St. Lawrence at Cornell 2023-02-18

[andyw2100]

...  Give me a big brick any day. Like that Scrivens kid.

For me Scrivens is the best Cornell goalie in the past 20 years.

He certainly knew how to have some fun in goal too.

How did I know, before clicking on the link, that that's what it would be?

[ACM]

Does anyone have a list (names) of the 1973 & 2003 players who returned this weekend?

God help me for expecting that they would be introduced, individually, by name... SMH!

I'm sure somebody does, but it's not me.

[upprdeck]

if someone was to go thru the last 2 games and stop the video as the scoring chances were showing up and ask you how many goals we would score I would think the answer would be like 10+

You could feel it sat as chance after chance nothing went in that they would lose on a simple play that led to a 2 on 1 for a tap in goal..

even the fact we lost 3 in a row with the amount of quality chances and not for a lack of effort or total lack of play on defense..

[abmarks]

Down to 15, yet still given 93% chance on chn.

They don't update that matrix in real time.   The current model now shows 50%

How does the matrix predict the likelihood of future outcomes? For example, let's say the 10th best team in KRACH plays the 25th best team. The 25th best team is at home. What odds does the model give the 25th best team to win? To my knowledge, there is no complete explanation of this anywhere. In past years, the model has yielded outcomes that are absurd. For example, a few years ago it gave each of several of the top seeds in the ECAC a >95% chance of winning their best-of-three quarterfinal series. That does not comport with the reality of hockey, which includes far too much randomness for the likelihood of a best-of-three series to be that certain. Several other similar absurdities led me to believe the model vastly overrates the likelihood of the higher ranked team winning. Which is to say, Cornell never had a 93% chance of making the NCAAs. That 93% chance was based on Cornell having an extremely high likelihood of beating Clarkson, StL, Yale, Brown, and their quarterfinal opponent.  

It is impossible to "disprove" a probabilistic model, but until the publishers of the model put forth a detailed explanation of how the predictions are calculated, I'm going to continue to believe it is unreliable.

After further review of the actual www, the replay shows the top of CHNs probability matrix page has the 3xplanation.

These are the results of 20,000 Monte Carlo simulations of the remaining games prior to Selection Day. The winner of each game in the simulation was determined randomly, weighted by KRACH. When that simulation was completed -- playing out the six conference tournaments -- a Pairwise was calculated based upon those results.

The numbers in the chart represent the percentage of times (among the total simulations run) each team placed in that spot in the Pairwise. Note that just placing in the top 16 does not indicate the team made it, due to automatic bids (AQ).

Please see below the chart for more information.

2 minute minor for not bothering to read

10 minute game misconduct for ranting after not bothering to read



Ps, 10 v 25 is hvd vs. UMD. Their respective krach numbers are 215.3 H vs 154.1 UMD.  That puts hvd at 58% to win.

Hvd Win probability = 215.3/(215.3+154.1) = 58%

[BearLover]

Down to 15, yet still given 93% chance on chn.

They don't update that matrix in real time.   The current model now shows 50%

How does the matrix predict the likelihood of future outcomes? For example, let's say the 10th best team in KRACH plays the 25th best team. The 25th best team is at home. What odds does the model give the 25th best team to win? To my knowledge, there is no complete explanation of this anywhere. In past years, the model has yielded outcomes that are absurd. For example, a few years ago it gave each of several of the top seeds in the ECAC a >95% chance of winning their best-of-three quarterfinal series. That does not comport with the reality of hockey, which includes far too much randomness for the likelihood of a best-of-three series to be that certain. Several other similar absurdities led me to believe the model vastly overrates the likelihood of the higher ranked team winning. Which is to say, Cornell never had a 93% chance of making the NCAAs. That 93% chance was based on Cornell having an extremely high likelihood of beating Clarkson, StL, Yale, Brown, and their quarterfinal opponent.  

It is impossible to "disprove" a probabilistic model, but until the publishers of the model put forth a detailed explanation of how the predictions are calculated, I'm going to continue to believe it is unreliable.

After further review of the actual www, the replay shows the top of CHNs probability matrix page has the 3xplanation.

These are the results of 20,000 Monte Carlo simulations of the remaining games prior to Selection Day. The winner of each game in the simulation was determined randomly, weighted by KRACH. When that simulation was completed -- playing out the six conference tournaments -- a Pairwise was calculated based upon those results.

The numbers in the chart represent the percentage of times (among the total simulations run) each team placed in that spot in the Pairwise. Note that just placing in the top 16 does not indicate the team made it, due to automatic bids (AQ).

Please see below the chart for more information.

2 minute minor for not bothering to read

10 minute game misconduct for ranting after not bothering to read



Ps, 10 v 25 is hvd vs. UMD. Their respective krach numbers are 215.3 H vs 154.1 UMD.  That puts hvd at 58% to win.

