National Semifinal: Cornell vs. Mercyhurst

[LakersFan]

To start this thread, here are some key numbers. On paper, these teams look pretty close, except for one key number.

Head to Head

10/23/09 Mercyhurst 4 Cornell 1
10/24/09 Mercyhurst 4 Cornell 1

Statistics (Ranks only include 33 teams)

Scoring
Mercyhurst 4.74 goals/game 1st (Z-score=3.35)
Cornell 2.88 goals/game 8th (Z-score=0.62)

Defense
Mercyhurst 1.60 goals/game 3rd (Z-score=-1.39)
Cornell 1.71 goals/game 6th (Z-score=-1.20)

Power Play
Mercyhurst 24.5% (2 short handed against)
Cornell 21.8% (3 short handed against)

Penalty kill
Cornell 93.2% (2 short handed for)
Mercyhurst 91.3% (21 short handed for, not a typo)

Z-score: the number of standard deviations away from the mean the observation is. Mercyhurt scores 3.35 standard deviations (.68 gpals/game) more goals per game than the DI average.

 

[LakersFan]

Re: National Semifinal: Cornell vs. Mercyhurst

Common Opponents

(L indicates a loss, T a tie)

Mercyhurst                              Cornell

10/16/09  MC 1 RPI 0                    11/7/09  C 3 RPI 1
10/17/09  MC 5 RPI 1                    2/19/09  C 2 RPI 1
                                        3/5/09   C 5 RPI 4

11/20/09  MC 2 NU 2                     11/28/09 C 2 NU 1
11/21/09  MC 2 NU 0                     11/29/09 C 1 NU 2 L
1/29/10   MC 2 NU 0
1/30/10   MC 1 NU 2 L

11/27/09  MC 8 Colgate 3                1/22/10 C 3 Colgate 3 T
11/28/09  MC 8 Colgate 2                1/23/10 C 6 Colgate 0
                                        2/26/10 C 2 Colgate 1
                                        2/27/10 C 5 Colgate 0

12/11/09  MC 4 SLU 2                    11/14/09 C 2 SLU 3 L
12/12/09  MC 6 SLU 2                    1/29/10  C 2 SLU 2

1/15/10   MC 6 Clarkson 3               11/13/09 C 2 Clarkson 0
1/16/10   MC 5 Clarkson 5 T             1/30/10  C 1 Clarkson 2 L
                                        3/7/10   C 4 Clarkson 3

1/22/10   MC 8 SU 1                     2/5/10   C 1 SU 2 L
1/23/10   MC 4 SU 3
2/26/10   MC 4 SU 1
2/27/10   MC 3 SU 1
3/6/10    MC 2 SU 1

 

[LakersFan]

Re: National Semifinal: Cornell vs. Mercyhurst

Both teams are on long win streaks.

Mercyhurst Last 10

Team                    Rutter Ranking

Robert Morris       4- 1     -0.477
Robert Morris       4- 0     -0.477
Robert Morris       9- 2     -0.477
Wayne State         8- 0     -0.304
Wayne State         4- 1     -0.304
Syracuse            4- 1     0.041
Syracuse            3- 1     0.041
Robert Morris       7- 1    -0.477
Syracuse            3- 1     0.041
Boston Univ.        4- 1     0.274

Average Rutter Ranking = -0.212

Cornell Last 10

Team               Rutter Ranking

PRINCETON    6- 0    -0.021
Brown        4- 0    -1.024
Yale         1- 0    -0.389
RENSSELAER   2- 1     0.053
UNION        6- 1    -1.674
COLGATE      2- 1    -0.224
COLGATE      5- 0    -0.224
RENSSELAER   5- 4     0.053
CLARKSON     4- 3     0.546
Harvard      6- 2     0.670

Average Rutter Ranking = -0.223

Toughest Eight Opponents (By Rutter Ranking)

Mercyhurst

MINNESOTA DULUTH       3- 4     0.982 L
MINNESOTA DULUTH       5- 2     0.982
CLARKSON               5- 5     0.546 T
CLARKSON               6- 3     0.546
Cornell                4- 1     0.464
Cornell                4- 1     0.464
Boston University      4- 1     0.274
St. Cloud State        4- 4     0.243 T

Cornell

MERCYHURST             1- 4     1.406 L
MERCYHURST             1- 4     1.406 L
Harvard                6- 2     0.670
Harvard                4- 4     0.670 T
HARVARD                4- 3     0.670
CLARKSON               4- 3     0.546
Clarkson               1- 2     0.546 L
CLARKSON               2- 0     0.546

 

[dave1381]

Re: National Semifinal: Cornell vs. Mercyhurst

I'll add this from the Harvard-Cornell thread

I think you sell Cornell a bit short. Certainly they're an underdog, but they'll come in the Frozen Four much more battle-tested than Mercyhurst has been over the past month.

