PDA

View Full Version : Pairwise Analysts, Please Come Help Us 2013



Pages : [1] 2 3 4 5

Numbers
01-13-2013, 11:59 PM
The idea I have for this thread is not to do the usual "This is the current PWR. The field would be... The bracket would be...."

Instead, I would like this to be a place where we discuss the positives/negatives of different teams' situation.

For example, I root for Minnesota. Right now their RPI is solid. But, more than that, they have an 8-0-0 non conference record. That is a very strong situation, because of how it affects the ComOpp part of any comparison. For this reason, their PWR standing is really STRONGER than it appears.

That's the kind of analysis that I think we need. As the end of the season draws nearer, whose TUC record is better, and how stable their TUC opponents are (compared to the TUC cliff) becomes important. Who might lose a compare to who because of ComOpp or head-to-head becomes important.

Is it worth it to have a thread for that analysis? This is what I am hoping for.

SCSU Euro
01-14-2013, 01:22 AM
The idea I have for this thread is not to do the usual "This is the current PWR. The field would be... The bracket would be...."

Instead, I would like this to be a place where we discuss the positives/negatives of different teams' situation.

For example, I root for Minnesota. Right now their RPI is solid. But, more than that, they have an 8-0-0 non conference record. That is a very strong situation, because of how it affects the ComOpp part of any comparison. For this reason, their PWR standing is really STRONGER than it appears.

That's the kind of analysis that I think we need. As the end of the season draws nearer, whose TUC record is better, and how stable their TUC opponents are (compared to the TUC cliff) becomes important. Who might lose a compare to who because of ComOpp or head-to-head becomes important.

Is it worth it to have a thread for that analysis? This is what I am hoping for.

Their PWR isn't stronger than it appears.... it is exactly as strong as it appears. For someone with your screenname, you should have no problem quickly learning the phrase "Simple Math."

SanTropez
01-14-2013, 09:02 AM
Good thread idea, and yes Numbers is correct it is stronger than it appears.

8-0 non conference will do that.

UncleRay
01-14-2013, 09:50 AM
If Maine manages to win a few down the stretch and becomes a TUC, there will be a lot of happy teams out there since everybody has been beating us like a red-headed stepchild rented mule.

FlagDUDE08
01-14-2013, 09:53 AM
The idea I have for this thread is not to do the usual "This is the current PWR. The field would be... The bracket would be...."

Instead, I would like this to be a place where we discuss the positives/negatives of different teams' situation.

For example, I root for Minnesota. Right now their RPI is solid. But, more than that, they have an 8-0-0 non conference record. That is a very strong situation, because of how it affects the ComOpp part of any comparison. For this reason, their PWR standing is really STRONGER than it appears.

That's the kind of analysis that I think we need. As the end of the season draws nearer, whose TUC record is better, and how stable their TUC opponents are (compared to the TUC cliff) becomes important. Who might lose a compare to who because of ComOpp or head-to-head becomes important.

Is it worth it to have a thread for that analysis? This is what I am hoping for.

You should have seen the computer program that RHamilton put together. It was a thing of beauty.

Patman
01-14-2013, 10:30 AM
Look, assessing such a thing isn't an easy thing as such... One such way you might be looking at it is via sensitivity. How sensitive is the method, or the position, to changes in complexion. That may not dictate your team's true wiggle room but it might mock it a bit.

One way may be to delete games at random. Another might be to insert random losses against forthcoming opponents.

These would only be tools that are rather incomplete. I've had some ideas that would be computation and leg-work intensive (assessing probabilities via model).

There are ways and with the PWR being an impure thing as it is... I don't think coming up with a single STRAIGHT FORWARD metric would be reasonable.

Numbers
01-14-2013, 11:26 AM
Look, assessing such a thing isn't an easy thing as such... One such way you might be looking at it is via sensitivity. How sensitive is the method, or the position, to changes in complexion. That may not dictate your team's true wiggle room but it might mock it a bit.

One way may be to delete games at random. Another might be to insert random losses against forthcoming opponents.

