# Nonsense Statistics to Make You Feel Better (but Probably Won't Help)

Submitted by FauxMo on October 19th, 2018 at 12:50 PM

Since I cannot work and occupy my mind, due to the impending game, I am trying to make myself feel better and get over these nerves. Mike Despacito's record at MSU vs. UM against the spread is incredible. It horrifies me. The guy is either good or lucky, and it really doesn't matter which (though, of course, it does).

So, I wanted to know - dispassionately - how likely we are to win this game tomorrow, throwing out his specific record ATS against us and the intense personal rivalry between the two programs. What are UM's chances of winning this game as 7 points favorites? I found a great website that has data on performance by game spread/line since 2003.

Based on this data (which I assume is accurate, because it's on the Internet), there have been 577 games where the closing line was -7 for one of the teams since 2003. In those games, the favored team won the game (regardless of score) 404 times, for a winning rate of 70.0%. In other words, a team favored by 7 points wins the game 7 of 10 times. FYI, if the line moves a half point in either direction before the game tomorrow, which it very well might, the odds of winning change very little; at either -6.5 or -7.5, the odds of winning outright drop a little to 68.5% and 67.5%, respectively.

However, while the team favored by 7 points may win 7/10 games, they cover the spread less than 50% of the time (46.0%). So don't be surprised if we win a close one tomorrow; I doubt anyone would be anyway.

(Now, to anticipate objections, I am sure someone will say, "BUT, even though we are 7 point favorites and that gives UM a 70% chance to win the game, that isn't factoring in the fact that D'Anterini has beat the spread against UM every time since 2007!!!" To which I respond... Of course the oddsmakers are taking this into account when they set the line. Based purely on records, offensive and defensive performance, and so forth, so far this season, UM should probably be a bigger favorite. And if we were, say, a 9-point favorite, which seems more accurate to me, our odds winning outright would leap to 83.7%)

So, we have a 70% chance to win tomorrow. I feel no better. But maybe you will...

maize-blue

Michigan scores on average 12 more points a game and gives up 7 less.

ScruffyTheJanitor

Reminds me of when we used to sing a song from Rent in choir and me and a couple of friends would randomly just add a soft "AAAAAAIIIIIIIDS" every now and then, like it was a separate baritone part of the song. Man, we were such jerks.

St Joe Blues

Hahaha! We sang "I'm on Top of the World" by the Carpenters one year in choir, and in the concert the basses added this:

"Something in the wind has learned my name

(Shooby doo wop, shooby doo wop)

And it's telling me that things are not the same

(Shooby doo wop, shooby doo wop)"

We were as smooth as the Isley Brothers with the shooby and the doo wop, if I do say so myself.

All the choir director could do, in front of hundreds of parents, was continue to play the piano and turn beet red in anger.

raleighwood

This made me feel good at first.....and then I realized that Michigan has been outscored 41 - 37 on the road this year.

I'm clinging to the fact the Michigan won in East Lansing two years ago as a rationale that it can (will) happen again.

IncrediblySTIFF

What happened yesterday is not an indication of what will happen today -- none of these stats mean anything if we don't win

LFG though

ijohnb

Time is a flat circle.

- Matthew Mcconaughay, 2014

- Freidrich Nietzsche, some other time

agp

I think the implication is even greater. If he's right - not just on a per day basis, but every single possible event has no relation to what came prior. Upside: nothing has any meaning and consequences do not exist. However, things like predicting football games are certainly harder since everything is now a 50/50.

agp

If I based yesterday on the belief that prior events are predictive of future outcomes, and then today realized that things that happen yesterday do not influence today based on an MGoBlog board post, what should I believe today since I can't both live like yesterday doesn't matter and also alter my today based entirely on what I thought yesterday. Feel like I'm caught in some broken existential for loop.

Don

"I don't think people understand how much they rely on prior probabilities to make decisions - big and small - in their daily lives."

I understand it perfectly, which is why I avoid clowns and mimes with near-religious fervor.

Greg McMurtry

One time I was eating at a restaurant with my sister, and a clown showed up there. Her frightened reaction to seeing it was “what the fuck is that asshole doing here!?”

Ron Utah

This is objectively false.  Even if each day were a random walk (it's not), the human brain would not allow it to play out that way because of its natural biases.

I'd recommend Thinking: Fast and Slow as a great place to start if you'd like to understand why you're wrong.

IncrediblySTIFF

Ron,

I love your football takes. You're well rounded and reasonable in everything you say. Even this.

I'm also going to read this book you suggested because I get off on books like that.

Anyway, I bought a 1 billion dollar lottery ticket today.  I am not likely to win.  But the stats on how impossible my chances are worth absolutely jack shit tomorrow if I do win.  Just like everything statistically leading into this game vs MSU isn't worth it's weight the pixels that it takes up on my screen if Michigan doesn't win

Stiff

rc15

So there is a 14% chance of some Dantonio voodoo happening... I don't know if that seems high or low.

mdurham07

So being a 7.5 point favorite is worse than a 7 point favorite?  Maybe just too small of a sample size?

FauxMo

Actually, statistically speaking, pretty large samples for both groups, though not enough to make the difference in outright win percentages from -7 to -7.5 statistically significant, I don't think.

agp

If we assume 404/577 and then back calculate to get 390/577 (assuming same sample size for -7 and -7.5 groups), the pval for 2-sample proportion is 0.41ish -- so yeah. We can't reject that they're the same. Highly powered as well.

FauxMo

Since you clearly know what you are talking about, here are the actual numbers:

-6.5, n=432, win %=68.5% (296-136)

-7.0, n=577, win %=70.0% (404-173)

-7.5, n=400, win %=67.5% (270-130)

And now that I look at the sample sizes for all three groups, and without wanting to compute actual difference tests, my guess is that if we go to a -7.5 spread, we will actually have a significantly smaller chance of winning the game outright. Let's hope big money doesn't go towards Michigan!

BlockM