2017 Pairwise thread

[LTsatch]

As stated elsewhere on this board, a pairwise thread is sort of passe' due to obvious results. That said I still like discussing this topic and miss the expert input of the members here. USCHO member Numbers posted this on another thread and I think it would be a good starting point to this years pairwise discussion. Anyone know what happened to Priceless? anyway, see below.
Quote Numbers
"No. Very odd. Usually Priceless and Patman and the other guys have it going by now. However, with RPI being so strong an influence on the PWR now, there are far fewer details to discuss. Most anyone interested can look and the PWR rankings and see all they need to see.

If you want commentary, it's available elsewhere. If you want some here....

UMD and Denver are almost assured the top 2 spots.
Harvard is almost assured the #3.

Minnesota and WMU are most likely going to end up 4/5

Lowell, BU, Union, Penn St, Cornell, and Providence are in very good position to make the tourney.

At that point there is a large gap, and there is a huge pileup which it makes no sense to try to sort out yet from:
NoDame, Wisc, OSU, SCSU, NoDak, Omaha, Air Force, and BC all with chances to make it as an At-Large.

After that, you'd better win, baby.

Conference affiliations are mushed up all over in there, too, so it's not really possible to comment on what to watch for.

There are a couple of intriguing possibilities:

1- North Dakota misses the tournament altogether
2- Big Ten teams end up 13 and 14 in the PWR, with Minnesota at 4. That would mean that Minny would play the #15 overall.
3- Associated with that, the possibility that NoDak ends at 13, which precludes Denver or UMD from playing at its nearest Region.

That's all you can really say right now."

 

[Numbers]

Re: 2017 Pairwise thread

Thanks for the quote, LTSatch. I appreciate it.

One further matter of note, is that Providence's place in the PWR is out of place with respect to their RPI because of 2 lost comparisons: With Notre Dame and with BC.

The first commentary on this is that: If Providence does not win this week against Mass, their RPI will take a huge hit, and this disparity will become moot.

Should they sweep (most likely), then the question becomes: How could they flip those 2 compares? In the case of the NoDame compare, it's simple: Beat NoDame the following week. That turns the H2H, and cancels the CommOpp piece. In the case of the BC compare, it's more complicated. Providence will have to flip the CommOpp part, because BC swept them. How could that happen? It's going to be very difficult. First, Prov would need to sweep UMass. Anything else, and this compare can't change. Then, Prov needs to sweep NoDame. That would gain them a bit. The problem is that BC will likely play either Vermont or Merrimack, and losing to either isn't going to change their CommOpp much. Without doing all the math, it seems that only a very slight chance exists for Prov to flip this compare.

That results in the following conclusion: Providence, as stated in the OP in this thread, has a strong position because of their already strong RPI.

HOWEVER, Notre Dame is not in as strong a position. Should Providence sweep UMass, and then defeat NoDame, NoDame would be in a very weak and vulnerable position.

I would say at this point that next week's supposed Prov/NoDame series is the first one that we are aware of which has make-or-break ramifications.

 

[LTsatch]

Re: 2017 Pairwise thread

Thanks for the quote, LTSatch. I appreciate it.

One further matter of note, is that Providence's place in the PWR is out of place with respect to their RPI because of 2 lost comparisons: With Notre Dame and with BC.

The first commentary on this is that: If Providence does not win this week against Mass, their RPI will take a huge hit, and this disparity will become moot.

Should they sweep (most likely), then the question becomes: How could they flip those 2 compares? In the case of the NoDame compare, it's simple: Beat NoDame the following week. That turns the H2H, and cancels the CommOpp piece. In the case of the BC compare, it's more complicated. Providence will have to flip the CommOpp part, because BC swept them. How could that happen? It's going to be very difficult. First, Prov would need to sweep UMass. Anything else, and this compare can't change. Then, Prov needs to sweep NoDame. That would gain them a bit. The problem is that BC will likely play either Vermont or Merrimack, and losing to either isn't going to change their CommOpp much. Without doing all the math, it seems that only a very slight chance exists for Prov to flip this compare.

That results in the following conclusion: Providence, as stated in the OP in this thread, has a strong position because of their already strong RPI.

HOWEVER, Notre Dame is not in as strong a position. Should Providence sweep UMass, and then defeat NoDame, NoDame would be in a very weak and vulnerable position.