These would only be tools that are rather incomplete. I've had some ideas that would be computation and leg-work intensive (assessing probabilities via model).

There are ways and with the PWR being an impure thing as it is... I don't think coming up with a single STRAIGHT FORWARD metric would be reasonable.

Patman,
Thanks for commenting. You are one of the guys I hoped to see, based on past years.

My thought right now is not a quantitative sort of analysis. Reasonably, there is too much hockey left for that, because we really won't have any idea of who the real bubble is for awhile yet. Even a team like Mich State, which right now is playing with the TUC cliff, could end up in the tournament field.

What I am hoping for is some comments from a few on which teams position may be stronger or weaker than it appears because of TUC record or CommOpp records. And, I know that right now, that might not mean much.

But, 4 weeks from now, there might be more clarity. It would be nice if, at that point, we could say (hypothetically), "Well, Mankato is currently 14th in the PWR, but their Out of Conference record gives them help against HE teams, so they have more cushion than Merrimack..." Note, I don't mean to comment specifically about Mankato or Merrimack here. But, that's an example.

I like digging into the depths of the information, so I like noting Minny's record (as above), but it's a tedious work, and I don't want to look close at 20 teams. So, I thought that collectively, we might be able to come up with few thoughts that would be good to keep in mind nearer the season's end.

Ralph Baer
01-14-2013, 11:32 AM
You should have seen the computer program that RHamilton put together. It was a thing of beauty.

I suspect that he will only bring it back if RPI becomes a candidate for the Tourney -- which I am too stubborn to rule out. :D

FlagDUDE08
01-14-2013, 11:39 AM
Patman,
Thanks for commenting. You are one of the guys I hoped to see, based on past years.

My thought right now is not a quantitative sort of analysis. Reasonably, there is too much hockey left for that, because we really won't have any idea of who the real bubble is for awhile yet. Even a team like Mich State, which right now is playing with the TUC cliff, could end up in the tournament field.

What I am hoping for is some comments from a few on which teams position may be stronger or weaker than it appears because of TUC record or CommOpp records. And, I know that right now, that might not mean much.

But, 4 weeks from now, there might be more clarity. It would be nice if, at that point, we could say (hypothetically), "Well, Mankato is currently 14th in the PWR, but their Out of Conference record gives them help against HE teams, so they have more cushion than Merrimack..." Note, I don't mean to comment specifically about Mankato or Merrimack here. But, that's an example.

I like digging into the depths of the information, so I like noting Minny's record (as above), but it's a tedious work, and I don't want to look close at 20 teams. So, I thought that collectively, we might be able to come up with few thoughts that would be good to keep in mind nearer the season's end.

One place where you can start is how your conference-mates are doing in NC play. If it's a fairly strong year for your conference as a whole, then that will improve your chances. On top of your conference-mates, it's also important to look at how the teams you faced in NC play are doing. A couple of years ago, both RPI and UNO played Alabama-Huntsville. RPI swept the series, but UNO split it. Because of that split, RPI was able to gain enough of a common opponents advantage to push UNO out of the tournament, as RPI was the last at-large in. I understand that you don't want to do that for every team you play, but it will give you a general idea of how your team fares.

burgie12
01-14-2013, 12:29 PM
Sioux Sports' PWR display (http://siouxsports.com/hockey/rankings/pwr/) is an excellent tool for delving into individual matchups. (Especially since they've now corrected the Dec 8th game between Robert Morris and Ohio State.)

The one team that really jumped out to me is BU. Their 4-6-0 TUC Record will significantly hurt them down the stretch (especially the fact that they're 3-6-0 against teams in the Top 10 of the RPI, teams that are extremely unlikely to fall off of the cliff), as will their 1-2-0 record against the WCHA and their 0 games played against CCHA teams this season. Their poor performance against the WCHA renders the COp comparison lost against nearly every WCHA team right off the bat and no games against CCHA teams means that they have one less opportunity to hide their poor TUC Record against most CCHA teams.

