Next Year's Schedule (Honestly)
Nothing on Michigan's schedule jumps out at me so I suppose that that is good, but honestly 12-0? really? Let's take an honest, semi-unbiased approach to this year....
WESTERN MICHIGAN - 75%
NOTRE DAME - 50%
EASTERN MICHIGAN - 95%
INDIANA - 80%
@ Michigan State - 55%
@ Iowa - 20%
DELAWARE STATE - 98%
PENN STATE - 15%
@ Illinois - 40%
PURDUE - 70%
@ Wisconsin - 55%
OHIO STATE - 35%
(Not playing Minnesota or Northwestern)
*This is a premature look at the year based on guesses and possibly false perceptions.
This (of semi-pure speculation) shows that a 7-5 or 8-4 season is expected while a 6-6 year is not unlikely. FWIW the odds of a undefeated regular season by these predictions are 0.025%
March 31st, 2009 at 10:40 AM ^
Toledo and Appy State would like a word with you about your percentages.
March 31st, 2009 at 10:42 AM ^
That is why I did not put 100%, I knew this would come up.
Michigan had a 98% chance against Appy St. They had a 2% game.
Probably a little lower of a % with Toledo. Still, upsets happen.
March 31st, 2009 at 11:44 AM ^
Your percentage regarding the Appalachian State game is probably...incorrect.
An easier way to figure it would be to figure out Michigan's winning percentage against Division II/FCS schools. If Michigan has won 49/50 games against those opponents, then I'd guess your percentage of 98 makes sense. I'm guessing Michigan's win percentage is not that high, though.
March 31st, 2009 at 10:59 AM ^
Don't all point estimates need confidence intervals?
March 31st, 2009 at 11:18 AM ^
You are killing noahtahl.
I think the question can be rephrased as:
Has anyone really been far even as decided to use even go want to do look more like?
In which case, the obvious answer is yes.
What??? Seriously, what the hell did you say???
I think CPS made a good point via the eloquent use of a theoretical question, drawing a simple and concise solution, similar to what one might conclude if asked, for instance, "how is babby formed?"
March 31st, 2009 at 10:44 AM ^
I agree on everything except ND. I would make it more like 55-60%. The only reason that we lost is that we spotted them 21 points with fumbles and had a redzone fumble and interception.
We are now at home. Our OLine will be better. Minor should be able to run all over them. Hopefully, we fumble less. Our secondary will hopefully be able to handle their one receiver look this year.
I see a lot of hope for this game, but it is a rival, so it can't be too high of a percentage.
March 31st, 2009 at 12:15 PM ^
Unless Micheal Floyd and Golden Tate morph into one...they will be pretty lethal as a pair.
Iowa at 20%? Lower than OSU? Green and King were the heart and soul of that team and the talent level at Iowa has plateaued the last couple years (defections, recruiting coming down to earth).
It's strange to think about, but despite their ineptitude over the past couple years, ND will probably have the best offense we will face this year. I wouldn't put a ton of stock in last year's game - that game was a in a torrential downpour, otherwise they would have thrown the ball a lot more. I really don't see how our secondary is going to be markedly better for at least another year.
As far as running all over them. I think we're going to see 8 men in the box all game long... ND has a deep and experienced secondary and ND will be, at most, Tate's 2nd college game.
But then again with their coaching, you never know.
As far as the secondary goes I'd like to think it will be better. Moving Stevie Brown to the LB/S position and having Brandon Smith available will help. Despite being young, they are talented. And, I don't think the secondary could be any worse than last year.
March 31st, 2009 at 10:46 AM ^
....isn't every game 50%? Especially considering there are no more ties in college football?
The rest are just externalities you have no way of knowing about. This is like when Herbstreit says "he has all the intangibles." Really? And how did you quantify that?
March 31st, 2009 at 11:20 AM ^
Only random number generators and apparently hockey games are 50/50.
This blog, I would argue, is all about illuminating the externalities that are quite tangible.
you will find my externality to be QUITE tangible. Giggity giggity goo.
March 31st, 2009 at 11:24 AM ^
By that logic, the odds of me getting mauled by a grizzly bear dressed up as Britney spears while singing the entire Wu-Tang album "36 Chambers" at 4 times the speed (in a Chipmunk voice) is 50/50 - because it'll either happen, or it won't. The rest are externalities. Are you SURE there's not a rap-obsessed Grizzly bear wandering the streets of Boston looking for me right now? That's just your guess at an externality.
March 31st, 2009 at 11:33 AM ^
I was hoping that he was joking...
Still hoping. 50/50 guy you are joking, right?
March 31st, 2009 at 12:14 PM ^
....I wish we could do "face recognition" when we post so we know the emotional setting from which a posting comes onto the MGOBLOGosphere.
But yes, I was joking about.
March 31st, 2009 at 12:21 PM ^
apologies.
March 31st, 2009 at 11:53 AM ^
Stay a long way from Vegas, dude.
Sure, if you're looking at the pure mathematical numbers. UM can only win or lose the game. However, these percentages are a guess based upon the likelyhood that UM will win.
March 31st, 2009 at 11:09 AM ^
Yeah I'd say its 50/50 whether we win any game this year. It's 50/50 we win. 50/50 we lose. Obviously.
