The Sagarin ratings

[SmellyCat]

I mentioned this in another thread, but I've been projecting out the rest of the season according to the Sagarin ratings. Not perfect, but it helps me understand who is expected to rise out of the pack. Here are the expected current numbers, as of today:

1) Purdue 18-2
2) Rutgers 14-6
3) Indiana 13-7
4) Illinois 12-8
5) Maryland 12-8
6) Iowa 12-8
7) Michigan State 11-9
8) Michigan 10-10
9) NORTHWESTERN 10-10
10) Wisconsin 9-11
11) Ohio State 7-13
12) Penn State 6-14
13) Nebraska 5-15
14) Minnesota 1-19

Note: my tie-breaker is just the higher Sagarin rating - I haven't looked at the actual Big Ten tiebreakers to determine seeding for the Big Ten tourney. Regardless of the 8-9 tiebreaker, looks like NU would get another shot at Michigan. I feel like the sixth seed is a decent spot, as NU would get a chance to beat one of the bad teams in the BTT, but obviously anywhere above that means NU is winning enough games that the BTT is an afterthought when it comes to getting an invitation to The Dance.

NU's remaining games, with Sagarin's current projections:
Purdue (lose by 4.54)
Indiana (lose by 1.49)
Iowa (win by 0.03)
at Illinois (lose by 7.18)
at Maryland (lose by 6.24)
Penn State (win by 5.15)
at Rutgers (lose by 6.73)

The Iowa game flipped from loss to win after last night, but obviously that margin is minuscule, and given how NU played against Iowa last week...well, I don't have to finish that sentence.

Let's start by winning against Purdue this weekend to really throw everything out of whack!

 

[AdamOnFirst]

For this analysis to be remotely useful you need to translate the point spreads into % chance of winning and calculate wins that way.

 

[SmellyCat]

I think they're at least "remotely useful." I'm not writing a thesis here, and I know all about Expected Value - I'm just trying to see what the rest of the season could like like if there are no upsets. Obviously there will be, and the 50-50 Iowa game is counted more or less as half a win or half a loss, but for the purposes of this simplistic exercise, as of today, it's one win for NU and one loss for Iowa. Just like the Indiana game is counted as one loss for NU and one win for Indiana, even if NU has 45% chance of winning according to the spread. NU's EV for those two games would be (.50) +(.45) let's say. Or 0.95. They'd have 1.00 wins instead. Close and crude, but it's what I've got on a Friday morning when I should be doing my day job.

 

[EagerFan]

Does Sagarin account for Rutgers losing their starting PF for the season? I’m thinking they lose more than expected from that.

 

[TheC]

I don't even want to look at projections. Projections are what made us (me) think the season was over after blowing a "must-win" game at home against Michigan last week. Instead, we did the most unlikely thing of knocking off Wisconsin and Ohio State back-to-back on the road. I have no idea what is going to happen from here on out. We may not win another game. We may win 4 or 5. I don't know so I'm just going to watch and hope.

 

[SmellyCat]

I don't even want to look at projections. Projections are what made us (me) think the season was over after blowing a "must-win" game at home against Michigan last week. Instead, we did the most unlikely thing of knocking off Wisconsin and Ohio State back-to-back on the road. I have no idea what is going to happen from here on out. We may not win another game. We may win 4 or 5. I don't know so I'm just going to watch and hope.

I know the feeling, and I go back and forth on it, but it's something that makes me feel like I have control when I clearly do not, and the games are few and far between. Obviously, just win baby!

 

[TheC]

I don't even want to look at projections. Projections are what made us (me) think the season was over after blowing a "must-win" game at home against Michigan last week. Instead, we did the most unlikely thing of knocking off Wisconsin and Ohio State back-to-back on the road. I have no idea what is going to happen from here on out. We may not win another game. We may win 4 or 5. I don't know so I'm just going to watch and hope.

I just realized I used the "H" word in this post about Northwestern basketball. Why do I do this to myself??

 

[SmellyCat]

I just realized I used the "H" word in this post about Northwestern basketball. Why do I do this to myself??

As long as we’re quoting this movie, I got a text last night after Barnhizer’s big three with a picture of “BROOKS WAS HERE” carved into the wood.

 

[ParisCat]

We have a rare scheduling quirk coming up — 3 straight at home. It would be REALLY nice to win 1 of those. We’ve been a great road team but with 3 of the last 4 in the road, all against very good teams, it would be nice to take advantage of our home court this week.

