skinny dudes to 2018 buff dudes and 18 year old boy Lebron to 35 year old (I'm a man!) Lebron are not particularly strong arguments.
35 year old me is 6'2" 210 lbs, and 18 year old me was 5'11" 140 lbs. If I had been lifting weights and playing hoops for that entire 17 years, I'd probably be a "good" 6'2" 210, instead of a "bad" 6'2" 210.
|05/14/2018 - 10:42am||Does this have any implications for online gambling?||
In particular, online poker?
|05/01/2018 - 12:54pm||Do they come annotated?||
A window into the sports machinations of the OP have a value beyond...well, value.
|04/17/2018 - 1:40pm||you're on the right track, but...||
NCAA galaxy brain move will be to replace kickoff with a commercial. Touchdown -> commercial break -> special kickoff replacement commerical brought to you by Arby's -> commercial break.
|04/15/2018 - 1:27pm||Taxes?||
Dont you live in your parents basement and get paid by Brian in pizzas?
|04/11/2018 - 4:47am||Yes, he was||
Jmo had excellent feet for a big man. Watch his hedge/recover defending PnR. Also watch him in transition. He was absolutely quick and explosive.
|03/26/2018 - 10:04am||somebody took the sister jean bait.||
M-dog's gonna be pissed.
|03/17/2018 - 5:47pm||Nelson county has jobs?||
You in the alcohol industry? I live near cardinal point, we love it out here.
|03/13/2018 - 4:06pm||full version of gif with bonus WTF reaction|
|03/13/2018 - 4:00pm||honestly the best recorded chug of all time||
there is only one GOAT* on this thread, and it ain't Tom Brady.
*With regard to smashing beer better than any frat boy ever could.
|03/10/2018 - 7:08am||We are talking past each other a bit.||
I commented specifically on your assertion that "performance over the last 10 games does not correlate with tournament success."
As always, we know it is sos-adjusted margin of victory that matters. So, to restate my claim: it is obviously true that adjusted MOV in that last 10 games is predictive of tournament success.
Finally, as mentioned elsewhere, those are some terrible scatter plots.
|03/09/2018 - 10:56pm||you are apparently right about the committee||
but I would like to see some support (I doubt any exists) for your assertion that there is not correlation between a team's performance in its last 10 games and its performance in the tournament. It seems like these would obviously correlate.
Would you not expect to see a correlation between a team's performance throughout the entire regular season and its post season performance? Yes, we clearly would. So why would one think that the most recent 10 games are not important?
You are apparently right about the committee not using most recent 10 games as a metric for inclusion/exclusion, but you are obviously wrong to think that the performance in a team's most recent 10 games does not have any predictive value on tournament performance.
Now, of course, if you are arguing whether the most recent 10 games is more important than the entire body of work, that is a different story.
|03/08/2018 - 3:27pm||OOoO,||
|01/23/2018 - 2:59pm||Yeah, I know||
The point is that by conceding the tip, you get the next possession. If that first possession is worth way less than others, it may be worth it to just give it up, right?
On second thought, that is dumb, though. The team winning the tip starts the game winning 0.8 - 0. Which is obviously a good thing for that team.
|01/23/2018 - 2:55pm||That's what I thought.||
So does a possession that is interrupted by an inbounds (after oob on defense, or kicked ball, or whatever), really drop in efficiency by 20%?
|01/23/2018 - 2:26pm||Oh. You think so? I don't||
Oh. You think so? I don't think I would have thought that. If there is a tie-up and the ball doesn't change possession, does that count as a new possession?
|01/23/2018 - 2:22pm||Post tip ppp||
Does the fact that it is so low imply it is actually a better strategy to purposefully concede the tip?
In return, you get to start the 2nd half with possession, and that is just off a normal inbound.
Or, does it simply imply that the first offensive possession of the game is rocky?
|01/22/2018 - 10:49am||Just piling on||
The implication that auto mechanics is somehow below "smart" people is ridiculous. I happen to be getting a PhD in aerospace engineering and you sure as hell better believe that I defer to my automechanic when it comes to taking care of my car.
