Category Archives: forecasts

How many teams will each conference put in the playoffs? (2017 edition)

In the college hockey community, speculation about who is likely to make the NCAA tournament begins to reach a fevered pitch this time of year. That also leads to inevitable boasting from fans of those conferences having a strong season.

Looking at the top 14 (or 16) of PWR gives an interesting benchmark of performance to date, but this is a situation in which it can be interesting to try to determine the effects of the known remaining regular season schedule. By simulating the rest of the regular season (assuming teams will continue to perform as they have to date), we can see how each conference is likely to fare at the end of the regular season.

How many teams will each conference put in the playoffs?

As mentioned above, a good quick first check is how many teams each conference currently has in the top 14. (The cut line of 14 is based on a historical guess, but also fits quite well with the current situation of two conferences with autobids not having any teams in the top 14).

Teams currently in the top 14 of PWR
Atlantic Hockey 0
Big Ten 3
ECAC 3
Hockey East 3
NCHC 5
WCHA 0

But, by simulating the rest of the regular season (assuming teams will continue to perform as they have to date), we can see how the remaining schedule might affect those numbers.

Likelihood of each conference’s number of teams in top 14 PWR at the end of the regular season
0 1 2 3 4 5 6
Atlantic Hockey 41% 59%
Big Ten 2% 35% 57% 6%
ECAC 9% 91%
Hockey East 2% 37% 48% 13% 1%
NCHC 21% 50% 29% 1%
WCHA 100%

Comparing the two tables, Atlantic Hockey and Hockey East are more likely to make a small gain before the end of the regular season than stay at their current level. That’s not surprising, as Atlantic Hockey and Hockey East teams currently hold PWR spots 15 and 16-18, respectively. The NCHC is more likely than not to lose a team, with St Cloud St and North Dakota sitting at #12 and #14, respectively. In fact, the NCHC seems about as likely to end with just 3 teams in the top 14 as its current 5. The Big Ten, with teams in PWR spots 11 and 13, is also reasonably likely to lose a team.

Methodology

Forecasts include the results of games played through Tuesday of this week.

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.

Resources

NCAA field shaping up with three weeks left in regular season

With three weekends of left of regular season play for most conferences (the Big Ten has four), the field of who is likely to make the NCAA tournament has really shaped up in the last week.

You may want to follow along with the data, this narrative is all based on the forecast of PWR by wins, which shows the number of wins necessary to be likely to end the regular season at a particular PWR ranking.

Keep in mind that conference tournaments could result in additional movements, though by then we generally know that teams in the mid-teens need to win to get an at-large bid.

Near locks

Only the top two, #1 Minnesota-Duluth and #2 Denver actually seem to be locks regardless of outcomes. Neither is likely to fall below the #4-7 range, even if swept, so would likely still get an at-large bid even in the worst case 0-2 scenario in the subsequent conference tournament. Note that’s not yet a mathematical certainty, just quite likely.

#1 Minnesota Duluth
#2 Denver

Likely in unless they collapse

#3 Harvard through #9 Mass.-Lowell need at least one win to be likely to finish in the top 13, and thus well positioned for a tournament spot. Because there is an extra weekend of Big Ten play remaining, #4 Minnesota and #6 Penn State actually 3-4 wins to be well positioned for a tournament spot.

#3 Harvard
#4 Minnesota
#5 Boston University
#6 Penn State
#7 Western Michigan
#8 Providence
#9 Mass.-Lowell

Control their own fate

#10 North Dakota through #21 Wisconsin plus #25 Miami control their own fate in reaching the NCAA tournament. #10 North Dakota appears to need 3 wins in its remaining 6 games to be top 13 going into conference tournament play, while #20 Air Force needs to win out its final 4 to stand a decent chance of finishing the regular season in the top 13. #21 Wisconsin has much more room to move because of the extra weekend of Big Ten play remaining. #25 Miami is the lowest ranked team that still seems to control its own ability to jump onto the bubble, with a sweep out of 6 wins likely to land them in the 10-13 range to close out the regular season.

#10 North Dakota
#11 Union
#12 St Cloud St
#13 Cornell
#14 Ohio St
#15 Boston College
#16 Vermont
#17 Nebraska-Omaha
#18 Notre Dame
#19 St. Lawrence
#20 Air Force
#21 Wisconsin
#25 Miami

#22 Quinnipiac through #24 Bemidji State and #26 Canisius and below are unlikely to crack the top 15 in the regular season, so would need near perfection and some luck to sneak into position for an at-large bid. Then, continued success in the conference tournament would be required not to fall out of position. These teams should plan to do well in their conference tournaments.