 

[PurpleFaze]

Tough remaining schedule but if they get 2 wins so finish 10-10 in conference they should make the tourney

 

[AdamOnFirst]

Does Sagarin account for Rutgers losing their starting PF for the season? I’m thinking they lose more than expected from that.

Neither KenPom or Sagarin adjust for personel like this.

 

[benoitdenizetl]

Tough remaining schedule but if they get 2 wins so finish 10-10 in conference they should make the tourney

I know some people are suggesting this, but I dunno. I really think we need 3 to have a great shot. Our strength of schedule is not good and the computers don’t love us.

 

[SmellyCat]

I know some people are suggesting this, but I dunno. I really think we need 3 to have a great shot. Our strength of schedule is not good and the computers don’t love us.

Two more wins at least keeps NU on the bubble. Three more means we don't have to sweat. Two more would be awesome, but I'd still be nervous. Even one more win (especially if it's a good one) might be enough if there aren't a lot of bid thieves, but obviously that would make us all nervous.

The weak schedule is going to hurt, but the strong Quad 1 record helps more than the weak schedule hurts.

This link is illuminating. Look how much better NU is at Q1 compared to these other bubble teams. Granted, I don't know how the website picked these eight teams, as OSU doesn't seem close to the bubble right now, but NU's Q1 record is way better than these others. And there's still five more Q1 games left, not including the BTT. Away record is strong too compared to the others. I like this website (just found it today), as you can compare resumes of up to eight teams at once.

Here's NU compared to the remaining teams on the schedule.

 

[Jonny2TheP]

We have a rare scheduling quirk coming up — 3 straight at home. It would be REALLY nice to win 1 of those. We’ve been a great road team but with 3 of the last 4 in the road, all against very good teams, it would be nice to take advantage of our home court this week.

This is probably irresponsible, but I am expecting to win the Indiana game. I know they are a good team, and it's not because we beat them in the first matchup, but I just think we match up well against them. While they are a good 3 point shooting team (38% - 2nd in the conference), they have attempted the least 3 point shots in the conference with nearly 40 less than the next lowest team. They like to play inside through TJD, which I think is a good matchup for how we like to double the post and make teams beat us from the outside. On top of that, it is a home game and Indiana (like most teams) hasn't been as good on the road.

We shall see. It's one of those weird things where I think Indiana is better than Iowa, but I feel like winning against Indiana is more likely based on how we match up.

 

[GeorgiaCat]

I know some people are suggesting this, but I dunno. I really think we need 3 to have a great shot. Our strength of schedule is not good and the computers don’t love us.

Our NET (overall) SOS is decent — 54. It is our Non-Conf SOS that is abysmal — 281.

 

[GeorgiaCat]

Two more wins at least keeps NU on the bubble. Three more means we don't have to sweat. Two more would be awesome, but I'd still be nervous. Even one more win (especially if it's a good one) might be enough if there aren't a lot of bid thieves, but obviously that would make us all nervous.

The weak schedule is going to hurt, but the strong Quad 1 record helps more than the weak schedule hurts.

This link is illuminating. Look how much better NU is at Q1 compared to these other bubble teams. Granted, I don't know how the website picked these eight teams, as OSU doesn't seem close to the bubble right now, but NU's Q1 record is way better than these others. And there's still five more Q1 games left, not including the BTT. Away record is strong too compared to the others. I like this website (just found it today), as you can compare resumes of up to eight teams at once.

Here's NU compared to the remaining teams on the schedule.

Look at NU vs Rutgers beginning with Strength of Schedule. Please tell me why Rutgers is 21 and we are 49, especially if NET is supposed to based on results not projections.

 

[SmellyCat]

Our NET (overall) SOS is decent — 54. It is our Non-Conf SOS that is abysmal — 281.

Yeah, I should've been more clear. It's the OOC SOS that's bad. Once all those Big Ten games come in, the overall SOS isn't bad, and it will get a bit better in the coming weeks with Purdue, Rutgers, Indiana, Illinois, etc. still on the schedule.

 

[PurpleWhiteBoy]

Does Sagarin account for Rutgers losing their starting PF for the season? I’m thinking they lose more than expected from that.

Rutgers will not play nearly as well without Mag.
They didn't have a deep bench to begin with.

NU, on the other hand, has essentially added Martinelli to the roster in the last 5 games.
Combine that with the experience gained by Barnhizer and Nicholson and we've actually improved compared to our early season games.

 

[Sec.112]

This link is illuminating. Look how much better NU is at Q1 compared to these other bubble teams.

Their Q1 record is also string compared to most of the 8- and 7-win B10 teams.

The good news is that so far they don't need to depend on another team. They can take care of their own business.