He's a very intelligent guy, and has developed the relevant expertise. For some reason, I doubt that anyone would trust you to fix their car. Is that because you are "too smart"?
p.s. When/if I ever graduate, I hope that I will have enough time to learn to do at least some simple work on our cars. Seems like a fun hobby and a way to save a lot of money. It also seems like an intellectually challenging project to take on. It doesn't seem like easy work for stupid people.
|01/18/2018 - 11:31pm||Really?||
I don't. For all the usual "lacked heart!" nonsense I see around here, I see a tough matchup with a team that plays 5 out, has decent athleticism, and can switch all screens and Wagner can't take advantage when switched onto a small, and can't beat their 5 off the dribble. I'd rather not see Nebraska again this year.
|01/17/2018 - 2:32pm||Not until he's long retired.||
I know Beilein is not JoePa, but seriously, let's not rush to name things after people.
|01/17/2018 - 2:28pm||Didn't look anything up, but||
Didn't look anything up, but I'm fairly confident the 2013 team did not improve a ton offensively from their early season ratings. IIRC, that team started the year 15-0 or something like that and was briefly #1 in the polls.
This is not to say there is no case for this team to improve offensively, it is just to point out that the 2013 team does not likely bolster that argument.
|01/11/2018 - 10:30pm||Lost out||
We were definitely recruiting him until the end.
2014 was kind of a rough class, we missed out on KBD, Jae'sean Tate, Vincent Edwards and Traevon Blueitt.
All those guys have turned out good to very good.
We ended up with Ricky Doyle, Kam Chatman, Aubrey Dawkins, Austin Hatch, MAAR, and DJ Wilson.
|01/10/2018 - 12:59pm||Not an exaggeration||
"I don't think it's an exaggeration to say Michigan loses by ten if you replace Teske with Mark Donnal."
I'm not even sure it's an exaggeration to say, "I think Michigan wins by 1 if you replace 5 or so of Wagner's minutes for Teske."
(especially the last 6 seconds)
|01/09/2018 - 1:37pm||Wolverines boil things?||
Wolverines boil things?
|01/05/2018 - 11:16pm||Ron America||
I only want Matt Canada if Ron America has already said no.
|01/04/2018 - 3:48pm||I sort of agree with you on Crawford, but not in the same way...||
Not so much that I'm really optimistic about him, but instead that I think he's shown more than DPJ. I know DPJ has the recruiting profile and all that, but his numbers this year are terrible, and they include a fair amount of drops as well. (Caveat: all receiver's numbers are terrible, because QB play has been bad, as everyone knows.)
Anyway, here are their advanced stats. As I said, none are good. But DPJ's are *particularly* bad, and Crawford's are definitely better.
Giving up on getting the formatting right. Here is a direct link to the page: https://www.footballstudyhall.com/pages/2017-michigan-advanced-statistical-profile
|12/13/2017 - 7:20am||*whoever||
|12/12/2017 - 12:41pm||interesting...||
I didn't realize anyone survived the great WLA purge of 201x. I always felt chitownblue was actually a pretty reasonable guy who made strong arguments...
Am I placing you wrong, or were you one of them?
|12/07/2017 - 8:27pm||He's referencing a new tweet,||
He's referencing a new tweet, which is like 7 minutes old. Read the article. It's a stronger statement than we had this am.
|11/29/2017 - 5:34pm||nice work||
It would be really great if you could expand this analysis to other P5 conference teams. It would be nice to find the correlation between defensive line havoc rate, and holds drawn. There may also exist a correlation between LB havoc rate and holds drawn.
If there is, then Michigan's results would stand out as an *even greater* outlier.
|11/21/2017 - 3:35pm||This is a good post||
Thanks OP for the analysis.
|11/13/2017 - 4:55pm||I almost did. It's a longshot and there's basically no point...||
but, when TaiStreetsMyHero posted this calculation the first time four weeks ago, our probability of winning the division was like 0.00062%, or something crazy like that.
Our chances of winning the division have increased about 1500x since then, so I have decided these posts are "good luck."
If we're not mathematically eliminated next week, expect another post.
|11/08/2017 - 5:59pm||I explicitly said I am *not*||
I explicitly said I am *not* to be confused with the guy with camaro. : )
I do appreciate when people who have been around here long enough to remember coach gsimms get the reference.
|11/08/2017 - 5:56pm||fun discussion. just got off the phone with my brother||
and his critique was the same as clion.
I think, because the effect pushes in both ways as you said, that really it only serves to increase the variance of outcomes.
The mean outcome is not likely to be pushed very much. I think maybe the outlier cases are underpredicted (Rutgers, Michigan) because this variance is unaccounted for.
|11/08/2017 - 5:51pm||You're right that the events||
You're right that the events are not strictly independent. Our predictions become stronger as we collect more information and I am neglecting that here.
