Category Archives: Uncategorized

So long, and thanks for all the fish!

I’m taking a break (indefinite?) from updating this site. There are a number of reasons now is the right time, but first a look back!

It’s been about 20 years that I’ve been independently calculating PWR, first at, then 8 years ago, launching this site. I did it because while there were a few other sources for the PWR table itself, I noticed that when my friends and I were discussing PWR, we were doing a lot of repetitive work that a computer could be doing. I felt that by making that information more accessible, I could add a lot to fans’ understanding of PWR.

What we were all always interested in about PWR was knowing what changes were likely, so my first innovation was the teams’ comparison details, which puts in front of the reader precisely how a teams’ PWR is calculated, making it much easier to guess at likely changes in the future.

About 15 years ago, I took it a step further, realizing that by simulating a weekend’s games, I could forecast likely PWR changes, based on whether a team wins or loses. This truly automated the work we all used to do notepads and the backs of envelopes, providing (somewhat) easy to interpret probability distributions of a teams’ likely PWR based its own performance. This sort of “UND needs to win about half its remaining games to get a bid” prediction was quite revolutionary at the time.

Finally, realizing that the reason we care about forecasting PWR is that we really want to know who’s going to make the tournament at the end of the season, I created the analysis of the wins needed to get a particular PWR rank table, which lets you see at a glance where every team is likely to fall based on its performance in its remaining games. I’m again proud of the simplicity of the presentation which, though it seems obvious in retrospect, was quite groundbreaking in making that sort of information accessible to everyone.

So why stop now? This year really is a perfect storm of conditions for me, as the amount of worked peaked right at a point my interest has fallen. First, because of the pandemic, I just haven’t paid attention to this site or college hockey much in the last two years. Second, over the years other have built on those innovations (imitation being the sincerest form of flattery), so me reviving this work is less necessary for the community. Third, the data source for college hockey scores that I (and the rest of the college hockey community) have been using for 20 years also shut down this year. Finally, the NCAA implemented the first major change to the PWR formula since 2013-14–crediting an OT win as only .55 of a win, thus necessitating a fair amount of work for me to get the rankings accurate again.

So, I’m going to hit the “pause” button on updating this site. Because I don’t want to post inaccurate information, I’ll take down the PWR/RPI tables, but as I mentioned above there are other great sources! I’ll leave the blog here as a historical archive for some time.

So long, and I’m sure I’ll see you around!

The NCAA tournament field is nearly set going into the last day of conference tournaments

Though 7 conference tournament games remain, there’s not a lot of drama left around who will make the NCAA tournament.

You can always follow along on your own:
College Hockey Tournament possibilities

Who’s in?

These 12 teams are in for sure:

  • St Cloud St
  • Massachusetts
  • Minnesota St
  • Minnesota-Duluth
  • Quinnipiac
  • Denver
  • Ohio St
  • Northeastern
  • Clarkson
  • Arizona St
  • Cornell
  • Harvard

Who else can make it

Both participants in the NCHC and ECAC tournaments are already in.

The winner of the Big Ten tournament will also make it (and the loser will not):

  • Penn State or
  • Notre Dame

The winner of Atlantic Hockey will also make it (and the loser will not):

  • American International or
  • Niagara

That’s 14 teams in, and leaves 2 slots.

There are two more conference championships, Hockey East and the WCHA, whose winners can secure an invitation. These teams can lock up an invitation with a win:

  • Bowling Green
  • Boston College

But, we’re already counting their opponents, Minnesota St and Northeastern as in. So, if Minnesota State or Northeastern win, that leaves an additional slot and an opportunity for another team to make it at-large.

If BC loses, Bowling Green and Providence make it at large
If BC wins, Bowling Green makes it if they win, otherwise Providence makes it

What needs to happen for my team to make it?

Restated in terms of what they need to do to make it:

  • Penn State needs to win
  • Notre Dame needs to win
  • American International needs to win
  • Niagara needs to win
  • Boston College needs to win
  • Bowling Green needs to either win or have Northeastern defeat Boston College
  • Providence needs either Northeastern to defeat Boston College or Minnesota St defeat Bowling Green

What will be the conference representation in the NCAA tournament

The ECAC is likely to place the most teams into the NCAA tournament this year. Here’s a complete rundown of the possibilities.

Big Ten0%0%100%0%0%0%
Atlantic Hockey0%100%0%0%0%0%
Hockey East0%0%6%72%22%0%

As you know from Who can make the NCAA tournament?, the winner of the Atlantic Hockey tournament is the only team from that conference that will make the NCAAs.