#22 Quinnipiac
#23 Northeastern
#24 Bemidji State
#26 Canisius
#27 Clarkson
#28 Robert Morris
#29 Merrimack
#30 Princeton
#31 Michigan Tech
#32 MSU-Mankato
#33 Yale
#34 Connecticut
#35 Army
#36 New Hampshire
#37 Colorado College
#38 Dartmouth
#39 Holy Cross
#40 Bentley
#41 Michigan
#42 Bowling Green
#43 Maine
#44 Arizona State
#45 Colgate
#46 Sacred Heart
#47 Mercyhurst
#48 RIT
#49 Michigan State
#50 Lake Superior
#51 Northern Michigan
#52 Ferris State
#53 Alaska
#54 Alaska-Anchorage
#55 American Int’l
#56 Alabama-Huntsville
#57 Rensselaer
#58 Brown
#59 Massachusetts
#60 Niagara

Methodology

Forecasts include the results of games played through Thursday of this week.

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.

Resources

New forecast presentation – PWR by wins

I’m pleased to announce an improvement to the way I present PWR forecasts this year. There were two guiding principles to the design of the new presentation:

  • The question people are really asking until conference tournaments begin is, “what will it take for my team to make the playoffs (or finish top 4)?”
  • Everyone is interested in something a little different—some are fans of a single team and just care about that team, some want to check up on rivals, and some want to dig through all the data to look for interesting outcomes.

My forecast posts in past years gave some insight into what it takes for a team to make the playoffs, but was limited to the teams I chose or scenarios I found interesting. To help expand that analysis to all teams, I sought a useful way to present all the data.

The table on the PWR By Wins page shows you how many wins each team needs to likely finish at each PWR ranking. 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 out of its remaining scheduled games.

This is the first public presentation of this, so I’m sure there will be some tweaks and improvements in coming weeks. Check it out, and let me know if there’s anything I can do to make this data more useful to you.

What does it take for each team to finish at each PWR rankings: PWR By Wins

Methodology

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.

The 1 seeds – UND and three of QU, SCSU, PC, BC, DU

The 1 seeds

These teams have a shot at 1-seed:
North Dakota (lock)
Quinnipiac
St Cloud St
Providence
Boston College
Denver

Extracted from PWR Possibilities:

Team Result 1 2 3 4 5 6 7 8
UND Win 0 34% 41% 21% 4%
Win 1 63% 30% 7% < 1%
Win 2 100%
Quinnipiac Win 0 < 1% 35% 39% 25% < 1% < 1%
Win 1 4% 46% 38% 12% < 1%
Win 2 57% 42% 1%
SCSU Win 0 < 1% 29% 40% 24% 7%
Win 1 8% 18% 48% 25% 1%
Win 2 62% 27% 11%
Providence Win 0 1% 58% 32% 9%
Win 1 7% 44% 41% 7%
Win 2 14% 51% 30% 5%
Boston College Win 0 82% 18%
Win 1 < 1% 4% 81% 15%
Win 2 < 1% 29% 41% 29% < 1%
Denver Win 0 98% 2% < 1%
Win 1 1% 3% 93% 2%
Win 2 31% 36% 8% 24% < 1%

UND will finish in the top 4 and get a 1-seed.

Quinnipiac, St Cloud St, and Providence are each guaranteed a top 4 finish, and a 1-seed, if they win their conference tournament.

Quinnipiac is nearly (but not quite!) a lock for a 1-seed regardless of outcome, and St Cloud St is nearly (but not quite!) a lock for a 1-seed if they win 1.

Boston College and Denver each need at least one win, preferably two.

BC to #1 overall

An example:
http://pwp.uscho.com/rankings/pairwise-predictor/?uniq=pwp_56e6c9fa056fc

Obviously Boston College needs to win and North Dakota, Quinnipiac, St Cloud St, and Providence need to lose enough to be passed by BC. Finally, it takes an unusual combination of other outcomes to give BC a small RPI advantage over those teams (note, for example, that the linked scenario actually has Providence win one then lose one to give BC just enough of an RPI edge; note also that the top 4 in that example end up within .0003 in RPI).