At BU's current RPI, once the TUC Record comes into play, they will lose the Dartmouth and North Dakota comparisons. Their comparison against Wisconsin is also hanging on by just a thread (both teams have a 0.400 TUC win percentage). Further, once their TUC Record comes into play against Colgate, potentially Cornell, and Denver, BU's RPI is the only thing keeping them winning those comparisons.

The Terriers could easily drop 4 comparison wins, just by playing out the rest of the regular season, pushing them much closer to the cut line than they are currently.

BUPhD
01-14-2013, 12:40 PM
The issue here is that PWR, as I understand it, is not meant to be a predictive tool. Hence the disclaimer on all bracketology using it that, "This assumes the season ends today." There's no real consideration as to whether a team has more TUC games under its belt or more on the plate to be won or lost, etc,

So the PWR at the end of the season is 'simple math' but what it seems speculators like to do is predict where the PWR will move from week to week on given results. Calculators will do this of course, but the sheer number of variables makes any 'snapshot of this week' a poor indicator of where things might end up.

FlagDUDE08
01-14-2013, 01:04 PM
The issue here is that PWR, as I understand it, is not meant to be a predictive tool. Hence the disclaimer on all bracketology using it that, "This assumes the season ends today." There's no real consideration as to whether a team has more TUC games under its belt or more on the plate to be won or lost, etc,

So the PWR at the end of the season is 'simple math' but what it seems speculators like to do is predict where the PWR will move from week to week on given results. Calculators will do this of course, but the sheer number of variables makes any 'snapshot of this week' a poor indicator of where things might end up.

I believe what has been used by some is to use KRACH ratings to figure out the odds of teams winning games and proactively adjust the PWR to make a prediction as to who has the best odds of making it to the tournament. I wonder how accurate KRACH is in predicting the winner of games based upon various ranges of chance of winning (50%-60%, 60%-70%, etc.)?

Numbers
01-14-2013, 01:16 PM
Sioux Sports' PWR display (http://siouxsports.com/hockey/rankings/pwr/) is an excellent tool for delving into individual matchups. (Especially since they've now corrected the Dec 8th game between Robert Morris and Ohio State.)

The one team that really jumped out to me is BU. Their 4-6-0 TUC Record will significantly hurt them down the stretch (especially the fact that they're 3-6-0 against teams in the Top 10 of the RPI, teams that are extremely unlikely to fall off of the cliff), as will their 1-2-0 record against the WCHA and their 0 games played against CCHA teams this season. Their poor performance against the WCHA renders the COp comparison lost against nearly every WCHA team right off the bat and no games against CCHA teams means that they have one less opportunity to hide their poor TUC Record against most CCHA teams.

At BU's current RPI, once the TUC Record comes into play, they will lose the Dartmouth and North Dakota comparisons. Their comparison against Wisconsin is also hanging on by just a thread (both teams have a 0.400 TUC win percentage). Further, once their TUC Record comes into play against Colgate, potentially Cornell, and Denver, BU's RPI is the only thing keeping them winning those comparisons.

The Terriers could easily drop 4 comparison wins, just by playing out the rest of the regular season, pushing them much closer to the cut line than they are currently.

Thanks, Burgie. This is the sort of thing I was hoping for.

Now, as a question: Should we have some kind of Google Drive spreadsheet with infor about various schools? I could make a rudimentary one. Not with any Math, mind you. Just qualitative info like what you shared here.

Numbers
01-14-2013, 01:20 PM
The issue here is that PWR, as I understand it, is not meant to be a predictive tool. Hence the disclaimer on all bracketology using it that, "This assumes the season ends today." There's no real consideration as to whether a team has more TUC games under its belt or more on the plate to be won or lost, etc,

So the PWR at the end of the season is 'simple math' but what it seems speculators like to do is predict where the PWR will move from week to week on given results. Calculators will do this of course, but the sheer number of variables makes any 'snapshot of this week' a poor indicator of where things might end up.

Thanks, BUPhD.

You are right, in one sense. Like some posters remind us every year, "Really, the PWR only exists on one date - Selection Sunday morning."

