Re: Hockey East - Bye, Home, and Road: by the numbers - 2016-17 edition
Interesting times in Hockey East as we look to not only the league tourney, but also the NCAAs.
UML and BU look to have wrapped up NCAA bids, and ND and PC are trying to strengthen their positions. UVM, BC, and even NU have a shot at getting on the plus-side of the bubble without the auto-bid from the HE tourney title. Everyone else is "auto-bid or bust". That means as many as four teams could end their seasons this weekend.
While UML, BU, BC, and ND have the advantage of byes into the next round, PC, UVM, and NU have the advantage of the chance to chalk up two more wins to add to their NCAA-eligibility resumes. For ND and especially BC, playing those extra games might have been worth missing out on the bye - except, of course, winning is no guarantee. Healthy and rested also gives a better chance in the games ahead that you do play. Still, BC swapping seeds with PC might have been the best thing holistically for the league's NCAA chances.
---
On the first night of this year's HE playoffs, the only underdog in the league to score was UConn. In fact, CT put two in the net in the final 90 seconds, but the second was waved off after a long review (three men, one replay box) judged the play to be offside. That would have given a 3-2 result, with the final three goals (a 2-0 NU lead becoming a 3-2 NU final) scored without the CT goalie on the ice. Fun times.
PC, UVM, and NU all winning improves all of their PWR chances, even if NU hasn't moved yet.
---
NU not only needs their RPI to improve (which it does with the win), but also needs to find a way to flip about eight pairs to have a chance for an at-large bid. Even though they are one place behind BC in the PWR table, they are three pairs behind. NU losing the pair to Quinnipiac below them is the reason for the 3-way tie at 21st (38 pairs) - rather than NU 39, SLU 38, QU 37, NU drops one and QU rises one. It doesn't look like that pair will flip, since QU has a H2H point and is way ahead on COP. That puts NU in a deeper hole to start from. That said, they win the COP battle with the seven of the next ten teams above them, and are tied with an in-conference eighth, UVM. For the other two, AFA and UNO, there is no H2H record, so flipping the RPI flips the pair. So even from 21st place, NU's chances boil down to piling up RPI points, and hoping that other teams' losses above them will help NU gain in relative RPI. UVM dropping in relative COP to NU during the HE tourney would help, too.
---
In the reverse of the NU/QU situation, BC has an "extra" pair by virtue of their pair win over PC, 10 places above. That's why it's a three-pair gap, not one, between BC and NU. Rather than BC at 40 and NU at 39, BC wins an extra above and NU loses an extra below.
For the Eagles, all the pairs they need to flip to climb into the NCAAs are simply RPI comparisons. OSU's upset loss to MSU and UNO playing Denver (losing game one) can help, even while BC is idle. AFA and ND are also off, while UVM, SCSU, NoDak, and Wisco all won their weekend opener, making that climb a little steeper.
---
UVM's win pulls them beyond the losses of OSU and UNO and into the 16 slot. With AFA at 17 and a WCHA auto-bid to consider, that's still not "in" just yet, but it is at least on th upper side of the bubble.
---
Like BC, Notre Dame is idle. Also like BC, the Irish have the "extra" pair above them from PC, which pulls them up even with the team above in the PWR, in their case, NoDak. At the moment, ND looks to be the last at-large in the tourney, but that 14th position is vulnerable to any team (outside of one AHA and one WCHA) earning their way in via auto-bid. That positioning is also highly volatile. The RPI gap between 11 (PSU: .5513) and 20 (BC: .5345) is only 168 points - and ND is right in the thick of it (.5442). We saw that, around Beanpot time, BU and Harvard were tied in RPI, but in short order, each moved about a hundred points in opposite directions. Given how close all of those 10 teams are, and that we have a couple of weekends of end-of-season, highly-purposeful hockey on tap, it's extremely unlikely that what we're looking at now will be the final arrangement.
---
Providence still has a realistic shot at moving up into the 2nd tier, perhaps leaving HE with three 2nds if BU and UML stay in the 5-8 range as well. The loss of the BC and ND pairs is what pulls PC down from 51 pairs - and 9th place in PWR between Union and Cornell - to 49 and into a tie at 10th w/ PSU. What PC needs to get those pairs "back" and restore order in the table, is not RPI - which PC already leads - but COP. Since the HE playoffs are essentially nothing but common opponents for all three teams (other than H2H games), the thin margins in COP that sway the pairs for BC and ND could swing back to PC.
