Preseason Analysis and Probabilities

MtownJacket said:
Physics
Click to expand...

that explains it.

Sorry to criticize, but I TA CS1372 (C for EE/CmpE)... and if somebody asked me why their code didn't work and I saw 14 nested FOR loops when I tried to help... I would tell them they were on their own until they fixed that.

On another note, I think it's really hard to come up with probabilities like this at this point or at any point during the season for that matter. The best chance for predicting the outcome of any single game is probably immediately before that game. Momentum from past games and injuries are sure to affect your chances in following games.

For instance: With a healthy roster, I would give GT a .55 chance of beating VT... if we are missing any key players for that game though, I could see us closer to .4...
 
MtownJacket said:
I see what you mean by home field advantage but its not just adding 5% to the home team its taking 5% away from the visiting team so its a 10% swing. I actually thought 5% might have been a little high.
Click to expand...

Right, the terminology is confusing but we're on the same page.

I'm saying that the 5% extra, or 10% swing is FAR too low, rather than too high as you suspected.

In major college football it's more like a 25% to 30% swing, with the home team winning 62-65% of the time. I was surprised the first time I heard this, too (I think I first heard it from a sports betting expert being interviewed on the radio).

A typical major college home field advantage is 4-4.5 points. GT's is rated at 3.75 by Phil Steele, and we're not one of the tougher places to play.

Most football betting is done on point spreads, but you can also bet the money line. If the point spread is 4 to 4.5, the underdog will pay 2:1 and the favorite will pay 1:2, not counting the vigorish.

I averaged a couple of money line to point spread conversion charts from sports books to get the above table, which I think is really useful for this kind of analysis because most of us are used to thinking in terms of point spreads.

This is why I feel VERY good about GT's schedule this year. IMO we get five of our seven toughest opponents at home, and we play Duke, MSU and Vandy (all of whom IMO we should beat anywhere) on the road. It's true we're snakebit at UVA, but I don't think that actually matters as much under this coach as fans think it does -- and we've got a LOT better team so we should win there.

Even though we'll have a better team in 2010 than this year, I don't think we'll win more games and would say we might win fewer due to playing probably 6 of our 8 best opponents on the road.

Thanks for the Python code! :biggthumpup:
 
GTjojo said:
that explains it.

Sorry to criticize, but I TA CS1372 (C for EE/CmpE)... and if somebody asked me why their code didn't work and I saw 14 nested FOR loops when I tried to help... I would tell them they were on their own until they fixed that.

On another note, I think it's really hard to come up with probabilities like this at this point or at any point during the season for that matter. The best chance for predicting the outcome of any single game is probably immediately before that game. Momentum from past games and injuries are sure to affect your chances in following games.

For instance: With a healthy roster, I would give GT a .55 chance of beating VT... if we are missing any key players for that game though, I could see us closer to .4...
Click to expand...

I know it looks sloppy to a cs major but there are 2^14 or 16384 different possible outcomes. I couldn't think of a different way to loop through 14 games without 14 different events. If you can do better then do better and post it, otherwise you can read the comments at the beginning of the code and follow them.
 
GoldenTornado said:
This is why I feel VERY good about GT's schedule this year. IMO we get five of our seven toughest opponents at home, and we play Duke, MSU and Vandy (all of whom IMO we should beat anywhere) on the road.
Click to expand...

Hmmm, thought I'd clarify after reading this.

The point is that it makes a big difference to play your tossup games at home, but it doesn't make much difference where the game is played when one team is heavily favored.

Take VT. This year we play VT at home, next year we play Duke at home.

If we're evenly matched against VT, then we would be a 5-6 point dog on the road and a 4-point favorite at home (according to Phil Steele). That's a swing of about 35% -- from 30% likely to win on the road to 65% likely at home. It just makes a huge difference.

On the other hand, say we're 7 points better than Wake Forest. That would make us a 4-point favorite on the road and an 11-point favorite at home. Here the swing is only about 16% -- from 65% on the road to 81% at home.

And when you're talking about a team like Duke, this is even more true. Then you're talking about something like a 5% swing -- say 90% on the road vs. 95% at home.

So from a probability analysis point of view, it's much better to get your most even matchups at home, and your much better or worse opponents on the road. Since we won't have any much better opponents, we want to get the Dukes and Vandys on the road as much as possible so they aren't taking up valuable home-game slots! :biggthumpup:
 
I plugged my numbers in and got fairly similar results to your "optimistic" table....