This is pretty unlikely, occurring only in about 1 of 940 scenarios in which Boston College wins its conference championship.

Quinnipiac missing a 1 seed

An example:
http://pwp.uscho.com/rankings/pairwise-predictor/?uniq=pwp_56e8cbd15d592

Even with a win, it’s possible for Quinnipiac to fall to 5th and miss out on a 1 seed. Success by other teams eligible for #1 is a key, in this case Boston College, Denver, and St Cloud St. Note how unusual this scenario is, with a 3-way tie in pairwise comparisons and the RPI tie-breaker settled by just .001 between the three teams.

This is pretty unlikely, occurring in only about 1 of 1900 scenarios in which Quinnipiac wins one, or 1 in 210 scenarios in which Quinnipiac loses its first game.

These teams should really try to win this weekend

Need to win

These teams clearly need to win this weekend. A pair of losses could plummet any of them into the 20s and out of reach of an at-large bid.

#13 Minnesota-Duluth
#14 Michigan Tech
#15 Nebraska-Omaha
#16 Cornell
#17 Minnesota
#18 Penn St
#19 Northeastern
#20 Clarkson
#22 Rensselaer

On the flip side, a pair of wins could propel any of them onto the 11-13 bubble in preparation for next weekend. Even with a pair of wins, 13 is a bit of a longshot for Clarkson and Rensselaer, while Minnesota will struggle to hit 14.

Minnesota-Duluth is the team most likely to survive a sweep, as the Bulldogs face a huge range of possible outcomes if swept—from 14 to 22 being somewhat likely, though 16-20 is most likely. Where they fall in the range will be determined by how the other teams on the list above perform — the more that lose, the better for Minnesota-Duluth.

umd

Michigan Tech and Nebraska-Omaha seem most likely to fall to 15 or below if they don’t emerge from the weekend victorious, so would face an uphill battle to get an at-large bid with no additional games to play.

mtech

uno

Cornell could land in the 13-14 range with a single win this weekend (though 15 is more likely), which would leave them alive but with no control.

cornell

Minnesota and Penn State would each have an additional shot to climb in the Big Ten tournament, even if swept this weekend. But, each would also be in a pretty deep hold without a win. Even with a sweep, Minnesota won’t climb much, with 15-18 the most likely outcome. The Gophers would need an additional conference tournament win to gain confidence in an at-large bid. Penn State fares a bit better with either a single win or a pair of wins, being able to climb to the top of the bubble with the latter.

minnesota

pennstate

Though Northeastern could climb to the 12-13 range with a pair of wins, 14-15 is much more likely. Clarkson and Rensselaer face a broad range of possible outcomes with a sweep, being able to reach 13-14, but with 15-18 more likely, and even 19-21 realistic. All three need to win, plan to win next weekend, and get some help from others.

northeastern

clarkson

rpi

Mathematically possible?

These teams are long shots for at-large bids. Even with a sweep this weekend it would take a lot of luck to come out ranked #15. A subsequent win and a loss in the conference tournament shouldn’t move either much, so it would take some PWR luck and some cutline luck to get in at-large. Better just plan to win the conference tournament.

#21 Robert Morris
#23 Miami

rmu

miami

Win the conference tournament

These teams could give their fans some false hope by cracking the teens with a pair of wins this weekend. Unfortunately, it’s pretty unlikely to climb from there to above the cutline with a win and a loss.

#24 Minnesota St
#25 St. Lawrence
#27 Dartmouth

mankato

slu

dartmouth

Other teams not yet mentioned don’t even seem likely to be able to crack the teens this weekend. It would take an extremely unusual confluence of events for any of them to move into position for an at-large bid.

Six teams have an opportunity to dramatically increase NCAA at-large odds this weekend

Likely in—#7 Yale, #8 Michigan, #9 Boston University, #10 Notre Dame

These teams are in good shape and likely to get an at-large bid if they win more than they lose in the coming two weekends.

If swept this weekend, their worst likely outcomes are falling to the 11-12 range, which would require keeping an eye out for potential additional downward movement and the cutlines (that is, how many of the at-large spots are taken away by autobids ranked lower than 16 in the PWR).

yale

michigan

bu

notredame

Of course, for Yale, Boston University, and Notre Dame, a pair of wins means another game. A loss in that game would probably result in giving back up some ground. But, that would still probably result in a safe position unless a lot of low ranked teams win their conference tournaments.