Similarly, the winner of the Big Ten tournament is the only team that will join Ohio State in representing the Big Ten, guaranteeing the Big Ten two representatives.

Arizona State will be the lone independent in the NCAA tournament.

The NCHC is guaranteed to get St Cloud St, Denver, and Minnesota Duluth in, and will most likely have 3 representatives. But, if (and only if) Colorado College wins the tournament, the Tigers will get an automatic bid, giving the NCHC 4 teams in the NCAAs.

The WCHA is only guaranteed one team, Minnesota State, and it is possible that the Mavericks will be the WCHA’s only representative. But, the WCHA is more likely to get 2 in by having Bowling Green join them (which could happen even if Minnesota State wins the conference tournament).

Hockey East is only guaranteed two teams, Massachusetts and Northeastern, but 3 or 4 is much more likely. Boston College or Boston University can make it by winning the conference tournament, and Providence can make it at-large.

The ECAC has the most possible different permutations of teams that could make the NCAAs, though landing 4 teams in the NCAAs is by far most likely. Quinnipiac and Clarkson are the only teams in for sure, but there is no situation in which they are the only two who make it. Brown can also make the NCAAs by winning the conference tournament. Cornell and Harvard can make it either by winning the conference tournament, or possibly by earning an at-large bid depending on other outcomes. The ECAC should see no fewer than 3 teams and no more than 5 in the NCAAs.

Conference tournament play-in weekend PWR outlook

This weekend most teams are playing best of 3 conference tournament play-in series, with the exception of two Big Ten semifinal games. The range of likely outcomes isn’t particularly broader than usual, because the best a team can do is still two wins and the worst two losses. But, this weekend’s interesting twist is that the worst outcome would also result in no additional games until NCAA tournament selection, so some teams will start to lose control of their own PWR destiny this weekend.

Follow along with the one-week forecast:

Wins needed to likely have PWR rank on March 18, 2019

#1 St Cloud is probably going to be #1 a week from now, no matter what happens this weekend. The Huskies’ RPI of .6175 is simply unassailable, with #2 Massachusetts at a distant .5827.

The top 6—#1 St Cloud, #2 Mass, #3 Minn St, #4 Minn Duluth, #5 Quinnipiac, and #6 Ohio St—should all still be top 10 after this weekend. Their worst case scenarios for the weekend would also mean the end of their conference tournament play, so a top 10 finish this weekend would make it very likely that they would still be positioned to receive at-large bids to the NCAA tournament by the time conference tournaments are over.

It might appear that Ohio State really belongs in the group below this one, with its RPI of .5595 being much closer to #7 Denver’s .5572 than #5 Quinnipiac’s .5714. But, because Ohio State can lose at most one more game, it’s difficult for them to drop much below #10 at the worst (i.e. under a 1% chance). Even then, everything below them would have to go just wrong for the Buckeyes to miss out on an NCAA bid from there.

From #7 Denver down to #12 Cornell could end the weekend on the bubble, with a chance of making the NCAA tournament at-large, even with losses this weekend.

Those teams include:

  • #7 Denver
  • #8 Northeastern
  • #9 Providence
  • #10 Arizona St
  • #11 Clarkson
  • #12 Cornell

From #13 Western Michigan down to #22 Lake Superior St can keep their hopes of an at-large bid alive by winning this weekend, but will probably move out of position for an NCAA bid if eliminated.

Those teams include:

  • #13 Western Michigan
  • #14 Harvard
  • #15 Notre Dame
  • #16 Bowling Green
  • #17 Union
  • #18 North Dakota
  • #19 Penn State
  • #20 Mass.-Lowell
  • #21 Minnesota
  • #22 Lake Superior St

#23 Northern Michigan and below stand little chance of making the NCAA tournament at-large, so their only likely paths to the NCAAs are through a conference tournament championship and automatic bid.

First 2019 PWR forecast available (and, what is it?)

After the holiday break is generally considered a good time to start paying attention to the PairWise Ranking (PWR). Remember that PWR is a ranking whose calculation mimics the NCAA’s tournament selection process, so the final PWR ranking perfectly predicts the teams that will be selected for the NCAA tournament. The 2014 article, When to start looking at PWR (revisited), examines how today’s PWR is reasonably predictive of the final PWR.