I'd give Cornell and better chance of beating Mercyhurst than any of the 4-5 seeds against the No. 1 seed, except for probably Brown over Minnesota in 2002 and Minnesota over UNH in 2006 (and I'm not just saying that because those were the teams that actually won). The No. 1 seed has lost in the semis in every Olympic year.

Here's some more encouraging stats: Cornell has just laid 10 goals on two of the best defensive teams in the country, Harvard and Clarkson (sure, neither is what they were earlier this season, but still).

Here is how Mercyhurst has done against the top offensive teams it has played, aside from your season-opening series: (I go by WCHODR rankings)
No. 3 UMD: 2 goals, 4 goals
No. 12 St. Cloud: 4 goals
No. 13 Clarkson: 3 goals, 5 goals
No. 14 Colgate: 3 goals, 2 goals

The WCHODR model expects a 3-2 Mercyhurst win. I expect that could be a fun higher-scoring game than that. Cornell clearly knows how to finish. Mercyhurst hasn't faced a team with the kind of weapons Cornell has since mid-January.

 

[dave1381]

Re: National Semifinal: Cornell vs. Mercyhurst

Of course the idea that Cornell is more battle-tested coming on will need some defending since the stats Lakersfan yield the opposite conclusion.

The last two weeks of RPI/Clarkson/Harvard are certainly a better test than RMU/Syracuse/BU. I don't think having played Brown or Union in the past month puts Cornell at any disadvantage.

 

[dave1381]

Re: National Semifinal: Cornell vs. Mercyhurst

Also, Lakersfan noted in the Mercyhurst thread that most teams simply can't handle Mercyhurst's pressure. It is worth noting that Harvard certainly tried to put a lot of pressure on Cornell in the beginning, and Cornell just totally burned them and the game was pretty much over. Now Mercyhurst is certainly much more talented than Harvard, but it is worth noting this Cornell team has finished well lately, beating two top 10 teams while being badly outshot.

I'm expecting this game to be pretty exciting -- on par with Minnesota-UNH four years ago.

 

[ARM]

Re: National Semifinal: Cornell vs. Mercyhurst

The last two weeks of RPI/Clarkson/Harvard are certainly a better test than RMU/Syracuse/BU. I don't think having played Brown or Union in the past month puts Cornell at any disadvantage.

This seems to be one of those years where every opponent is flawed in some way, but it is nice if that opponent can at least present challenges to an NCAA team along the way. Clearly there is no other offense out there comparable to the Lakers. Cornell has at least seen MC, although something more akin to a beta-test model. In their last two games, the Big Red have faced size/strength (Clarkson) and speed (Harvard) equal to anything they are likely to see the rest of the way. Granted, Mercyhurst may use their speed more effectively than do the Crimson, but at least it gets a team accustomed to playing at a high tempo. Beyond RPI, I don't know how much grooming they got in the rest of their Last 10.

Similarly, the Lakers prep games in their Last 10 are probably limited to Syracuse and BU. I'm sure they'd have liked to have faced another top opponent down the stretch, but that's been a traditional problem. They overcame it last year, and this year, there is a much smaller game between a Syracuse and an NCAA team.

 

[dave1381]

Re: National Semifinal: Cornell vs. Mercyhurst

Similarly, the Lakers prep games in their Last 10 are probably limited to Syracuse and BU. I'm sure they'd have liked to have faced another top opponent down the stretch, but that's been a traditional problem. They overcame it last year, and this year, there is a much smaller game between a Syracuse and an NCAA team.

Understood, and I'm not by any means saying I expect Cornell to win .

A lot of people will just think Cornell has no chance, given their season opening results against Mercyhurst and the fact Cornell was under-the-radar most of the season.

I'm saying the last two weeks of preparation will be a factor that will make the game more even than it otherwise would be.

