Hockey East Communications Reveal HEPI Formula, Frustration Among Coaches

With COVID roadblocks preventing Hockey East from keeping to a planned conference slate and causing significant scheduling imbalances, the league announced last month that the league standings would be set using a “customized objective mathematical formula never before used in college sports” called the “Hockey East Power Index” (HEPI, for short) in order to mitigate those imbalances and number of games played.

However, as we’ve pretty frequently pointed out, the league never did release the formula they used to seed the teams, and in fact we were explicitly told that they had no plans to do so:

Hockey East commissioner Steve Metcalf later joined The BU Hockey Show’s podcast to further confirm that the league would not be releasing the formula, despite repeated calls for them to release it in the interest of transparency from supporters across the conference.

In light of this, BC Interruption submitted a Public Records Request under Massachusetts Public Records Law to see if we could get some clarity on how the calculations worked. We submitted the request to the University of Massachusetts, as a public institution, for any communications between the school’s administrators and the league office concerning the Hockey East Power Index.

We received those communications this week, which provided both the formula used to calculate the HEPI as well as emails from some frustrated coaching staffs regarding the formula itself and the league’s transparency.

First thing’s first — let’s get into the formula that the league used to calculate the Hockey East Power Index.

Highlight in email below is ours:

We’ll quickly condense the formula here before going into detail:

• HEPI was calculated using Winning Percentage and Strength of Schedule, with strength of schedule defined as “Opponents’ HEPI Rating” (more on this below).
• Home/Away weighting was used in the same way the RPI uses it (weight of 1.2 for road wins and home losses, weight of 0.8 for home wins and road losses).
• The formula’s weighting is 33% Winning Percentage and 67% Opponent’s HEPI.
• Overtime wins and shootout wins both count as 0.67 wins; overtime losses and shootout losses both count as 0.33 wins.

Now for the details —

The basis of the HEPI is essentially the same thing as what the NCAA is calling the “NPI,” short for “NCAA Percentage Index.” It’s very similar to RPI, but the math is cleaner.

RPI is calculated using a team’s winning percentage (25%), opponent’s winning percentage (21%), and opponent’s opponent’s winning percentage (54%) — that latter two combined forming the team’s strength of schedule (75%). Instead of having to go down the complicated path of calculating an OppWin% and an OppOppWin%, it just replaces those components of the RPI entirely with one number: an average of your opponents ratings themselves.

We actually only know about the NPI thanks to the Women’s NCAA Summer Meetings. Page 4 of the annual meeting report breaks down how it works in detail — again, it just uses your opponent’s NPI as the strength of schedule component in the existing RPI calculation. It notes that Tim Danehy of CollegeHockeyStats.net helped to develop the formula.

Brian Smith’s email above also mentions Tim Danehy, which makes sense — Danehy helped develop NPI, and he appears to have taken the same formula and brought it over to Hockey East for HEPI.

Hockey East did make tweaks to the calculation of NPI, which is likely why they called it something else. There are a few notable differences:

First, the weights for Win% vs. SOS were changed from the RPI’s 25%/75% to 33%/67%. Brian Smith noted that the change was “due to the schedule imbalance this year,” seemingly implying that the change was made to make strength of schedule more of a factor, but in practice they actually decreased how much strength of schedule mattered from 75% to 67% compared to what’s used nationally.

By far the biggest difference between RPI and HEPI, however, is its weighting of overtime and shootout wins. In men’s hockey, with 3 on 3 overtime followed by a shootout now standardized across leagues, the NCAA counts 3 on 3 overtime wins as only 55% of a win, and treats any game going to a shootout as a tie. Hockey East, however, chose to treat its overtime wins and shootout wins the same, opting for each to count as 67% or a win — a pretty major departure from how it’s used nationally.

This disconnect in how the league treats overtime/shootout games and how the NCAA treats them seems to have caused some frustration among several coaching staffs.

It’s obvious from the email above that at least one coach — Greg Carvel of UMass — was frustrated enough that he’d tried discussing it quite often with Commissioner Metcalf. “He’s sick of hearing about it from me.”

But also, the repeated concern from Providence’s Nate Leaman that the HEPI’s ranking of a team could influence the way bubble NCAA tournament teams are perceived is very valid. We just saw last week in the women’s tournament’s selection that Minnesota clearly seemed to have been penalized (to the benefit of Minnesota-Duluth) because of how the WCHA decided to set its league standings.

Yellow text below is Commissioner Steve Metcalf:

A few notes here on this email in particular.

• It’s clear that the league took a long time to get the formula to the coaches despite some repeated requests:

Nate had asked Brian for the equation used to determine league standings. This is still a major point of contention and you’ve explained yourself enough times that I don’t need you to do it again [...] I’m really struggling with the lack of consistency between the nationally agreed upon OT percentage breakdowns and what our league decided to use. I’m not alone [...] We are still waiting on Brian to supply the equation used to determine HE standings.

• This is not exactly the response I would have given to one of my coaches if they were asking me how the league standings were going to work:

Win your games and don’t worry about this!!

• Not really related to the subject at hand, but interesting that the league made a point to schedule games that would most help its top teams:

When possible strategically schedule games that will help to teams that could use them [to help qualify for the NCAA tournament]

• It seems that at least some of the league coaches were the ones against awarding the Regular Season Championship trophy.

Carvel:

Strong sentiment that no trophy be handed out for a Regular Season Champ

Metcalf:

I’d like to award.

• It’s made clear that the coaches were not consulted on the use of HEPI, which is, honestly, just baffling:

The group is curious why the coaches were not included in any discussions about the Hockey East Ranking system and you only discussed it with ADs and voted on it without any input from the coaches?

This point was further driven home in another email from UMass’ Greg Carvel to Providence’s Nate Leaman:

All in all, it’s just bewildering the lack of transparency here. First and foremost, the HEPI formula is actually pretty reasonable! It’s nothing radical; it’s just making use of an already-existing NCAA formula using different weightings. So it doesn’t make any sense as to why the league couldn’t have just come out and made these public.

Second, it’s a surprise that the coaches were left so out of the loop on all this. Given that the formula was a mystery to the coaches for quite some time and that many didn’t like going in a different direction from the nationally agreed upon weights for overtime and shootout wins, it’s no wonder that they were the ones hesitant to award regular season titles based on a formula they weren’t consulted on and didn’t like once they saw.

In the end, there are certainly bigger things to worry about, obviously, than whether teams were seeded #5 or #6 or #7 in the conference tournament — but the main takeaway here is that hopefully the league office will opt for a bit more transparency both publicly and to its own members going forward.