Current PWR Ranking

The reason you’re really interested in the PWR is because you want to know who is going to make the NCAA tournament. So what you really want to know is what the PWR is going to be at the end of the season. To help bridge the gap from today’s PWR to the end of season PWR, I’ve developed a forecast that shows you what the PWR is likely to be at the end of the regular season. It’s explained in the 2017 article, New forecast presentation—PWR by wins, and can be used to answer questions such as:

  • How many wins does my team need to make the tournament?
  • Can my team make the top 4?
  • What are some unlikely tournament seeding outcomes that could occur?

PWR By Wins (What does it take for each team to finish at each PWR ranking?)

The table on the PWR By Wins page shows you how many wins each team needs to likely finish at each PWR ranking.

So, for example, if Mass. wins half its remaining scheduled games (7 of the 14), it will probably end the regular season in the 10-12 range, and thus likely to make the tournament.

If you want more detail on a specific team, you can click a team name to see the probability curves of how likely that team is to end the regular season with each PWR ranking with a given number of wins in its remaining scheduled games.

That helps you see, for example, that though Mass. has a decent change of finishing anywhere from #3 to #9 with 8 more wins, #5-#7 are much more likely.

The forecasts are usually updated in the first half of the week. You can always browse all the data any time, but I’ll also scour the data and post interesting results and observations in this space in coming weeks.

How it works

The page notes when the forecast was last run (assume that it includes all games that had been completed as of that time).

Each forecast is based on at least one million monte carlo simulations of the games in the described period. For each simulation, the PairWise Ranking (PWR) is calculated and the results tallied. The probabilities presented in the forecasts are the share of simulations in which a particular outcome occurred.

The outcome of each game in each simulation is determined by random draw, with the probability of victory for each team set by their relative KRACH ratings. So, if the simulation set included a contest between team A with KRACH 300 and team B with KRACH 100, team A will win the game in very close to 75% of the simulations. I don’t simulate ties or home ice advantage.

What Yale, Northeastern and Minnesota Duluth need

Remember that 11 teams are already in.

These teams can’t get in at-large, but can claim a spot by winning their conference tournament:

  • Minnesota
  • Robert Morris or RIT
  • Minnesota State or Ferris State

These teams are hoping for at-large bids:

  • Yale (idle)
  • Minnesota-Duluth
  • Northeastern

The Big Ten Championship is the biggest determinant of who gets in, because it’s the only one in which it’s not yet known whether its winner will take away an at-large spot. If Minnesota wins the Big Ten and takes away a spot, then Northeastern can only get in by winning, otherwise Yale and Minnesota-Duluth get the two spots. If Minnesota wins and Northeastern does take one of the spots, then Minnesota-Duluth needs to win to get the last spot, otherwise it goes to Yale.

Here’s how each team gets in


Michigan wins
Northeastern loses
Minnesota-Duluth loses


Michigan wins
Minnesota Duluth wins
Northeastern loses


Michigan wins
Northeastern wins

Friday night update

These 11 teams are locks:
North Dakota
St Cloud St
Boston College
Boston University
Notre Dame

These teams are hoping for at-large bids:
Yale (idle)

These teams can claim a spot by winning their conference tournament:
Robert Morris or RIT
Minnesota State or Ferris State

Each of Minnesota-Duluth and Northeastern is also in if they win their conference tournament.

So, given that the Big Ten is the only remaining conference in which it’s unknown whether the winner will consume an at-large spot or not, Minnesota is the wildcard. If they win, they will take a slot, denying one of Yale, Minnesota-Duluth, or Northeastern. If Michigan wins, all three get in regardless of their own outcome.

What could be decided by today’s games

We’re not particularly likely to know a lot more about who’s in or out after today’s games because there aren’t a lot of edge cases that just depend on one or two outcomes. Instead, there’s a decent size group of teams fighting for the middle and we probably won’t know who makes it until they’re all either eliminated or have won their conference tournaments.

But, referring back to NCAA tournament possibilities (with less math), here are some things to watch for:

Teams still playing that can make it at-large

#11 Harvard
#13 Minnesota-Duluth
#14 Northeastern
#15 Michigan Tech

The more of these that lose, the better for the others and idle Yale and Notre Dame (who are also hoping for at-large bids).

Teams that won’t chew up an at-large spot if they win

Big Ten – Michigan
ECAC – Quinnipiac
Hockey East – Boston College, Providence, Mass.-Lowell
NCHC – North Dakota, St. Cloud St, Denver

If all the teams listed for a given conference lose, there will be one less at-large spot available. So, most relevant today are Michigan and Quinnipiac.

A few things that could be decided today

Again, it takes some deep combinations for anything big to happen today, but here the most interesting things I’ve found.