 

[hdhntr1]

Look at NU vs Rutgers beginning with Strength of Schedule. Please tell me why Rutgers is 21 and we are 49, especially if NET is supposed to based on results not projections.

Well they beat us on our floor for one. After that I have got nothing

 

[SmellyCat]

They’ve played Purdue and we haven’t?

 

[AdamOnFirst]

Yeah, I should've been more clear. It's the OOC SOS that's bad. Once all those Big Ten games come in, the overall SOS isn't bad, and it will get a bit better in the coming weeks with Purdue, Rutgers, Indiana, Illinois, etc. still on the schedule.

The SOS also isn’t really very material to the selection committee. What is material is stacking up a resume of good and very good wins and avoiding stacking up a resume of bad losses. NU has ZERO bad losses and wont have one this year. NU also already has a fairly meaty resume of good and very good wins which will only continue to grow as long as NU stays on the bubble (ever remaining game with the possible exception of PSU at home is a quad 1 game with a few opportunities for huge wins. The overall SOS # isn’t really a factor beyond its influence on that.

 

[FeralFelidae]

Tough remaining schedule but if they get 2 wins so finish 10-10 in conference they should make the tourney

Might need to win one in the conference tourney to keep that conference record at or above .500.

 

[AdamOnFirst]

Might need to win one in the conference tourney to keep that conference record at or above .500.

Agreed. At 10-10, that first conference tourney game starts to become pretty big. Likely win and in, lose and sweat. Puts a play-in situation more in the table too. Still PROBABLY okay though. We have to remember that to get to 10 wins, we will at WORST have to add a game like a home win vs Iowa or a road win vs Illinois to our resume.

 

[PurpleWhiteBoy]

NET has a basic flaw.
It treats these the same...
Purdue goes to Northwestern and wins by 20.
Rutgers goes to Northwestern and wins by 1.
As Oprah would say "You get a quad 1 win and You get a quad 1 win..."

The computers, on the other hand, determine that "therefore" Purdue is 19 points better than Rutgers. They apply that fundamental logic to all the games and derive rankings.

Both assumptions are incorrect to a degree, but NET is more incorrect.

 

[Jonny2TheP]

NET has a basic flaw.
It treats these the same...
Purdue goes to Northwestern and wins by 20.
Rutgers goes to Northwestern and wins by 1.
As Oprah would say "You get a quad 1 win and You get a quad 1 win..."

The computers, on the other hand, determine that "therefore" Purdue is 19 points better than Rutgers. They apply that fundamental logic to all the games and derive rankings.

Both assumptions are incorrect to a degree, but NET is more incorrect.

This is a pretty good example of why computer rankings bother me (and yes, I understand that there's not one single perfect system). Not all one-point wins are the same. For example, our 1 point win at Indiana was a pretty bad indicator of how much we controlled that game. We were up by 11 points with two minutes left and up by 9 points with 40 seconds left. We never trailed after Indiana was leading 5-4 in the first couple minutes of the game and led by double digits for about 30 minutes of game time. Oh, and Indiana hit a 50 footer at the buzzer to get it down to a one point final margin.

It is what it is, and at the end of the day, all that really should matter is that we won. But some computer rankings see that as a close one point win, when in reality we controlled the entire game and it was never a one possession game in the 2nd half until the buzzer sounded.

 

[GeorgiaCat]

NET has a basic flaw.
It treats these the same...
Purdue goes to Northwestern and wins by 20.
Rutgers goes to Northwestern and wins by 1.
As Oprah would say "You get a quad 1 win and You get a quad 1 win..."

The computers, on the other hand, determine that "therefore" Purdue is 19 points better than Rutgers. They apply that fundamental logic to all the games and derive rankings.

Both assumptions are incorrect to a degree, but NET is more incorrect.

I’m not sure you are correct. Two things…

While it is not listed, there are Quad 1 wins and Quad 1A wins. I believe we have just one Quad 1A win - Indiana. So the committee will look beneath just the number of quad 1 wins and losses.

Second, I looked up what goes into NET rankings. This is verbatim: The NET includes more components than just winning percentage. It takes into account game results, strength of schedule, game location, scoring margin, net offensive and defensive efficiency, and the quality of wins and losses.

According to this, a 20-point loss is different than a 1-point loss. I think this is why Rutgers is 28 spots higher in the NET rankings. Rutgers is better than us on both offense and defense efficiency. Of our seven losses, 4 were very large margins (17, 16, 16, 17). And most of our quality wins are single digits (except for Liberty and Illinois). Rutgers has one less double digit loss (and none greater than 13) and they beat Purdue.

