danielb927
Orange Krush Class of 2013
- Rochester, MN
Trivia question before I put up some more charts: which team do you think benefitted most (in terms of winning the B1G) by way of our upset over Purdue today?
BIG Predictions
danielb927
Orange Krush Class of 2013
- Rochester, MN
Both good guesses. Maryland sees some benefit, but Michigan St. actually made out the best over the weekend by far. Iowa, in fact, is worse off than they were on Friday. I'll post the data tomorrow sometime along with an attempt to figure out why MSU's now leading the pack.
IllinoisFan733
- Chicago
My guess is remaining schedule strength.
lstewart53x3
- Scottsdale, Arizona
Agreed. MSU is leaps and bounds better. Iowa, Maryland, and Indiana probably too.
frozenrope9190
- Aurora, IL
Still thinking 9-9 for the Illini. Think we can be 8-6 the rest of the way?
I can't imagine any team looking good against Malcolm and Kendrick last night.
ILLINOIS BABY!
sacraig
- The desert
If we reach 9-9 I will be absolutely thrilled. I think whether that is possible depends entirely on if/when Thorne and/or Black return.
My guess:
...MSU gets a bit more benefit from the tiebreaker against Illinois as our chances of winning went from very low to low. (there Illinois win looks better)
...Iowa has already beat Purdue, so they probably lose some of their tie breaker advantage in ties with PU win by Purdue dropping a game, so there PU win is less significant (+they play PU again)
...MD hasn't played IA, MSU, PU or us yet...so our game doesn't move the needle much for them. Plus MD B1G schedule has been weakest of MSU/PU/Iowa so they still have the ground to cover in B1G.
danielb927
Orange Krush Class of 2013
- Rochester, MN
Okay, here goes nothing. First, let me start with the state of the Big Ten on January 8th, the day before this weekend's games (by the way, Dan, is there a way to make these attachments a tasteful but readable size within the body of the post?)
Iowa sat on top of the predictions by virtue of their 3-0 record with wins already over MSU and @Purdue. MSU and Purdue were at 2-1 but Pomeroy ranks them as better teams than Iowa, so their odds were pretty similar. Maryland and IU were in there as well with unbeaten records.
Just to refresh memories, the first tiebreaker is head-to-head within a tied group, and after that it goes to record against other conference teams starting at the top of the standings and working down. If that fails, it goes to nonconference winning percentage, and if that fails, it goes to a coin toss. I keep loose track of this and right now it appears there's something like a 1-2% chance it goes to NC-win%, and a few-in-a-million chance of a coin toss. Wouldn't it be fun if a coin toss determined a bye or something like that?
Any rate, last Friday tiebreakers were giving Iowa about a 3% boost, which is a decent amount but not as huge as it could have been if they didn't have to play both MSU and Purdue twice this year. Purdue's the team they'd really have the edge on, given that their road win was a huge upset. Keep that in mind...
There's some other cool stuff going on - for instance, PSU's seeing the biggest benefit from the tiebreaker, and in a way that's important (first round bye vs. a bottom 4 seed). Why's that? Other suggestions are welcome, but I think it's because three of their four single-play home games are Illinois, Wisconsin, and Minnesota - all teams they'd be likely to tie. In particular, situations where PSU ties Illinois or Wisconsin are probably likely to include a PSU home win over those teams, and a corresponding tiebreaker boost.
Also worth noting is the Minnesota-Rutgers race for the bottom and the tiebreaker influence there.
Okay, so that was so three days ago. What's the situation now?
Iowa sat on top of the predictions by virtue of their 3-0 record with wins already over MSU and @Purdue. MSU and Purdue were at 2-1 but Pomeroy ranks them as better teams than Iowa, so their odds were pretty similar. Maryland and IU were in there as well with unbeaten records.
Just to refresh memories, the first tiebreaker is head-to-head within a tied group, and after that it goes to record against other conference teams starting at the top of the standings and working down. If that fails, it goes to nonconference winning percentage, and if that fails, it goes to a coin toss. I keep loose track of this and right now it appears there's something like a 1-2% chance it goes to NC-win%, and a few-in-a-million chance of a coin toss. Wouldn't it be fun if a coin toss determined a bye or something like that?
Any rate, last Friday tiebreakers were giving Iowa about a 3% boost, which is a decent amount but not as huge as it could have been if they didn't have to play both MSU and Purdue twice this year. Purdue's the team they'd really have the edge on, given that their road win was a huge upset. Keep that in mind...
There's some other cool stuff going on - for instance, PSU's seeing the biggest benefit from the tiebreaker, and in a way that's important (first round bye vs. a bottom 4 seed). Why's that? Other suggestions are welcome, but I think it's because three of their four single-play home games are Illinois, Wisconsin, and Minnesota - all teams they'd be likely to tie. In particular, situations where PSU ties Illinois or Wisconsin are probably likely to include a PSU home win over those teams, and a corresponding tiebreaker boost.
Also worth noting is the Minnesota-Rutgers race for the bottom and the tiebreaker influence there.
Okay, so that was so three days ago. What's the situation now?
danielb927
Orange Krush Class of 2013
- Rochester, MN
First, a note about the numbers. Over the weekend, Illinois pulled a big upset over Purdue, and for the most part nothing super exciting happened otherwise. Maryland and Northwestern both won as only slight favorites on the road, so the particular set of six outcomes that happened had about a 1.5% chance of happening. That means that of the 1,000,000 simulations used to produce the previous charts, about 15,000 of them were "right", and the other 985,000 were "wrong". That's as a result of six games, and there are 99 left in the season, so you can see why at this point a million simulations barely scratches the surface of total possible outcomes. In fact, I probably haven't simulated an "all favorites" case once so far.