If Quinnipiac, Michigan, and Providence all lose, Cornell is out.

If Michigan Tech and Minnesota Duluth lose, Harvard controls its own destiny (in with a win).

If Michigan Tech and Northeastern lose, Yale is in and Harvard controls its own destiny (in with a win).

If Northeastern and Minnesota Duluth lose then Yale, Harvard, and Notre Dame are in.

Correction – removed incorrect scenario that involved Harvard losing. Harvard losing is not sufficient to make any changes to any teams’ outlook.

How Cornell can make the NCAA tournament


You may want to read the Background on how PWR matters for tournament selection and How the “can make it at-large teams” can make it sections of NCAA tournament possibilities (with less math), if you haven’t already.

How Cornell can make it

Idle Cornell is currently sitting at #16 in the PWR, but has the possibility of landing anywhere between #13-#19. Because the Atlantic Hockey conference tournament winner will come from outside the top 16, only teams that finish in the top 15 stand any chance of an at-large bid.

So, Cornell is looking for some combination of climbing to the 13-15 range and having 3 to 5 of the conference tournaments won by teams ranked ahead of them.

Climbing in RPI ranking

Because Cornell is idle, their two levers to move their RPI ranking are to have their opponents’ (and opponents’ opponents’) win percentage rise to increase their own RPI, and to have teams that are just ahead of them (and still playing) stumble and drop below them in RPI.

Looking at Cornell’s weighted games played against teams that are still playing, Cornell has the largest potential RPI boost from Quinnipiac, Dartmouth, Harvard, and St. Lawrence winning. Of course, those teams all play each other, so it’s best to go with Quinnipiac as the champ and Harvard as runner up to get the most upward bounce. Providence winning its conference is also common in Cornell’s successful scenarios. Cornell’s weighted games played vs teams that are still playing are the following:
Quinnipiac 5
Dartmouth 2.4
Harvard 2.2
St. Lawrence 1.6
Providence 1
Ohio State 1

The other way Cornell can pass teams in PWR is to have teams immediately ahead of them have their own RPI fall enough to be passed. All of #13 Minnesota-Duluth, #14 Northeastern, and #15 Michigan Tech are active, so each losing can clear the way for Cornell to rise.

Conference tournaments won by high ranked teams

In addition to winning more PWR comparisons, Cornell needs conference tournaments won by teams ranked ahead of them. We learned above that Quinnipiac and Providence wins are key to raising Cornell’s RPI, so those two winning their conference tournaments are the cornerstone of most successful scenarios for Cornell. That leaves the following additional possibilities for the necessary 3-5 high-ranked conference tournament champions:
Big Ten – Michigan
NCHC – North Dakota, St Cloud St, Denver
WCHA – Michigan Tech

Remember, how many of those Cornell needs to happen depends on how many teams Cornell can pass in PWR.


Needing both to increase their PWR ranking (through a combination of raising their own RPI and others’ falling) and to have top teams win most conference tournaments makes this a pretty long short for Cornell, with the Big Red advancing in only about 1% of possible scenarios. Those scenarios generally include:

  • Increasing Cornell’s RPI (e.g. wins by Quinnipiac, Harvard, and Providence)
  • and/or decreasing RPI of teams directly ahead of Cornell (e.g. Minnesota-Duluth, Michigan Tech, and/or Northeastern losing)
  • and having enough top teams win their conference tournaments that Cornell’s final PWR rank is included at-large (e.g. Quinnipiac, Providence, and Michigan as winners)

Example of Cornell making it from #15

Example of Cornell making it from #13

Edit – original version stated that Northeastern was idle, updated to reflect that they’re still playing.

How Notre Dame could miss the NCAA tournament


You may want to read the Background on how PWR matters for tournament selection and How the “can make it at-large teams” can make it sections of NCAA tournament possibilities (with less math), if you haven’t already.

How Notre Dame could miss

Given that Notre Dame is already #12, there are myriad ways to get the Irish to 12-13, totaling about 60% of total remaining scenarios (see Notre Dame PWR details).

Because that 12-13 range is the lowest the idle Irish can fall, the most important contributor to excluding Notre Dame is to have conference tournaments won by low-ranked teams. That chews up spots that could otherwise have gone to at-large teams, what we call “moving the cutline”.

For example, the following scenario has Minnesota-Duluth flip its comparison with Notre Dame to push the Irish down to 13. It then has 4 other conferences won by teams that wouldn’t have otherwise auto-qualified:

Those factors come together in about 7% of remaining scenarios.