I would say at this point that next week's supposed Prov/NoDame series is the first one that we are aware of which has make-or-break ramifications.

Glad to see you expand on your points Numbers! With no dog in the race unless Yale has a very unexpected run in the ECAC's I will look forward to everyones future predictions. BC is also an intriguing team. What do you think of Cornells status?

 

[MagnessMan]

Re: 2017 Pairwise thread

Thanks for getting this started LTsatch. Always lots to discuss in March. The only other tidbit that I would add is that once again St Cloud is in contention within the PWR but one game under 500. They host CC at home. CC has been playing much better since January riding one of many hot Freshmen goalies. It will be interesting to see if the Huskies can pull off the sweep to get back in the race. Agree on the BIG10 possibilities. Very different picture than last year there.

 

[Numbers]

Re: 2017 Pairwise thread

Thanks for getting this started LTsatch. Always lots to discuss in March. The only other tidbit that I would add is that once again St Cloud is in contention within the PWR but one game under 500. They host CC at home. CC has been playing much better since January riding one of many hot Freshmen goalies. It will be interesting to see if the Huskies can pull off the sweep to get back in the race. Agree on the BIG10 possibilities. Very different picture than last year there.

As regards to the Huskies.... I am not at all sure, but I believe it makes sense, in a quantitative sense, to say this:
Given that WMU went tie/win with CC recently and didn't suffer much penalty in RPI, a split by the Huskies would perhaps lower their RPI just a little. Of course, that result would leave them needing a good NCHC playoff run.. a sweep and a semi-final loss likely would not help them very much.

So, to me, it seems SCSU needs at least a tie and a win this weekend.

 

[Fighting Sioux 23]

Re: 2017 Pairwise thread

As regards to the Huskies.... I am not at all sure, but I believe it makes sense, in a quantitative sense, to say this:
Given that WMU went tie/win with CC recently and didn't suffer much penalty in RPI, a split by the Huskies would perhaps lower their RPI just a little. Of course, that result would leave them needing a good NCHC playoff run.. a sweep and a semi-final loss likely would not help them very much.

So, to me, it seems SCSU needs at least a tie and a win this weekend.

A couple points in response to your post:

1) Remember that it matters whether results are posted on the road or at home. WMU's road win/tie against CC is not equal to an SCSU home win/tie against CC, and SCSU would likely suffer a harder hit in the RPI than WMU did. If SCSU splits with CC, it will take a hard hit in the RPI.

2) The NCHC has a Third Place game at the Frozen Faceoff. If SCSU splits against CC, a sweep in the first round, and then loses both at the Frozen Faceoff, they would finish 18-19-1 and eliminated from NCAA Tournament consideration. However, If SCSU splits against CC, sweeps either North Dakota or Omaha (with North Dakota being the significantly more likely opponent), and splits at the Frozen Faceoff, they would finish 19-18-1 and very much in NCAA Tournament consideration. The NCHC Third Place game could also have significant importance for North Dakota and Omaha if either are playing in the game, particularly if the opponent is Denver or Minnesota Duluth.

 

[Siouxfaninseattle]

Re: 2017 Pairwise thread

A couple points in response to your post:

1) Remember that it matters whether results are posted on the road or at home. WMU's road win/tie against CC is not equal to an SCSU home win/tie against CC, and SCSU would likely suffer a harder hit in the RPI than WMU did. If SCSU splits with CC, it will take a hard hit in the RPI.

2) The NCHC has a Third Place game at the Frozen Faceoff. If SCSU splits against CC, a sweep in the first round, and then loses both at the Frozen Faceoff, they would finish 18-19-1 and eliminated from NCAA Tournament consideration. However, If SCSU splits against CC, sweeps either North Dakota or Omaha (with North Dakota being the significantly more likely opponent), and splits at the Frozen Faceoff, they would finish 19-18-1 and very much in NCAA Tournament consideration. The NCHC Third Place game could also have significant importance for North Dakota and Omaha if either are playing in the game, particularly if the opponent is Denver or Minnesota Duluth.

And it was a win in that third place game combined with a Wisconsin win in OT that got UND to the tournament in 2014. I could see it being a make or break game for one of the NCHC teams this year.