 

[LakersFan]

Re: National Semifinal: Cornell vs. Mercyhurst

A lot of people will just think Cornell has no chance, given their season opening results against Mercyhurst and the fact Cornell was under-the-radar most of the season.

I'm saying the last two weeks of preparation will be a factor that will make the game more even than it otherwise would be.

My initial motivation for my data collection was to combat the argument "Cornell is on a 10 game win streak against better competition," which is clearly not the case. The last three games have been against better competition, and as Dave says, that will help Cornell.

Cornell looks a lot better when you dive into the numbers to someone outside the ECAC. In my rankings, the did edge up into the top 3 for a moment, before losing to SLU and having a run of sub-par performances. Hurst and Cornell have similar records against common opponents, except St. Lawrence (Cornell only played one game against Syracuse).

According to my rankings, Mercyhurst should win this game 81.8% of the time. I think that is a fair number when you look at the differences in offensive production and the fact that Mercyhurst has beaten them twice (after which Cornell beat Harvard and Clarkson over the next 14 days, which to me suggests that Cornell was pretty good to start the season).

 

[ARM]

Re: National Semifinal: Cornell vs. Mercyhurst

A lot of people will just think Cornell has no chance, given their season opening results against Mercyhurst and the fact Cornell was under-the-radar most of the season.

Personally, I don't think the opening series has any relevance beyond both getting a first-hand look at the other team's personnel. So many of Cornell's impact players are young, and even veteran teams evolve over 5 months.

One intangible -- which team performs better in any empty arena? There were few fans in Minneapolis yesterday, even fewer in Duluth, so hopefully, there are some hardcore fans looking to head cross country to support these teams that drew well in their final home games.

 

[Trillium]

Re: National Semifinal: Cornell vs. Mercyhurst

One intangible -- which team performs better in any empty arena? There were few fans in Minneapolis yesterday, even fewer in Duluth, so hopefully, there are some hardcore fans looking to head cross country to support these teams that drew well in their final home games.

That surprises me--I thought the crowds in Minnesota were supposedly the best in D1?

 

[Skate79]

Re: National Semifinal: Cornell vs. Mercyhurst

I don't think looking back to October does either team any good at this point because they both have made significant progress since their opening series. They're different teams now and physically are more spent than last October. Conditioning will play a role especially in the later stages of the game.

One of the keys will be penalties. Cornell is dangerous on the shorthand and the Lakers will have to be wary of this. Assuming both goalies are on, a key turnover, penalty or losing a faceoff can be the difference between winning and losing. Mercyhurst has the experience of being in this situation. This is a first for Cornell and it will be interesting to see how their young players handle it.

 

[ARM]

Re: National Semifinal: Cornell vs. Mercyhurst

Cornell is dangerous on the shorthand and the Lakers will have to be wary of this.

According to the record book, the converse is also true.

 

[dhmn]

Re: National Semifinal: Cornell vs. Mercyhurst

That surprises me--I thought the crowds in Minnesota were supposedly the best in D1?

The boys high school hockey tournament championship games Saturday *HOPEFULLY* had something to do with the pathetic turnout for NCAA games... a team from the Duluth area played in the small school final in the afternoon while two cities teams played the large school final.. in the same time frame as the NCAA games in their area would be. While I wish that was the actual reason it's hard to really know.

 

[ARM]

Re: National Semifinal: Cornell vs. Mercyhurst

Unless my memory or 'rithmetic is off, Cornell becomes the 12th different program to advance to the Frozen Four in the event's 10-year history.

Frozen Four Appearances
# Team (Years)
7 Minnesota-Duluth (2001,2002,2003,2007,2008,2009,2010)
7 Minnesota (2002,2003,2004,2005,2006,2009,2010)
5 Harvard (2001,2003,2004,2005,2008)
5 St. Lawrence (2001,2004,2005,2006,2007)
4 Dartmouth (2001,2003,2004,2005)
4 Wisconsin (2006,2007,2008,2009)
2 Mercyhurst (2009,2010)
2 New Hampshire (2006,2008)
1 Boston College (2007)
1 Brown (2002)
1 Cornell (2010)
1 Niagara (2002)

 

[D2D]

Re: National Semifinal: Cornell vs. Mercyhurst

I can tell you that the high school tourney definitely has an impact on attendance for the Gophers mens team when they (usually) host the first round of the WCHA playoffs every year. I always thought it was a shame they had to be on the same weekend, but there's not much either can do about it.