8 9.5%
9 18%
10 25%
11 23%
12 14%
13 5%
14 < 1%

Obviously games 13 and 14 depend on the first 12. Worse than 5-3 with a win over VT, or 6-2 with a loss to them, and we probably only play 13.

On the other hand, if we were somehow to win the first 13 then our chances are much lower than 65% to win the 14th -- b/c it will be against Florida!
 
MtownJacket said:
Interesting point. It makes you wonder how we pick up phrases that people use all the time but make no sense.
Click to expand...

The phrase is "I couldn't care less." You just sound like moron for saying it wrong.

Minus well leave stingtalk now.
 
Dude, you came out of lurking to post this? During your time at GT did you ever paint your chest yellow and your face black, and wear a yellow wig?

Just curious.
 
MtownJacket said:
I thought .3 was pretty fair. Its the same thing as when we play wake forest we are just in the other team's shoes. I would say that florida playing us would be very similiar to us playing wake.
Click to expand...

It's the probability of winning the BCS championship ONCE WE ARE ALREADY IN THE GAME. It is an entirely separate event GETTING to the point where we have qualified for the game.

Scanning the week by week point spreads from Golden Tornado I would advise not using this as a betting guide. To suggest that we will be favored in all but the FSU game is not realistic.
 
BarrelORum said:
Dude, you came out of lurking to post this? During your time at GT did you ever paint your chest yellow and your face black, and wear a yellow wig?

Just curious.
Click to expand...

PIMPLMAOF!

Best thread of the off season. Mods?
 
MtownJacket said:
I saw attempts at looking at probabilities to break down GTs football season on various boards but I never saw one that actually computed the probabilities of a certain amount of wins given the probability of each game so I decided to do it myself:

http://www.prism.gatech.edu/~tosburn3/gt2009optimistic.xls

http://www.prism.gatech.edu/~tosburn3/gt2009realistic.xls

http://www.prism.gatech.edu/~tosburn3/gt.txt

The last file is code in Python which I used to compute the probabilities. You can download Python 3.0 for free and run the code with your own probabilities to see what you get.
Click to expand...

Convert to a google doc and host it someplace?
 
As a former TA for a programming class, without even running the code to check for correctness I'd give you no higher than a C.
 
Yukonwreck said:
Scanning the week by week point spreads from Golden Tornado I would advise not using this as a betting guide. To suggest that we will be favored in all but the FSU game is not realistic.
Click to expand...

You're entitled to your opinion, of course, and I ain't claiming to be Nostradamus. :p But I can promise you those numbers didn't come out of my ass. I really doubt anyone else has wasted as much of his life doing quantitative analysis of GT, PJ and the ACC going into this season. :ugh:

If you think GT is gonna go 9-3 or 10-2, then you can only be quibbling over tiny differences. The point spreads I posted set the regular-season over-under at only 9.0.

If you think we're only gonna win 7-8 regular season games this year.... well, I just don't agree and I'm willing to put my money where my mouth is. ;)
 
turbanicusgt said:
I'm a nerd, I like this kind of stuff. I'd take 9 or 10 wins. Though, what is up with the .3 to win the BCS championship game? I like those odds but don't quite believe them.
Click to expand...

I agree that the .3 is off. If we are able to get to the national championship in the first place, then the chance should be higher. It should really be a .5.
 
Diseqc said:
As a former TA for a programming class, without even running the code to check for correctness I'd give you no higher than a C.
Click to expand...

As long as we're handing out grades for no good reason, you get F's on today's quizzes in basic adult politeness and reading comprehension.... :laugher:

Line 3:

# ... I don't care if you can make better or more efficient version
 
TechFan4Life said:
Kyle was trying to say...If you COULD care less, that would imply that you did, infact care at least a little about the codes. Really what you should have said was I COULDN'T care less - because you don't care at all thus making caring less... impossible :)

(ps - not sticking up for Kyle :))
Click to expand...
Have your codes figured in the following facts:
1. Our offensive line is still not able to blow people off the ball
2. Our defensive line was decimated by graduation
3. Special teams play, to be charitable, is mediocre
4. Teams with issues on both lines and special teams rarely win 10 games much less championships.
5. Rednecks like good ol boys in south Georgia tend to know a little bit about winning football.
 
You must log in or register to reply here.
Back
Top