Right off the bat, it's clear that Michigan St. "gained" the bulk of the championship probability that Purdue gave up in dropping down to 5%. Maryland and IU also picked up some chance. Weirdly, Iowa dropped a couple points. Their drop could be explained as simply as the fact that they had a bye, I'm not sure if it's more complicated than that.
Illinois gave itself a nice boost, a full seed spot up from Friday (now projected at 9 or 10, previously 10/11).
My first thought was that this might be a bug - MSU and Iowa are both projected to go 14-4 on KenPom, how is MSU 50% more likely to win? Is it all tiebreakers?
I really wanted it to be tiebreakers. Iowa and MSU are pretty likely to split their games. The next tiebreaker would be record over probably IU and/or Maryland. MSU plays those teams both once, at home, while Iowa plays Maryland once on the road and IU twice. MSU's gotta be a strong favorite to win that tiebreaker, and I think that's showing up to some degree in the red numbers for IU and Maryland getting the 1 seed. But it's certainly not giving MSU 14 percentage points over Iowa, and I'm not entirely sure what is - so much so that I'm a little worried the code might have a bug.
Other interesting notes: Minnesota has the biggest number of the year in the tiebreaker chart, a 5.5% boost to their chance at the 13 seed. That means that with tiebreakers, Minnesota is favored over the field for the 13 seed, but without them, the field is a slight favorite. Vegas lovers take note.
One final note of clarification - the tiebreaker effect charts are showing the difference in percent chance between a case with normal tiebreakers and a case where every tiebreaker is just a coin toss.
Right off the bat, it's clear that Michigan St. "gained" the bulk of the championship probability that Purdue gave up in dropping down to 5%. Maryland and IU also picked up some chance. Weirdly, Iowa dropped a couple points. Their drop could be explained as simply as the fact that they had a bye, I'm not sure if it's more complicated than that.
Illinois gave itself a nice boost, a full seed spot up from Friday (now projected at 9 or 10, previously 10/11).
My first thought was that this might be a bug - MSU and Iowa are both projected to go 14-4 on KenPom, how is MSU 50% more likely to win? Is it all tiebreakers?
I really wanted it to be tiebreakers. Iowa and MSU are pretty likely to split their games. The next tiebreaker would be record over probably IU and/or Maryland. MSU plays those teams both once, at home, while Iowa plays Maryland once on the road and IU twice. MSU's gotta be a strong favorite to win that tiebreaker, and I think that's showing up to some degree in the red numbers for IU and Maryland getting the 1 seed. But it's certainly not giving MSU 14 percentage points over Iowa, and I'm not entirely sure what is - so much so that I'm a little worried the code might have a bug.
Other interesting notes: Minnesota has the biggest number of the year in the tiebreaker chart, a 5.5% boost to their chance at the 13 seed. That means that with tiebreakers, Minnesota is favored over the field for the 13 seed, but without them, the field is a slight favorite. Vegas lovers take note.
One final note of clarification - the tiebreaker effect charts are showing the difference in percent chance between a case with normal tiebreakers and a case where every tiebreaker is just a coin toss.
Good in depth analysis.
The schedule the rest of the way is pretty favorable for th is team.
Looking at the line up Rutgers. Wisky and Minny are pretty favorable...beat the ones you should and are capable of and steal one or two....
Youngsters getting better and some minutes frrom those that can return would help immensely
The schedule the rest of the way is pretty favorable for th is team.
Looking at the line up Rutgers. Wisky and Minny are pretty favorable...beat the ones you should and are capable of and steal one or two....
Youngsters getting better and some minutes frrom those that can return would help immensely
IllinoisFan733
- Chicago
Also... there are opportunities to pad the resume. Selection Committee looks at who you beat, and I'm assuming they will give the team some leeway for those earlier losses since they were missing a All-B10 team player.
Illinois needs to take advantage of opportunities against Top 25 and Top 50 RPI teams on their schedule and steal a few. They also need to beat the teams they should beat. Not much room for error anymore.
Illinois needs to take advantage of opportunities against Top 25 and Top 50 RPI teams on their schedule and steal a few. They also need to beat the teams they should beat. Not much room for error anymore.
frozenrope9190
- Aurora, IL
Knocking off Indiana on the road would be a good second step in that direction. Nebraska first, though.
icengineer
- Southern Illinois
ray:
Yeah, do both.
Neb W
@IU L
@Min W
OSU W
Wis W
@Rut W
Iowa L
@NW L
Rut W
@Wis L
IU W
Min W
@MD L
@PSU W
Win 2 in the B1G tourney and we're:whoo:
@IU L
@Min W
OSU W
Wis W
@Rut W
Iowa L
@NW L
Rut W
@Wis L
IU W
Min W
@MD L
@PSU W
Win 2 in the B1G tourney and we're:whoo:
I don't really disagree with your specific predictions, but if we win two in the tourney, we will be 20-13. Does that really get us to the big dance?
IllinoisFan733
- Chicago
I could see the Illini picking up wins @NW and @Wisky. But there are other potential home losses marked down as W's... (Ohio State, Indiana) so it's most likely a wash. 10-8 would be a great conference record to finish with, and I think it would put us on the bubble even with only 1 B1G tourney win.
Purdue was a 9 seed last year with a 20-11 record.
IllinoisFan733
- Chicago
Total amount of wins is meaningless. Illinois could of been undefeated in the non-conference with a schedule full of cupcakes, but that won't help their resume.
It's the quantity of quality teams you do beat. Purdue was a good start. Need to string some more resume-building wins together.