Last night the Edina-Minnetonka championship game drew more than 19,000 potential hockey fans away from the Gopher-Clarkson womens game, but of those I would guess fewer than 1,000 would fall into the real potential category.

 

[dave1381]

Re: National Semifinal: Cornell vs. Mercyhurst

According to my rankings, Mercyhurst should win this game 81.8% of the time. I think that is a fair number when you look at the differences in offensive production and the fact that Mercyhurst has beaten them twice (after which Cornell beat Harvard and Clarkson over the next 14 days, which to me suggests that Cornell was pretty good to start the season).

Okay, I would disagree that 81.8% is a fair number. So KRACH has it at 80.2%, so even distributional assumptions can get Cornell another 1.6%.

If you think of all the model doesn't take into account, do you think the potential sources of bias all cancel each other out? Cornell is certainly the team whose results on the margin were most affected by U-22s. Cornell is certainly looking better in its last 3 games than it did all season, while I think Mercyhurst's performance has been steady and constant.

Yes, we don't have enough of a sample size to determine whether Cornell's three recent games mark a genuine improvement or good luck, but either way, 81.8% seems high to me.

Personally, I think the Mercyhurst home games against UMD and Clarkson are the closest thing you have to an approximation of the relative abilities of the teams playing today (I'll even ignore home ice advantage). Mercyhurst went 2-1-1 in those games, a 62.5% win percentage. I think the truth is closer to that than to 80%. How critical really is the information of Mercyhurst and Cornell's relative performance against 2nd-tier teams (counting Cornell opening weekend as a second-tier team ) in terms of predicting their performance against each other?

 

[dave1381]

Re: National Semifinal: Cornell vs. Mercyhurst

Actually, I'd be very interested to know what the estimated probability of Mercyhurst winning is IF you pretend the following five games never happened.
-- the first two Mercyhurst-Cornell games
-- the three U-22 games (Syracuse & Providence X 2 )
which makes Cornell a 20-3-6 team, much better than a 20-8-6 team

Now I think we'd agree it's an extreme assumption to say none of these games give information about the Mercyhurst-Cornell result this weekend, but I think this gives us a lower bound on Mercyhurst's probability of winning based on the kind of arguments I'm making.

 

[ARM]

Re: National Semifinal: Cornell vs. Mercyhurst

Actually, I'd be very interested to know what the estimated probability of Mercyhurst winning is IF you pretend the following five games never happened.
-- the first two Mercyhurst-Cornell games
-- the three U-22 games (Syracuse & Providence X 2 )
which makes Cornell a 20-3-6 team, much better than a 20-8-6 team

Now I think we'd agree it's an extreme assumption to say none of these games give information about the Mercyhurst-Cornell result this weekend, but I think this gives us a lower bound on Mercyhurst's probability of winning based on the kind of arguments I'm making.

I think it is valid when trying to use data to predict future results. It isn't purely math at this point, but math combined with science. If this was an analysis for a chemistry reaction, and we discovered that some of our data was based on the use of a watered-down solution, we'd throw that data away. And it does make a difference when top players are missing, as pointed out below:

St. Cloud vs. Mercyhurst.....Huskies definitely gave the Lakers a great game. Worth noting however, is that with 5 minutes remaining in the second period, the Laker line of Bram, Scanzano and Bendus left the ice and headed for the airport to catch up with the Canadian U-22 team. Mercyhurst was ahead 4-3 at the time of their departure.

Gutsy move by St. Cloud in the thrid period as they had a 5-3 power play for 50 seconds before the Lakers would return one of two in the box, and in the early seconds of the power play they pulled their goalie creating a 6 on 3 and got the game tying goal to make it 4-4.

 

[DC78-82]

Re: National Semifinal: Cornell vs. Mercyhurst

The boys high school hockey tournament championship games Saturday *HOPEFULLY* had something to do with the pathetic turnout for NCAA games... a team from the Duluth area played in the small school final in the afternoon while two cities teams played the large school final.. in the same time frame as the NCAA games in their area would be. While I wish that was the actual reason it's hard to really know.

Those of us who watched Saturday's UMD - UNH game online know all about that boys tourney. It was pretty much all the announcers talked about during the women's game