clock menu more-arrow no yes

Filed under:

Does Boston College Hockey Have To Win Out? A Rebuttal.

New, 7 comments

They don't. Let's all just settle down, okay?

Arthur Bailin, BC Heights

After putting together a nation's best eight-game unbeaten streak by taking down the highly ranked BU Terriers, the following game's loss to Maine slammed the brakes, at least temporarily, on the Boston College Hockey Don't-Call-It-A-Comeback Tour '15.

Time to panic? Soaring to Glory sure thinks so. STG put out an article earlier today asking, and then answering, the question of whether or not BC Hockey has to win out to hope to have a shot at an at-large bid to the NCAA tournament heading into the conference championships.

While I admire the effort put into trying to figure out how much of a shot BC has at this point in the season, I'm here to emphatically rebut much of that article and get the rest of you panicked souls off the Tobin.

Let's start out with a few factual inaccuracies.

The bad teams are a problem. Playing a few against Notre Dame, UConn, and Northeastern may hurt their PairWise Ranking, even if the Eagles should win those games. Just like playing lousy teams hurts your RPI in basketball, it’s no different in hockey since RPI is RPI.

Nah man, that's completely false. First of all, only the very highest ranked teams in RPI can play a team so much worse than them that a win will lower their RPI. BC is not one of them right now. #1 Minnesota State, #4 North Dakota, #7 Harvard, and #14 Providence (definitely an outlier) are the only ones that have at this point.

The worst team Boston College has played is UMass, and they don't play anyone worse than that the rest of the way. BC's win over UMass helped BC's RPI.

But all that is entirely useless information anyway. Men's hockey is not basketball -- RPI is specifically adjusted to remove "bad wins" that hurt your RPI. Take a look at the bottom of the USCHO Pairwise Rankings:

*Team's RPI has been adjusted to remove negative effect from defeating weak opponent.

What that means is that if a game's RPI is lower than a team's overall RPI, it's removed from the calculation entirely.

So, no worries about beating bad teams and having it hurt our RPI. Our future opponents aren't bad enough for that to matter, and it wouldn't lower our RPI anyway.

Alright, what's the next thing I have a beef with?

When it comes to specific scenarios for Boston College, the rest of Hockey East, and the rest of the teams in Division I, there are far too many variables and potential outcomes to calculate. Boston University losing a game in the middle of February while Denver ties another game with Yale winning another — there are probably thousands and thousands of outcomes for playing out the rest of the college hockey regular season.

This isn't really a big deal but bruh, there are literally over 393 sexquinquagintillion (a real thing) possible outcomes for playing out the rest of the regular season*. That's 393 with 171 zeroes after it. So, yeah, thousands and thousands, technically.

*Source: 350 non-Beanpot games left, 3 possible outcomes per game (win, loss, tie), so 3^350, times 2^2 because no ties in the Beanpot, and 3^350 x 2^2 = that big number.

Sorry, what were we talking about?

Okay, the last thing that was really, really bad was the end, where STG estimated +/- in the rankings based on "wild guesses based on past similar games played during the season."

UConn: Win (-1), Tie (-2), Loss (-4)
Providence: Win (+1), Tie (no change), Loss (-1)
Northeastern: Win (no change), Tie (-1), Loss (-2)
Merrimack: Win (+1), Tie (no change), Loss (-1)
BU or Harvard: Win (+3), Tie (+1), Loss (no change)
Vermont (per game):Win (+1), Tie (no change), Loss (no change)
UMass-Lowell: Win (+2), Tie (+1), Loss (-1)
Notre Dame (per game): Win (-1), Tie (-2), Loss (-3)

If Boston College wins out, they go up six spots in the PairWise and head into the Hockey East Tournament right on the border, but in with one spot to spare. It goes without saying any deviation would likely result in being out.

Alright, well, like I said earlier, I appreciate the effort, but this thing that you did is Bad and you should Feel Bad.

Pairwise doesn't work like this. For a long time, our wins weren't doing much because we were a decent amount below the next team above us. Also, a loss tends to hurt more than a win does. Law of diminishing returns and all that. Guessing how much we are going to go up or down based on a previous game is like guessing how much money you're going to get based on previous pulls on a slot machine.

There's a much, much more helpful method of calculating BC's target record going into the conference tournaments.

College Hockey News has a pretty slick tool that allows you to change past results and input future results to see the resulting Pairwise.

So, we can actually plug in possibilities and see where it spits out BC in the season ending Pairwise. Now, obviously BC's games don't happen in a vacuum—everyone else is winning and losing games too, after all—but if you assume the RPI/Pairwise bubble will remain largely unchanged at about .5500 as teams win/rise and lose/drop around it, it's a pretty decent estimator.

Assuming winning the games against the lowest ranked teams and losing the games against the highest ranked teams, this is what we get:

Note that this doesn't include the Beanpot final—the calculator can't recognize a win or a loss in this game because the opponent isn't known yet.

9-0-0: 4th overall (!!) (.5785 RPI) – undefeated
8-1-0: 10th overall (.5690 RPI) – loss to UML
7-2-0: 10th overall (.5605 RPI) – loss to UML and Merrimack
6-3-0: 15th overall (.5521 RPI) – loss to UML, Mack, and PC
5-4-0: 19th overall (.5443 RPI) – loss to UML, Mack, PC, and UVM.

And those numbers are actually better if we lose to a bad team and beat a good team because of the Quality Wins Bonus added into RPI. For example, going 6-3-0 with losses to UConn, NU, and Notre Dame puts us in 13th.

Factoring in the Beanpot final and how a win would help and a loss would hurt—6-4-0 wouldn't do it, but 7-3-0 sure would.

So that's the target: If BC goes 7-3-0, they should be on the right side of the bubble heading into the conference tournaments. Remember, the bubble teams around us are going to be losing games, too. We just need to lose less.

Let's all just settle down a little bit. There's a lot of season left.