Hvd Win probability = 215.3/(215.3+154.1) = 58%

I literally wrote my above post immediately after reading the exact web page you're quoting, because it's not at all clear what "weighted by KRACH" means. In addition to CHN, there is also the playoffstatus.com website which to my knowledge offers no explanation either.

Also, what you just said is incorrect. Harvard is 14th in KRACH and UMD is 19th. My above post asks about 10th in KRACH vs 25th in KRACH. (Notice that I specifically asked about KRACH because right before posting I read the exact webpage you accused me of not reading.)

But let's assume for the sake of argument that you're correct, and your interpretation of "weighted by KRACH" is accurate. In that case, you've proven my point that this is an absolutely abysmal way of predicting future outcomes. Let's say #1 Minnesota plays #60 Dartmouth. Per your formula, that gives Minnesota a 570.9/(570.9+25.7)=96% chance of winning, which is completely absurd.

By the way, per this method, Minnesota's change of beating Cornell? 78%. This is one of many insane results one derives from abmarks' interpretation of "weighted by KRACH." I hope for the sake of the model that this interpretation is NOT true. Repeatedly overweighting the better team's chance of winning (which has been an obvious problem with these models for years) yields ridiculous outcomes like saying Cornell had a 93% chance of making the NCAA going into last weekend.

[CU77]

Yes, this is what "weighted by KRACH" means, and yes, KRACH greatly overestimates the win probability of the higher-rated team.

[upprdeck]

how many times do we think Minn would beat Dartmouth if they played 100x.. not 96 for sure. But given OT rules so no ties they probably win 70-80? Given that why would any formula put them at 96%

[Dafatone]

how many times do we think Minn would beat Dartmouth if they played 100x.. not 96 for sure. But given OT rules so no ties they probably win 70-80? Given that why would any formula put them at 96%

I'd generally be okay with assuming that the best team would beat the worst team 90% or more, though maybe not 96%.

I guess Dartmouth is the 2nd worst ahead of Lindenwood. And this year is a bit of an exception because I think Dartmouth is a good deal better than their record based on goal differential.

[Trotsky]

CHA's list of Seniors told Muzyka to pound sand.  Is he off the roster because he entered the transfer portal for 2024?  He hung on in practice for all four years; too bad.

[arugula]

https://theathletic.com/4213022/2023/02/21/college-chl-european-hockey-free-agents/?source=dailyemail&campaign=601983

Malinski and Mitchell mentioned as possible NHL level players, Steinburg mentioned as possibly a free agent this summer

[Trotsky]

https://theathletic.com/4213022/2023/02/21/college-chl-european-hockey-free-agents/?source=dailyemail&campaign=601983

Malinski and Mitchell mentioned as possible NHL level players, Steinburg mentioned as possibly a free agent this summer

Paywall = nope.

[The Rancor]

https://theathletic.com/4213022/2023/02/21/college-chl-european-hockey-free-agents/?source=dailyemail&campaign=601983

Malinski and Mitchell mentioned as possible NHL level players, Steinburg mentioned as possibly a free agent this summer

Paywall = nope.

That I pay for NYT and still have an additional paywall for sports is offensive.

[Weder]

2. Sam Malinski, RHD, Cornell-ECAC: Malinski has been a top defenseman in college this season. He has excellent playmaking ability. He makes a lot of difficult passes in the offensive zone and has the ability to run a pro power play with his vision and shot. His skating is solid too and he can create with his skill and feet. Malinski isn't the biggest defender but he competes well enough that I think he has a shot to be a good pro.

12. Travis Mitchell, LHD, Cornell-ECAC: Mitchell has played a large role for Cornell this season. The senior defenseman and team captain often lined up on both special teams. He's a big-body defenseman who competes well and could be a solid pro defender. His skating is fine for his size but not a major selling point. Whether he can move pucks versus men will be Mitchell's main challenge as I wouldn't call him a natural offensive type.

[RichH]

https://theathletic.com/4213022/2023/02/21/college-chl-european-hockey-free-agents/?source=dailyemail&campaign=601983

Malinski and Mitchell mentioned as possible NHL level players, Steinburg mentioned as possibly a free agent this summer

Paywall = nope.

That I pay for NYT and still have an additional paywall for sports is offensive.

Here are two models:

Cable industry: charge all subscribers a fee to cover ESPN and regional sport networks licensing costs. Some watch. A lot don't.

NYT: Basic subscription for the news/opinion. Add-on model for a well-developed sports writing site, cooking database, games, and shopping review site. All are available a la carte. If you want sports coverage, you can get it without penalizing those that don't.

Which model is more offensive?

[upprdeck]

the latest NYT subscription i saw include the athletic.