Note that Michigan’s position is slightly different because the Big Ten is still in regular season play. Even if swept by Penn State this weekend, the Wolverines would still have an additional game in the Big Ten tournament to either climb or fall. That worst possible outcome this weekend could drop Michigan as low as #12, from which another loss could push them out of position for an at-large bid.

Likely in if they win—#11 Harvard, #12 Mass.-Lowell

These teams would be in trouble if swept this weekend. Harvard’s likely fall could be to as low as 16, and Mass.-Lowell as low as 15. While it’s possible to climb a little bit while idle and have the cutline low enough to sneak into an at-large bid, it would be a bit of a longshot. I’ll also note that there are some extremely unlikely cases that could push these teams as low as 18 and 17 respectively if swept.

Sweeping, on the other hand, has the possibility of putting either in great shape — as high as 8 or 9 (though 9 to 11 more likely) with only 1 potential loss remaining.

harvard

uml

One more team you should expect to see in the NCAA tournament

The first five—Quinnipiac, St Cloud St, Boston College, North Dakota, and Providence

On Feb. 23, I called the top 5 “starting to look like locks” for the NCAA tournament (in order of their ranking at the time)—#2 Quinnipiac, #3 St Cloud St, #5 Boston College, #1 North Dakota, and #4 Providence. While those 5 have shuffled a bit amongst themselves, they obviously remain the top 5, and are even more certain locks.

None of them are likely to drop out of the top 7 in this weekend’s conference tournament play. Because a worst case scenario for any of those teams would also mean no more games before NCAA selection, there would be limited opportunity for additional downward movement.

und

qu

scsu

providence

bc

Almost had it—Michigan

In that Feb. 23 article, I called (then #6) #8 Michigan too close and too many games remaining to call. The Wolverines have rewarded me for my caution, going 1-3, dropping two spots, and being at risk of hitting the bubble if they continue to lose.

michigan

The new “next one”—Denver

But what about the new #6 Denver (then #8)? They’ve been the hottest PWR team of 2016 with a 14-1-3 record since the New Year, including a 4-0 run since that article.

top10

The Pioneers are facing an elimination best-of-3 that is likely to result in no worse than a #8 PWR ranking, I’m comfortable saying you should expect to see Denver in the NCAA tournament.

Note, however, that Denver has not mathematically secured the bid. In very obscure scenarios, they could fall as low as 9 this weekend if swept (about a .1% of KRACH-weighted outcomes in which Denver goes 0-2). It’s conceivable, though incredibly unlikely, that the then idle Pioneers could slip enough further to fall below the cutline if almost all of the conference tournaments are won by teams outside the top 16. That scenario is extremely unlikely to still be in play after this weekend, but if it is, I’ll give full details of the likelihood and how it could happen.

denver

This weekend’s big PWR movers

Sorry for the lateness and any errors, had to crank this one out in a hurry…

More upside than you might expect

Not this week. Usually a team or two in the mid-20s have the opportunities for jumps approaching 10 spots. Plenty of teams in that range have the chance to jump about 5 spots, maybe even 6 if all goes well, but anything approaching a 10 spot jump from a team of any consequence is very unlikely.

More downside than you might expect

The high teens is a tough place to be this time of year.

#14 Penn State, idle last weekend but on the list of potential big drops two weeks ago, is in familiar territory — win 2 to stay in place, or be prepared to fall. A 10 spot drop to #24 is possible.

pennstate

#15 Minnesota-Duluth, which jumped 10 spots (as predicted) with an impressive pair of wins last weekend, doesn’t have much upside potential but stands to give back its ground if it doesn’t keep up its winning streak. An 11 spot drop to #26 is possible.

minnduluth

#16 Minnesota, which was also on last week’s potential big drop list, finds itself there again. Two wins won’t do much for the Gophers, but losses could result in up to a 9 spot drop to #25.

minnesota

#17 Michigan Tech, #18 Miami, and #20 Cornell also face potentially significant drops of up to 7-8 positions if they lose 2. They each, however, face more normal upside potential with success.

mtech

miami

cornell

Weekend PWR watch — first conference tournament round

Watching the top

#2 North Dakota used a pair of wins over #13 Nebraska-Omaha to climb past #3 St Cloud, which dropped a pair of games to desperate #15 Minnesota-Duluth. That leaves UND as the only team capable of taking over #1 from idle Quinnipiac this weekend.