 

[Numbers]

Re: 2017 Pairwise thread

A couple points in response to your post:

1) Remember that it matters whether results are posted on the road or at home. WMU's road win/tie against CC is not equal to an SCSU home win/tie against CC, and SCSU would likely suffer a harder hit in the RPI than WMU did. If SCSU splits with CC, it will take a hard hit in the RPI.

2) The NCHC has a Third Place game at the Frozen Faceoff. If SCSU splits against CC, a sweep in the first round, and then loses both at the Frozen Faceoff, they would finish 18-19-1 and eliminated from NCAA Tournament consideration. However, If SCSU splits against CC, sweeps either North Dakota or Omaha (with North Dakota being the significantly more likely opponent), and splits at the Frozen Faceoff, they would finish 19-18-1 and very much in NCAA Tournament consideration. The NCHC Third Place game could also have significant importance for North Dakota and Omaha if either are playing in the game, particularly if the opponent is Denver or Minnesota Duluth.

Thank you especially for the home/road reminder. I was wondering why WMU didn't lose any more that weekend. And, also, I had forgotten about the 3rd place game.

Again, with respect to SCSU....split with CC, win the NCHC quarters in 3...Now, they go into the Frozen Faceoff at level .500. A more dicey situation. And, I think it MIGHT be fair to say that their RPI hit from the CC split would mean they would be on the very edge of the bubble, and perhaps needing to win the tourney.

 

[Fighting Sioux 23]

Re: 2017 Pairwise thread

Thank you especially for the home/road reminder. I was wondering why WMU didn't lose any more that weekend. And, also, I had forgotten about the 3rd place game.

Again, with respect to SCSU....split with CC, win the NCHC quarters in 3...Now, they go into the Frozen Faceoff at level .500. A more dicey situation. And, I think it MIGHT be fair to say that their RPI hit from the CC split would mean they would be on the very edge of the bubble, and perhaps needing to win the tourney.

Very true. The bubble is jam-packed too. The difference in RPI between #12 (Notre Dame (in RPI)) and #20 (Boston College) is about the same as the difference between #11 and #12. The final 2-3 spots are going to come down to the wire. Sweeping your first round series vs. winning in 3, or winning the NCHC third-place game vs. tying could be the difference between making the NCAA Tournament or not.

 

[Fighting Sioux 23]

Re: 2017 Pairwise thread

And it was a win in that third place game combined with a Wisconsin win in OT that got UND to the tournament in 2014. I could see it being a make or break game for one of the NCHC teams this year.

Not only that, but Ohio State was up by 2 goals late in the third period before Wisconsin came back and won in OT. Crazy game. Ohio State jumped out to a 2 goal lead, and Wisconsin clawed back to tie it at 2 midway through the second. Ohio State punched back about a minute later and took a 3-2 lead into the third. With about 7-8 minutes left in the third, Ohio State looked to ice it with a goal to take a 4-2 lead, but then Wisconsin immediately responded and scored twice within about a minute of the Ohio State goal, and knotted it up a 4. Ohio State had a couple Grade A chances in OT, but Wisconsin scored to send UND to the NCAA Tournament.

The Sioux were so grateful that they only beat Wisconsin by three in the first round of the NCAA Tournament.

 

[Siouxfaninseattle]

Re: 2017 Pairwise thread

Not only that, but Ohio State was up by 2 goals late in the third period before Wisconsin came back and won in OT. Crazy game. Ohio State jumped out to a 2 goal lead, and Wisconsin clawed back to tie it at 2 midway through the second. Ohio State punched back about a minute later and took a 3-2 lead into the third. With about 7-8 minutes left in the third, Ohio State looked to ice it with a goal to take a 4-2 lead, but then Wisconsin immediately responded and scored twice within about a minute of the Ohio State goal, and knotted it up a 4. Ohio State had a couple Grade A chances in OT, but Wisconsin scored to send UND to the NCAA Tournament.

The Sioux were so grateful that they only beat Wisconsin by three in the first round of the NCAA Tournament.

Wasn't there a third thing that had to happen that last weekend for the Sioux to make tournament? Seems like there was but I can't remember what it was.

 

[Patman]

Re: 2017 Pairwise thread

usually I'd just reply in full... I don't initiate these things... i just try to be the resident math expert.

----

That and I feel some of these discussions aren't worthwhile unless I have better tools on hand and I don't want to take the time to develop them.