March 31st, 2009 at 11:15 AM ^
Too low @ Iowa
Too low PSU at home
Too high @ Illinois
Too high @ Wisconsin
I would think any of the away games on the schedule this year would be significantly below 50%
March 31st, 2009 at 11:17 AM ^
Not every game is 50/50. Think about it this way, if UM plays Indiana 100 times (this year's teams, like a simulation), we'd expect UM to win about 90 of those times, or something close to that. So UM has about a 90% chance of winning the the one game on the schedule this year.
Least that's the way I think about it.
50% implies you'd expect each team to win 50 of those games which clearly is not the case.
March 31st, 2009 at 11:18 AM ^
Let's put your theory to work. If all games are 50/50 then I will take Florida and Oklahoma next year and you take their opponents. We'll put a $1,000 a game up. We'll see how the "externalities" help you out. I'll quantify it by the dollars in my pocket.
March 31st, 2009 at 11:18 AM ^
There are people lurking who will rip you apart, demand your Michigan Man credentials and ask you to recite the Victors because you dont give us a 100-percent chance at beating OSU.
FWIW, I dont have much confidence in the Iowa game....even in the best of times, that is always a fight to the death.
March 31st, 2009 at 11:20 AM ^
Some of the criticisms on here are just stupid.
"Toledo and Appy St would like a word with you."
teehehehehe good one.
Understand what % means before you post.
Thank you, dpb, for the much needed clarification for some people. You get an A for the day.
March 31st, 2009 at 11:29 AM ^
You, on the other hand, are noahtahls Jesus.
March 31st, 2009 at 12:15 PM ^
Say something retarded and then come back with just bluster when called on it. Why dont you explain to use what exactly you meant by :
" Toledo and Appy State would like to have a word with you".
Are you saying putting percentages is meaningless or are you saying his percentages are wrong.
If you are saying the former, the the Michigan Math department would like a word with you. I can even recommend a particular professor that you can go talk to.
If you are saying the later, then your argument is self defeating. Any model that tries to predict the future is only as good as the estimated parameters. Again, I highly recommend a chat with the Math department.
Now if you have something useful to contribute, like the guy who pointed out (correctly) that these should come with confidence intervals ( god know how we can estimate that!), then come back ...mmkay?
I like banana nut Cheerios and I find them every bit as wonderful as honey nut Cheerios.
honey nut are better than banana nut
March 31st, 2009 at 12:25 PM ^
In_Rod_I_Trust had a word with you.
March 31st, 2009 at 11:35 AM ^
I would change that 15% to 65%. They have had their senior moment. They lost all of their WRs, all of their DBs, 3/5ths of the OL, their top 3 DEs. Also a good LB. Clark can't do it all by himself.
I would also give us a better than 20% shot at Iowa, say about 40-45%.
I would lower Wisky and to 45% and MSU to 51%.
March 31st, 2009 at 11:52 AM ^
Jamie Mac wants a word with you.
March 31st, 2009 at 12:41 PM ^
If every game was 50/50 then Vegas casinos would be out a lotta cash...I'd definately take 1:1 odds on Florida over Middle Tennessee State or whomever they play in their first couple games
March 31st, 2009 at 12:49 PM ^
Is Iowa supposed to be that good? I'm genuinely curious and haven't paid any attention to why they would be that good.
March 31st, 2009 at 12:51 PM ^
SPH wants a word with ALL of you.
Rutgers' chances at winning any game? 100%. Period. Shove it, stat nerds.
It is a week after MSU and is on the road. Remember what happened to PSU last year in Iowa.
MSU, between their incessant bragging and their victory over UM last year, has just officially renewed the rivalry on UM's end and will get UM's full attention this year. Since the team is young, I think a letdown in Iowa is unavoidable.
As for UM ever having an Appy State or Toledo loss again, I don't think it is very likely. Appy State was an indication of how far the program had sunk into mediocrity under Lloyd Carr, and Toledo was because UM didn't have a real QB last year.
I am not a stats person, but I would put both games in the "statistical aberration" pile.
was an indication of how far the program had sunk, that what did the Florida game later that season indicate?
that when we opened up the playbook, we were great. When we ran the predictable BS, we were mediocre.
(Other things contributed obviously before everyone jumps all over me, but this was my main observation)
A quick and rough Monte Carlo analysis of the percentages in the first post suggest an expected win total of approximately 7.16 with a standard deviation of 1.3. So it seems you're expecting a 7-5 season.
that number. I know what Monte Carlo analysis is and it has no direct application here without further assumptions(as far as I know). Maybe you have a new technique, could you explain?
I used excel to quickly put something together since I was interested in what kind of season by wins and losses those numbers correlated to.
Knowing the likelihood of victory set forth by the percentages, I used the random number generator combined with a simple IF statement to give out a 1 or a 0, with 1 being a random number generated that is between 0 and the percentage for that game, and a 0 for a number that is above. For example, if the likelihood of a victory is 55%, and the random number generated is 34, then the victory column would have a '1' set for that game. Add up the number of '1's for the season and you have your record.
I did this for all twelve games, giving out one season. I then did a monte carlo analysis of "simulating" a number of seasons (in this rough estimate I had 50 seasons) to find what the average win total was and the standard deviation of the results.
You were extremely bored this morning, weren't you?
Yes.
What you did is simulate a few ensembles of a random distribution. What the original poster did is a far more accurate. He used the probability of victory of each game and calculated the estimated value of the final win loss of the season ( "Estimated" here Math term that has more implication than the English meaning).
If you continue your method for an infinite period of time (or practically a very long time) you will get the exact same answer the original poster got.
+eleventy billion for effort though...
But my analysis killed more time.