 

[PurpleWhiteBoy]

I’m not sure you are correct. Two things…

While it is not listed, there are Quad 1 wins and Quad 1A wins. I believe we have just one Quad 1A win - Indiana. So the committee will look beneath just the number of quad 1 wins and losses.

Second, I looked up what goes into NET rankings. This is verbatim: The NET includes more components than just winning percentage. It takes into account game results, strength of schedule, game location, scoring margin, net offensive and defensive efficiency, and the quality of wins and losses.

According to this, a 20-point loss is different than a 1-point loss. I think this is why Rutgers is 28 spots higher in the NET rankings. Rutgers is better than us on both offense and defense efficiency. Of our seven losses, 4 were very large margins (17, 16, 16, 17). And most of our quality wins are single digits (except for Liberty and Illinois). Rutgers has one less double digit loss (and none greater than 13) and they beat Purdue.

as you wrote

"This is verbatim: The NET includes more components than just winning percentage. It takes into account game results, strength of schedule, game location, scoring margin, net offensive and defensive efficiency, and the quality of wins and losses."

They're just throwing words out there. I'm talking specifically about the whole "Quad 1, Quad 2 stuff," which is silly, in my opinion.

 

[AdamOnFirst]

as you wrote

"This is verbatim: The NET includes more components than just winning percentage. It takes into account game results, strength of schedule, game location, scoring margin, net offensive and defensive efficiency, and the quality of wins and losses."

They're just throwing words out there. I'm talking specifically about the whole "Quad 1, Quad 2 stuff," which is silly, in my opinion.

The win quad stuff is just a way to group the old committee concept of quality wins, big wins, win rate bs other tournament quality teams, bad losses, etc under a particular method. The committee has always looked at this stuff roughly like this.

 

[AdamOnFirst]

NET has a basic flaw.
It treats these the same...
Purdue goes to Northwestern and wins by 20.
Rutgers goes to Northwestern and wins by 1.
As Oprah would say "You get a quad 1 win and You get a quad 1 win..."

The computers, on the other hand, determine that "therefore" Purdue is 19 points better than Rutgers. They apply that fundamental logic to all the games and derive rankings.

Both assumptions are incorrect to a degree, but NET is more incorrect.

The “computer rankings,” as we are apparently calling them here, are just as must interested in projecting expected margins of victory as they are the actual winners because they are betting models to determine point spreads. Those point spreads can also be translated into win probabilities, of course. They aren’t necessarily intended to demonstrate RESUME, where obviously a one point win in a big game is HUGE, even if the computer sim considers it not only different from a one point loss… which in reality it mostly isn’t, statistically.

Pretty quickly, people realized the computer ranks systems predicted future results at a vastly superior rate than anybody had been able to do before, including Vegas line makers, so we look closely at those ranking systems to determine who is actually good. But they are NOT quite the same as looking at selection resumes, which is why the committees have additional stuff they look at. A committee has to mostly focus on what was actually accomplished over the course of the season rather than predicting future games.

 

[Styre]

Second, I looked up what goes into NET rankings. This is verbatim: The NET includes more components than just winning percentage. It takes into account game results, strength of schedule, game location, scoring margin, net offensive and defensive efficiency, and the quality of wins and losses.

According to this, a 20-point loss is different than a 1-point loss. I think this is why Rutgers is 28 spots higher in the NET rankings. Rutgers is better than us on both offense and defense efficiency. Of our seven losses, 4 were very large margins (17, 16, 16, 17). And most of our quality wins are single digits (except for Liberty and Illinois). Rutgers has one less double digit loss (and none greater than 13) and they beat Purdue.

This is correct. The committee used to use RPI, which does not include margin of victory or efficiency ratings and is based entirely on strength of schedule, but the NET formula is far more expansive.

 

[PurpleWhiteBoy]

This is correct. The committee used to use RPI, which does not include margin of victory or efficiency ratings and is based entirely on strength of schedule, but the NET formula is far more expansive.

It is also badly flawed and dramatically favors midpack Power 6 teams over mid-majors who win a lot of games.

They're using two step functions to evaluate wins and losses. Quad 1,2,3,4. The stats guys would bever do that, because it is so clunky.

 

[GatoLouco]

It is also badly flawed and dramatically favors midpack Power 6 teams over mid-majors who win a lot of games.

They're using two step functions to evaluate wins and losses. Quad 1,2,3,4. The stats guys would bever do that, because it is so clunky.

I find it BS that mid-major conference winners don't get an automatic bid. It just takes away more upsets. And stupid that any of them gets placed in the first four games.