Torvik's/Smelly's Big Ten predictions

[SmellyCat]

I've made a spreadsheet with all of the games in the Big Ten coming up and Torvik's predictions. Since someone has to get a win and someone has to get a loss, rather than doing the statistically appropriate thing and giving a team 0.74 of a win if they have a 74% chance of winning, I just gave the win the to the team that was over 50% and the loss to the one that was under, just to get a taste of where everyone stands, how hard their schedules are, etc. (Note: When it was exactly 50%, I gave the win to the home team).

Here's how it looks as of today:
Michigan (18-2, 26-5)
UCLA (16-4, 25-6)
Illinois (15-5, 23-8)
Oregon (14-6, 25-6)
Maryland (14-6, 24-7)
Michigan State (13-7, 22-9)
Purdue (12-8, 20-11)
Iowa (12-8, 21-10)
Wisconsin (12-8, 22-9)
Northwestern (11-9, 20-11)
Nebraska (9-11, 19-12)
Penn Stata (9-11, 19-12)
Ohio State (8-12, 16-15)
Indiana (7-13, 16-15)
USC (5-15, 13-18)
Rutgers (3-17, 10-21)
Washington (1-19, 10-21)
Minnesota (1-19, 9-22)

Obviously there is a long way to go and these numbers will change as more games are played, but it was a fun little project for me.

 

[PurpleWhiteBoy]

Can you break out the NU wins and losses?

 

[SmellyCat]

Sure. Here you go...

12/3​	at Iowa​	L​	-4.4​
12/6​	Illinois​	W​	-2.0​
1/2​	at Penn St.​	L​	-3.2​
1/5​	at Purdue​	L​	-4.7​
1/12​	Michigan St.​	W​	1.8​
1/16​	Maryland​	W​	0.0​
1/19​	at Michigan​	L​	-8.8​
1/22​	Indiana​	W​	5.7​
1/26​	at Illinois​	L​	-8.5​
1/29​	Rutgers​	W​	8.1​
2/1​	Wisconsin​	W​	2.4​
2/4​	USC​	W​	6.7​
2/8​	at Washington​	W​	3.9​
2/11​	at Oregon​	L​	-7.2​
2/16​	Nebraska​	W​	2.4​
2/20​	at Ohio St.​	L​	-3.2​
2/25​	at Minnesota​	W​	5.0​
2/28​	Iowa​	W​	3.6​
3/3​	UCLA​	L​	-0.4​
3/7​	at Maryland​	L​	-7.1​

The number in the fourth column is the expected difference between the two teams. The third column is either the expected result or the actual result if the game occurred (Illinois game was in bold because the expected result and actual result differed).

You'll note the Maryland game is 0.0. I gave the advantage to NU because they're at home but it can obviously go either way, as can most of these.

 

[NJCat]

The astute reader will note NU is predicted to win every home game except UCLA, and only beat Minny and Washington on the road......crucial to maintain the home court advantage.

 

[Deeringfish]

Sure. Here you go...

12/3​	at Iowa​	L​	-4.4​
12/6​	Illinois​	W​	-2.0​
1/2​	at Penn St.​	L​	-3.2​
1/5​	at Purdue​	L​	-4.7​
1/12​	Michigan St.​	W​	1.8​
1/16​	Maryland​	W​	0.0​
1/19​	at Michigan​	L​	-8.8​
1/22​	Indiana​	W​	5.7​
1/26​	at Illinois​	L​	-8.5​
1/29​	Rutgers​	W​	8.1​
2/1​	Wisconsin​	W​	2.4​
2/4​	USC​	W​	6.7​
2/8​	at Washington​	W​	3.9​
2/11​	at Oregon​	L​	-7.2​
2/16​	Nebraska​	W​	2.4​
2/20​	at Ohio St.​	L​	-3.2​
2/25​	at Minnesota​	W​	5.0​
2/28​	Iowa​	W​	3.6​
3/3​	UCLA​	L​	-0.4​
3/7​	at Maryland​	L​	-7.1​

The number in the fourth column is the expected difference between the two teams. The third column is either the expected result or the actual result if the game occurred (Illinois game was in bold because the expected result and actual result differed).

You'll note the Maryland game is 0.0. I gave the advantage to NU because they're at home but it can obviously go either way, as can most of these.

Did you take into account extreme travel? As I recall, in football this year teams traveling coast to coast suffered for it. I should think that will be even worse in BB due to multiple games a week.

 

[olshin]

0 shot Rutgers goes 3-17.

Not saying they go 17-3 either. Not as good as people expected - still an incomplete team.

Well I shouldn't say 0 shot. Of course there is. A slow start could snowball. But significantly higher chance they figure it out and upset some teams along the way.

 

[eastbaycat99]

The astute reader will note NU is predicted to win every home game except UCLA, and only beat Minny and Washington on the road......crucial to maintain the home court advantage.