Further, in the case of ND and PC, if PC wins another UMA game, next weekend would have PC and ND meet directly in South Bend. Being that it's the playoffs, that series can't split, so someone will gain an advantage - and stop the other's season at least until the NCAAs are laid out, if not for good. If PC were to win that series 2-1, they would take the PWR pair, even if COP stayed as is, in ND's favor. (COP doesn't include games H2H as teams don't play themselves. Example: ND plays PC, but PC doesn't play itself, meaning that PC is not a common opponent for those two teams, nor is ND for the reciprocal reason.) A 2-1 PC series win would level H2H at 2-2-1, making H2H a wash, leaving COP and RPI. That means RPI, since it's the tie-breaker. PC already wins RPI and wouldn't fall behind if they won the series.
PC could possibly move on to play BC H2H, either in the semis - if BC wins and either UML or BU (but not both) loses in the quarters - or in the final if they are the last two standing. If BC loses in their quarters and PC makes the Garden, that might swing COP in PC's favor (adding four wins to BC's two losses), but I haven't done the math (see rant below).
Picking up a single pair (like ND or BC) would pull PC up into a tie at 9 with Cornell, but PC already has the pair with Cornell, so would be slotted 9th. They would need two more pairs (three total) to go their way (relative to Union) to move up to 8th, but that could also include PC swapping the pair with Union (PC +1, Union -1) and PC winning the resulting tie at 51 pairs. The four teams immediately above PC (UML, BU, Union, Cornell) are all on Byes this week, so PC will have to look to solidify their hold on 10 this weekend and use next weekend's results to shoot for the 2nd tier.
---
UML and BU are about as tied as two teams could be. They split H2H. They tied in league points. They are even in COP. They are tied in RPI to the 4th digit. But UML wins the pair. Here's why: A close look at RPI shows UML ahead by a rounding factor in the 5th digit.
In the raw RPI, UML is 20 points up: .5577 to BU's .5557. Neither had adjustments from beating very weak opponents. UML had a Quality Win Bonus of .0036 while BU's QWB was .0055 - or 19 points more. Adding UML's numbers gives .5613, but adding BU's 20-pt deficit to their 19-point advantage also gives them .5613, instead of the apparent .5612. BU's score is clearly rounding up, something like .55574 + .00554 = .56128 = .5613. This pair will be subject to butterfly effects until one starts losing - unless they both do.
---
[rant]
The main reason I haven't "done the math" is that the current COP calculus is not a simple record comparison. They do the mathematically-silly adding of winning percentages (we're all taught early on in school that adding ratios is bad math). That means one has to look at prior results and see how a win or loss (no HE playoff ties) changes that slice of the winning percentage, then add that back in. Not that that's hard, more that I rebel against the stupidity.
For example, let's look at PC's COP. I do know that, since PC already swept UMA, beating them again has no - literally zero - impact on their COP record with ND and BC. PC has already included a 1.000 record vs UMA in those COP totals. Whether that's 2-0-0, 3-0-0 or 4-0-0, it's all 1.000. So, since PC won their first UMA game (giving them 1.000 out of the gate) last weekend, winning the next two, and possibly another tonight mean absolutely nothing positive in regards to PC's COP with BC and ND. All they are doing is staving off a loss (or tie last Saturday) that would drop the record from 1.000. PC winning three straight additional games against a team that ND could only split with? (ND has the same record against UMA that they do against BC, FWIW.) No impact.
OK. I get that beating up on the bottom-feeders is why they instituted this silliness in the first place. However, let's look at another scenario.
What if PC beats UMA and ND to face top-seed and QF-winning UML in the semis at the Garden? PC also already swept UML. So if they manage to beat the league's top team, and the nation's 4/6 ranked team (poll/PWR), for a remarkable THIRD time they get NO benefit from it in COP.
However, if they were to beat BU to change their H2H record from 0-1-1 to 1-1-1, they go from .250 (1 of 4 points) to .500 - adding .250 to COP. If they face BC and move from 0-2-0 to 1-2-0, they go from .000 to .333 - all of that .333 going into COP - but not into the COP with BC.
PC went 0-1-1 (.250) against ND two weeks ago. If they meet and PC wins 2-0-0, they move to .625 (2-1-1), a COP gain of .375. If PC wins 2-1, a 2-2-1 (.500) record adds .250 to COP - but not vs. ND's COP.
At this point, all league teams are common opponents with other league teams (other than H2H), but not all of them count in your COP calculation - whether a win against the top seed or the bottom.
Sure. Makes total sense.
If you suck at math...
[/rant]