und

In last week’s outlook I predicted that the top 12 were generally safe, with no one likely to drop below 14. Though there was some minor shuffling within the group, only Nebraska-Omaha fell out of the top 12, landing at #13 following a pair of losses to North Dakota.

top13

Safe this weekend

This week’s top 12 look similarly safe for another week, with no one likely to come out of the weekend below #13. Note that quite a few of these teams, including #9 Notre Dame, #11 Harvard, and #12 Mass.-Lowell, are insulated against much movement by being idle this weekend. Teams likely safe in the top 13 include:

#1 Quinnipiac
#2 North Dakota
#3 St Cloud St
#4 Providence
#5 Boston College
#6 Michigan
#7 Denver
#8 Yale
#9 Notre Dame
#10 Boston University
#11 Harvard
#12 Mass.-Lowell

uml

Around the bubble

Most teams from #13 Nebraska-Omaha down to #21 Minnesota State could end up on either side of the bubble—at #14 or higher after this weekend, or below 14. The notable exceptions are #16 Minnesota will struggle to climb above #15 and idle #19 Robert Morris isn’t likely to climb beyond #17.

#13 Nebraska-Omaha
#14 Penn State
#15 Minnesota Duluth
#17 Michigan Tech
#18 Miami
#20 Cornell
#21 Minnesota St

uno

mankato

Need to climb

A handful of teams can’t quite make it into position for an at-large bid (which we’re currently calling #14 and better), but can come out of the weekend in the teens, preserving hope for an at-large bid with additional conference tournament success. Those include:
#16 Minnesota
#19 Robert Morris
#22 Northeastern
#23 Clarkson
#25 Rensselaer

minnesota

bobmorris

northeastern

clarkson

renssalaer

Need to win their conference tournaments

Idle #24 St Lawrence, and #26 Bowling Green and below are unlikely to come out of the weekend in the teens, and thus face a steep climb to gain an at-large bid.

slu

bgsu

A couple data notes

The Eastern teams are already in conference tournaments, and thus playing best of three series this weekend. Their “win 2” curves include both the 2-0 and 2-1 scenarios. They are generally double-topped, and the 2-0 scenario is the better outcome.

There seems to be some confusion about whether the Feb. 20 Denver vs CC matchup was a neutral site game. As of now, USCHO, CHN, and the Denver site all list it as a home game for Denver; while collegehockeystats.net and the CC site list it is a neutral site game. This has a minor effect on RPI which does not affect any of the above outcomes; though these calculations assume it was a neutral site.

This weekend’s big PWR movers

Looking up!

As mentioned in yesterday’s PWR outlook article, all the teams from #20 St. Lawrence to #26 Clarkson have the potential to climb into the teens this weekend. But, #25 Minnesota-Duluth and #26 Clarkson, in particular, have the potential to make huge jumps.

The #25 Bulldogs could find themselves as high as #14, a jump of 11 spots, with a sweep of #2 St. Cloud St. Most likely with a sweep is a still unusual climb of 8-9 positions to #16-17.

duluth

#26 Clarkson could find itself as high as 16 with wins over Dartmouth and Harvard, though #19-21 are more likely. The likely outcomes for Clarkson with two wins cover a surprising #16-#23 span. Clarkson’s outcome with two wins would rest a lot on other games, with the Golden Knights cheering for sweeps from Alaska Anchorage over Minnesota St, St. Cloud St over Minnesota-Duluth, Maine over Northeastern, Alaska over Michigan Tech, and Canisius over Robert Morris.

clarkson

Going down?

Every team in the #14-22 range, except idle #15 Penn State, could fall more than five positions with a pair of losses. #14 Cornell, #17 Miami, #18 Michigan Tech, and #16 Minnesota are facing particularly steep falls.

Cornell could see itself fall as low as #24 with losses to Rensselaer and Union, though the spread of likely outcomes stretches a huge span of 17-24.

cornell

Miami faces a steep drop to as low as #29 if swept by Colorado College, with a still significant drop to 26-27 most likely.

miami

Michigan Tech could find itself as low as #28 if they drop a pair to Alaska, though #24-26 are more likely.

mtech

If Minnesota gets swept by Michigan, it will face a broad range of potential outcomes that stretches as low as #25. #19-23 are all reasonably likely in that scenario.

minnesota