Category Archives: forecasts

Pairwise Ranking outlook for this weekend

1 Reply

Poised to climb

#44 Michigan State (8-12-3) probably isn’t on anyone’s radar, but if they pull off an unlikely sweep at Minnesota, the Spartans can jump as many as 17 spots to the high 20s. That demonstrates the power of the quality win bonus (QWB) which would contribute .05 RPI points, or about a 3 spot jump, for two wins over the #1 (though perhaps then #2) team at home.

thisweek1

Lest Spartans fans get too excited, the rest of Michigan State’s schedule isn’t so kind, so it would really take winning out the season to complete the climb into the teens.

The only teams outside the top 20 with a good shot (over 1%) of climbing into the top 14 this weekend are:

  • #21 Western Michigan (could climb as high as #11)
  • #22 New Hampshire (could climb as high as #13)

thisweek2

thisweek3

Again, each would be helped in their ascent by the QWB.

Poised to fall

Highly ranked teams with significant downside potential include:

  • #10 Michigan (could fall as low as #21)
  • #14 Minnesota-Duluth (could fall as low as #24)

thisweek5

thisweek4

Bubble teams of interest

Finally, #16 Clarkson and #18 Yale are having their moments in the sun. Each could climb into place for an at-large bid, or fall deep into the 20s.

clarkson

yale

Methodology

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

How many teams will each conference put in the playoffs (and is the NCHC really underperforming?)

2 Replies

There has been a lot of discussion about the NCHC seemingly having a down year based on its current poor performance in the Pairwise Rankings. A comment by Goon on a different blog asked whether Hockey East and ECAC might fall as additional conference play leads them to beat up on each other.

To help assess the NCHC’s performance and answer Goon’s question, I ran simulations of the remaining games and tracked how many teams each conference had in the top 14 at the end of the regular season (a reasonable guess as to the PWR rank that would guarantee an invitation to the NCAA tournament).

First, here is the current distribution of the top 14 PWR ranks by conference:

Number of teams in the top 14 in PWR right now
Atlantic Hockey 0
Big Ten 3
ECAC 3
Hockey East 5
NCHC 2
WCHA 1

Next, ExileOnDaytonStreet on the USCHO Forum calculated each conference’s current members average tournament appearances per year over the past ten years:

Number of members that made that tournament per year
Atlantic Hockey 1.3
Big Ten 3
ECAC 2.3
Hockey East 4.1
NCHC 4.3
WCHA 1

When comparing the numbers across the two tables, keep in mind that the first table only includes the top 14 while the second includes all tournament participants. To directly compare across the two tables, two auto bids that wouldn’t make the tournament at large need to be added to the first table.

Finally, here are the results of the simulations:

chart_1

With 0 teams currently in the top 14 and a historic average of 1.3 in the tournament, AH is about on par as it is almost certain to end with 0 teams in the top 14 so one participant in the tournament as an auto bid.

chart_2

With 3 teams currently in the top 14 and a historic average of 3 in the tournament, the Big Ten is about on par with 2-3 teams in the top 14 a likely finish.

chart_3

With 3 teams currently in the top 14 and an average of 2.3 in the tournament, the ECAC is outperforming with 3-4 teams in the top 14 a likely finish.

chart_4

With 5 teams currently in the top 14 and a historic average of 4.1 in the tournament, Hockey East is about on par to slightly outperforming with 4-5 teams in the top 14 a likely finish.

chart_5

With 2 teams currently in the top 14 and a historic average of 4.3 in the tournament, the NCHC is definitely underperforming with 2 teams in the top 14 a likely finish (though 1 or 3 distinctly possible).

chart_6

With 1 team currently in the top 14 and a historic average of 1 in the tournament, the WCHA is about on par with 1 team in the top 14 a likely finish (though 2 distinctly possible).

The projections are KRACH-weighted, so implicitly assume that teams will continue to perform as they have to date. That helps isolate the specific effects of the remaining schedule. The remaining schedule is not likely to push Hockey East or the ECAC down from their current levels, though individual teams’ performances certainly could.

The conclusion — even if its members outperform and its auto bid goes to a team that wouldn’t qualify at large, the NCHC seems quite likely to finish with a down year relative to its members past performances.

Pairwise Rankings (PWR) and the NCAA hockey tournament outlook

Leave a reply

With eight weeks left in the regular season, some teams’ NCAA tournament chances are starting to firm up while others still have the opportunity to climb into or fall out of contention for an at-large bid.

To help get a picture of those possibilities, I ran simulations of the rest of the regular season and calculated each team’s potential Pairwise Rankings (PWR) (which mimic the NCAA’s tournament selection process).

A note on the 2014 PWR

The PWR was changed this year in a few significant ways:

  • RPI now includes a quality win bonus for beating top teams
  • RPI now weights home losses/road wins higher than home wins/road losses for the win percentage criterion
  • PWR no longer includes the Teams Under Consideration criterion

By shifting the credit for strong wins from a PWR criterion (TUC) to the RPI (QWB), the other PWR criteria are deemphasized. With only three criteria in PWR and RPI as the tie-breaker, PWR will only deviate from RPI when head-to-head results come into play.

Teams that are almost a lock

A few teams’ performances have already earned them strong enough RPIs that they’re almost locks for at-large bids to the NCAA tournament. That level seems to be above about a 57 RPI index.

minnesotaendofseason

#1 Minnesota (61.2 RPI) currently wins all of its PWR comparisons.  Even if the Gophers were to win only 4 of their remaining games, they would still end the regular season in the top 10.

bcendofseason

#2 Boston College (59.97 RPI) currently wins 57 of 58 PWR comparisons. Winning 4 of its remaining games would result in a top 10 regular season finish, while winning only 2 would still leave the Eagles in a likely position for an at-large bid.

quinnipiacendofseason

#3 Quinnipiac (57.63 RPI) currently wins 56 of 58 PWR comparisons. The Bobcats need to win 4 of their remaining 8 games to be certain to be in position for an at-large bid at the end of the regular season, though could still be in contention with just 2 wins.

Teams that are almost out of contention

On the other end of the RPI rankings, some teams have performed poorly enough to date that they have very little chance of making the tournament at-large. That line seems to be below about a 49 RPI index.

alaskaendofseason

#36 Alaska (49.39 RPI) currently wins 24 of 58 PWR comparisons. Even if Alaska won out, they would likely be just on the outside of an at-large bid at the end of the regular season.

bentleyendofseason

#39 Bentley (49.1 RPI) currently wins 19 of 58 PWR comparisons. Winning out would not provide Bentley enough of a boost to make the tournament at-large by the end of the regular season.

The outlook is similarly grim for the 21 teams below Bentley in RPI:

Bemidji State
Michigan State
Michigan Tech
Massachusetts
Merrimack
Harvard
Connecticut
Boston University
Canisius
Dartmouth
RIT
Princeton
Colorado College
Penn State
Holy Cross
Robert Morris
American Int’l
Niagara
Sacred Heart
Alabama-Huntsville
Army

Methodology

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