Losing at Iowa the way the game ended really hurt. Road B1G wins are a rare commodity.

 

[SmellyCat]

0 shot Rutgers goes 3-17.

Not saying they go 17-3 either. Not as good as people expected - still an incomplete team.

Well I shouldn't say 0 shot. Of course there is. A slow start could snowball. But significantly higher chance they figure it out and upset some teams along the way.

I agree. I think Indiana is probably underrated right now as well. Michigan has been really good but 18-2 seems unlikely.

I'm just going by what is predicted right now.

 

[SmellyCat]

Did you take into account extreme travel? As I recall, in football this year teams traveling coast to coast suffered for it. I should think that will be even worse in BB due to multiple games a week.

I only took it into account if Torvik did (all I'm doing is just reporting/interpreting his numbers). You're right that it will likely be a factor we're not aware of yet - definitely true for football.

 

[NUCat320]

I've made a spreadsheet with all of the games in the Big Ten coming up and Torvik's predictions. Since someone has to get a win and someone has to get a loss, rather than doing the statistically appropriate thing and giving a team 0.74 of a win if they have a 74% chance of winning, I just gave the win the to the team that was over 50% and the loss to the one that was under, just to get a taste of where everyone stands, how hard their schedules are, etc. (Note: When it was exactly 50%, I gave the win to the home team).

Here's how it looks as of today:
Michigan (18-2, 26-5)
UCLA (16-4, 25-6)
Illinois (15-5, 23-8)
Oregon (14-6, 25-6)
Maryland (14-6, 24-7)
Michigan State (13-7, 22-9)
Purdue (12-8, 20-11)
Iowa (12-8, 21-10)
Wisconsin (12-8, 22-9)
Northwestern (11-9, 20-11)
Nebraska (9-11, 19-12)
Penn Stata (9-11, 19-12)
Ohio State (8-12, 16-15)
Indiana (7-13, 16-15)
USC (5-15, 13-18)
Rutgers (3-17, 10-21)
Washington (1-19, 10-21)
Minnesota (1-19, 9-22)

Obviously there is a long way to go and these numbers will change as more games are played, but it was a fun little project for me.

The week between Christmas and New Year’s Day is truly the best week of the year.

Those pagans picked the right day for celebrating the Sun God.

Okay time to nap again.

https://www.wate.com/news/why-is-christmas-celebrated-on-december-25/

 

[PurpleWhiteBoy]

Sure. Here you go...

12/3​	at Iowa​	L​	-4.4​
12/6​	Illinois​	W​	-2.0​
1/2​	at Penn St.​	L​	-3.2​
1/5​	at Purdue​	L​	-4.7​
1/12​	Michigan St.​	W​	1.8​
1/16​	Maryland​	W​	0.0​
1/19​	at Michigan​	L​	-8.8​
1/22​	Indiana​	W​	5.7​
1/26​	at Illinois​	L​	-8.5​
1/29​	Rutgers​	W​	8.1​
2/1​	Wisconsin​	W​	2.4​
2/4​	USC​	W​	6.7​
2/8​	at Washington​	W​	3.9​
2/11​	at Oregon​	L​	-7.2​
2/16​	Nebraska​	W​	2.4​
2/20​	at Ohio St.​	L​	-3.2​
2/25​	at Minnesota​	W​	5.0​
2/28​	Iowa​	W​	3.6​
3/3​	UCLA​	L​	-0.4​
3/7​	at Maryland​	L​	-7.1​

The number in the fourth column is the expected difference between the two teams. The third column is either the expected result or the actual result if the game occurred (Illinois game was in bold because the expected result and actual result differed).

You'll note the Maryland game is 0.0. I gave the advantage to NU because they're at home but it can obviously go either way, as can most of these.

Thanks, Smelly!
I have a ratings calc that I update every now and then and the standard deviation of the actual result versus the predicted result is pretty high (about 12 points) so far this year. It used to be around 9-10 points in past years. Maybe it settles down in conference play.. but I could also see it as an indication that rosters are so fluid now that teams are more erratic as the coaches figure out who can do what and teams improve or stagnate.

Some teams are really volatile. Northwestern is not. In our last 8 games NU has played within 5 points of the expected score 6 times.
The other two games (DePaul and Northeastern) we outperformed by 12 and 10 points - and that was all in the last few minutes of the game.

 

[PurpleWhiteBoy]

I think Indiana is probably underrated right now as well. Michigan has been really good but 18-2 seems unlikely.

The 18-2 prediction for Michigan is a result of your method (adjusting for home court and awarding the better team a 100% chance of winning)

 

[SmellyCat]

The 18-2 prediction for Michigan is a result of your method (adjusting for home court and awarding the better team a 100% chance of winning)

Yeah I’ll freely admit my method is not too scientific. It’s mostly just a way to keep track of all the games so I can see what’s upcoming.