To me, the way things are now, its all about RPI unless something really crazy happens because all ties are determined by RPI and you have to up-end multiple pairs to get a real switch-around going.... maybe things would be more "interesting" if they went NFL style tie-breaking (highest places are determined first) and used the pairwise comparison as the breaker... but until then nothing interesting will happen.

The easiest cheating way to see how far you can fall is to hold the COP and COPOP constant and then seeing what the winning percentages will do to you... reason being is unless you play some team truly rancid the COP and COPOP won't move much. If almost everything in the RPI then its just down to the values of the RPI for those teams. How far can you fall, how much could the line budge upwards from the performance of the "field".

That's all I got. I keep saying "this is the year I build my simulation object of doom" and then I never do it. If I do, I'd love to have contingencies out the wazoo.

 

[Fighting Sioux 23]

Re: 2017 Pairwise thread

usually I'd just reply in full... I don't initiate these things... i just try to be the resident math expert.

----

That and I feel some of these discussions aren't worthwhile unless I have better tools on hand and I don't want to take the time to develop them.


To me, the way things are now, its all about RPI unless something really crazy happens because all ties are determined by RPI and you have to up-end multiple pairs to get a real switch-around going.... maybe things would be more "interesting" if they went NFL style tie-breaking (highest places are determined first) and used the pairwise comparison as the breaker... but until then nothing interesting will happen.

The easiest cheating way to see how far you can fall is to hold the COP and COPOP constant and then seeing what the winning percentages will do to you... reason being is unless you play some team truly rancid the COP and COPOP won't move much. If almost everything in the RPI then its just down to the values of the RPI for those teams. How far can you fall, how much could the line budge upwards from the performance of the "field".

That's all I got. I keep saying "this is the year I build my simulation object of doom" and then I never do it. If I do, I'd love to have contingencies out the wazoo.

It's been awhile since I've done an RPI calculation. Is it still a 25/21/54 breakdown?

 

[vizoroo]

Re: 2017 Pairwise thread

Very true. The bubble is jam-packed too. The difference in RPI between #12 (Notre Dame (in RPI)) and #20 (Boston College) is about the same as the difference between #11 and #12. The final 2-3 spots are going to come down to the wire. Sweeping your first round series vs. winning in 3, or winning the NCHC third-place game vs. tying could be the difference between making the NCAA Tournament or not.

Why not just sweep the little kitties (CC) to get a head start.

 

[manurespreader]

Re: 2017 Pairwise thread

To me the issue is not who is close, but where the cutoff will be. In other words, will the 15 team get in? I suspect not. So that makes the UND and St Cloud problem a bit bigger does it not?.

 

[ticapnews]

Re: 2017 Pairwise thread

North Dakota is really lousing up bracketology.

Providence
1. Denver v AQ2
Union v Providence

Cincinnati
4. Western Michigan v Notre Dame
Minnesota v Cornell

Fargo
3. Harvard v North Dakota (host)
Lowell v Pedo State

Manchester
2. Duluth v AQ1
Boston University v Wisconsin

 

[LTsatch]

Re: 2017 Pairwise thread

I am no numbers guy, but would it be an issue if Air Force wins the rest of their games except the AHA final? Could they still get in via RPI?

 

[ticapnews]

Re: 2017 Pairwise thread

I am no numbers guy, but would it be an issue if Air Force wins the rest of their games except the AHA final? Could they still get in via RPI?

There is a chance Air Force can earn an at-large bid, yes.

 

[Skate79]

Re: 2017 Pairwise thread

North Dakota is really lousing up bracketology.

Providence
1. Denver v AQ2
Union v Providence

Cincinnati
4. Western Michigan v Notre Dame
Minnesota v Cornell

Fargo
3. Harvard v North Dakota (host)
Lowell v Pedo State

Manchester
2. Duluth v AQ1
Boston University v Wisconsin

Wow, what a hose job for Harvard. Complete hose job. The #2 seed in the country and we don't stay in the East??? Screwed royally if that happens.

 

[riverchief1]

Re: 2017 Pairwise thread

Wow, what a hose job for Harvard. Complete hose job. The #2 seed in the country and we don't stay in the East??? Screwed royally if that happens.

Our thoughts on Lowell too. If NCAA really wants $ and fans in the seats, they'd find a way to keep Lowell east as otherwise, that will be 1000s of fans and money guaranteed that they'll lose as only parents and couple diehards will go west.