However, we do know the pieces of the comparisons that constitute the PWR. And, many sites publish a 'current PWR.' And, we are interested in "Who will make the field." So, what we are trying to do here is say "Well, BU's OOC results are going to be a drag for them. Afterall, they could easily lose both the TUC and COp parts of many compares. So, to stay in the 2 seed bad, (or whatever), they will need very good results the rest of the season - better than they have had."

We are not trying to predict every game. Rather, a qualitative feel for how safe the different schools' current Compasison Wins really are, for example.

LTsatch
01-14-2013, 01:32 PM
I think I will stick to the Priceless thread, because you guys are going way over my head. If you want to do a Yale comparison for chits and giggles, have at it, they are an interesting case so far.

Patman
01-14-2013, 01:38 PM
I believe what has been used by some is to use KRACH ratings to figure out the odds of teams winning games and proactively adjust the PWR to make a prediction as to who has the best odds of making it to the tournament. I wonder how accurate KRACH is in predicting the winner of games based upon various ranges of chance of winning (50%-60%, 60%-70%, etc.)?

Personally I take more stock in score predictive models... I think once we get past the Monday tournament Sioux sports does sims from remaing regular season games.

If I had my way I'd have a giant simulator available earlier in the season a la baseball prospectus... But schedules are fluid because of tournaments and tie-breakers are hard for this "as I need to" programmer.

Edit: you know, I've talked about this so long, I ought to just write up a document giving what i want to do and then gauge ability and interest in pulling it off.

Numbers
01-14-2013, 01:45 PM
Patman,

I did a quick look at Lowell just now. You might know more details. What I found was that their TUC record could hurt them. They currently lead the compares with both UNO and SCSU because of having a higher RPI. But, in both cases, they lose both COp and TUC once TUC comes into play.

So, I would say, their position, if they stay sort of 'on the bubble' in the RPI ranks, is a bit tenuous for the tournament field.

Do you concur?

Patman
01-14-2013, 01:53 PM
Patman,

I did a quick look at Lowell just now. You might know more details. What I found was that their TUC record could hurt them. They currently lead the compares with both UNO and SCSU because of having a higher RPI. But, in both cases, they lose both COp and TUC once TUC comes into play.

So, I would say, their position, if they stay sort of 'on the bubble' in the RPI ranks, is a bit tenuous for the tournament field.

Do you concur?

Without looking closely... Lowell did well in OOC but only split against the WCHA teams so that will leave an impact. Their TUC and that will hold Lowell down as the WCHA tends to have a fair number of teams in the pair wise. The only thing holding us up at the moment is strength of schedule.

I've noticed that in pair wise if you are to choose your losses then take them against crummy opposition. Of course, that isn't controllable.

Lowell has a lot of work to do and the TUC harm is nearly undoable due to their performances against BU, BC, and UNH... I'll leave out the DU game as parts of the picture were there... It just didn't come together that night.

---

I really ought to get to work... Don't be surprised if a PDF shows in my signature in line w the grander idea that I always seem to talk about.

UMLFan
01-14-2013, 02:14 PM
Lowell has a lot of work to do and the TUC harm is nearly undoable due to their performances against BU, BC, and UNHLowell hasn't played BU yet.

burgie12
01-14-2013, 03:44 PM
I think I will stick to the Priceless thread, because you guys are going way over my head. If you want to do a Yale comparison for chits and giggles, have at it, they are an interesting case so far.
Yale's heavily insular schedule does make them a very interesting case.

Add in the fact that 8 of Yale's 13 remaining regular season games are against current TUCs (and three more against Princeton and Rensselaer, who are each within 0.015 RPI of being a TUC) and it's practically impossible to make any definitive claims about them.

The best advice for Yale going forward is to continue winning the big games. The losses to RPI and Holy Cross may be maddening as a fan, but they are great for the team as explained a couple of seasons ago by Scott Brown (http://www.uscho.com/2009/02/24/big-wins-and-bad-losses/). The main point, though is that only 5 of Yale's 16 games so far have been played against TUCs and 8 of their remaining 13 will be. So, really, we just don't know enough about Yale's ability to make any statements on how strong or weak they are compared to their current Pairwise placement.