It makes the simulation more complicated though (I would have to simulate game scores, instead of pure W-L), and I think the improvement would be marginal.
|11/08/2017 - 4:08pm||I see. So he "ran" simulations...||
Obviously, he's seen my results and just copied them, ha. And no citation or anything!
|11/08/2017 - 4:04pm||I don't speak for anyone else here, but...||
I have actually made explicit how unlikely it is we run the table. Of course I hope it happens, but I will be happy with any W-L record as long as the following conditions are satisfied:
(1) Peters looks good enough to give me hope that we can have a solid offense overall next year.
(2) The run game looks good enough against the remaining top shelf competition to give me hope that we can have a good-to-very-good run game next year.
(3) The defense holds up well enough against the remaining top shelf competition to give me hope that we can have an elite defense next year.
If 1-3 are satisfied, then we are a contender. This year, unfortunately, most likely got ruined the second we lost a 5 TO game to Sparty (barring the miracle laid out in my OP).
|11/08/2017 - 3:51pm||By the time he was done, it was mostly correct||
But yes, I agree with you. I don't think adding in changing probabilities would significantly sway results (would probably only make outlier-type results more likely).
My only complaint was with his original analysis, which he fixed soon after, as you have noted.
In fact, his analysis has more or less been shown to be true. The events of the past 2 weekends basically were 1-in-1000 to happen, so it makes sense that our probability of coming in first has increased by about 3 orders of magnitude.
It might have been better if I had written something like, "The past two weekends were about as likely as getting heads on 10 consecutive trials, now all we need is to flip 7 more!"
|11/08/2017 - 1:37pm||Good point here||
Of all the crazy stuff that needs to happen, UM winning out is actually the furthest long shot (6.7%, as you said).
The probability that MSU drops 2 games in its last 3 is actually 36%!
The probability that PSU drops 1 game in its last 3 is actually 39%!
|11/08/2017 - 1:35pm||nice, thanks||
|11/08/2017 - 1:10pm||still||
Amazing that the results of the last two weeks have increased Michigan's chance of getting the title by about 1000x!
|11/08/2017 - 12:43pm||ah-ha||
That explains it. I was wondering why your numbers were *so* low for PSU getting upset in the last 3, but didn't realize we were using different predictive systems.
I think S&P is way more bullish on Michigan + upsets than Massey, so that definitely explains the 3x difference.
p.s. I was nervously waiting for your post because I was kind of waiting on you to make sure my math checks out. It's a bit hard to 'go inside' 50,000,000 simulations to make sre I had all the tiebreakers coded correctly.
|11/08/2017 - 12:40pm||paging the_knowledge||
Everytime he appears, we switch to a different timeline due to very irregular spatio-temporal anomolies.
Hopefully The_Knowledge has the wisdom and vision to put us in the proper universe
Where he will soar, leaving his doubters in a cloud of dust
|11/08/2017 - 11:58am||Just using current odds||
Right now they are the best guess of what the future odds will be.
But as we saw with the results of the last few weeks, "good" results have a residual effect that increases the chance of "good" results down the line.
The nice thing about writing a simulation is that it will be very easy to update with new numbers and results as they come available.
|11/02/2017 - 12:45pm||air force, not osu||
Air Force is 4-4.
He is saying, "Air Force is playing well, they are now 4-4, ergo I was correct to devote extra time to preparing for that game."
|10/30/2017 - 1:45pm||upvoted||
Thanks for the improvement on my vocabulary choices. My only excuse is that I'm not a statistician, I'm an engineer and was too lazy to double check the terms I was using for accuracy. My word choice was poor, but I do not think my criticisms were inaccurate.
I think it's possible that you never saw the original version of his post, which is what I was complaining about. You are correct that now he is only missing cases which were an order of magnitude less likely, but his original post came up with a number that was 3-4 times less than the number he is now posting, so he *was* originally off by a not insignificant amount in his analysis.
I also don't agree that creating a simulation is such a horrible choice for this type of analysis (you are, of course, correct that calling it an uncertainty analysis was poor word choice. I use these types of simulations to analyze uncertainties in the experiments that I run, so that is why that descriptor was on my mind). My code that I wrote for PSU's end of season took 5 minutes to write, and 1 minute to run (MATLAB is sloooow). Writing in win probabilities for all the other 6 remaining Big Ten East teams would probably take another 10 minutes or so. (Note: this is more time than I was willing to spend, which is why I simulated PSU's season only. It is not, however, likely to be more time than the OP spent with pencil and paper trying to cover all possible scenarios.)
The only thing that would be difficult about simulating the remnants of the B1G football season would be sorting out the various tiebreakers after the results come in, but even that can be coded in without too much difficulty. Maybe add another 15 minutes to write in those rules, and at that point I would guess that I have spent a similar amount of time to the OP as he worked on the straight-up math problem.