ELynah Forum

Quote from: BearLover on March 29, 2026, 11:20:28 PM

Quote from: Tom Lento on March 29, 2026, 11:14:19 PM

Quote from: BearLover on March 29, 2026, 01:55:24 PM

Quote from: Tom Lento on March 29, 2026, 01:47:10 PM

Quote from: BearLover on March 28, 2026, 10:03:58 AMHarvard/Dartmouth was their best weekend for sure, but it was also their only great weekend the entire season. This never felt like a top 10 team. That's how the computer rankings shook out, though. KRACH has us down at 12 but that's still really good. I don't believe our underlying metrics were that great even if our results (wins and losses, accounting for SOS) were very good.

I haven't looked since a few weeks before the end of the RS, but at least in aggregate Cornell's advanced metrics seemed to me to be in line with a solidly top 15 team. Top 10 in the ratings was maybe lucky, but not egregiously so.

If you've gone through the numbers a bit more it'd be interesting to get more detail there. I didn't get to watch the team much this season but I got the feeling from following along here and checking some xG/Corsi/Fenwick data that Cornell's play style results in more possession than it seems based on the eye test.

That may be because they've got a metrics-sound approach that doesn't yield enough sustained pressure (from what I understand Corsi and Fenwick, in particular, can suffer from this), or it may be because everybody around here got used to watching Schafer's endless cycle approach to puck possession, or it may just be noise.

If you're looking at the publicly available data on eg. CHN, those aren't adjusted for SOS. So top 15 possession metrics in the ECAC is in reality worse than top 15. I don't have access to anything better than what's publicly available, and even the CHN xG stat is not precise as it does not take into account type of shot, so I can't really comment with any greater degree of certainty.

Yeah that's a good point. I don't really know how to discount the stats accordingly so unless someone wants to do a more rigorous analysis this will all be on vibes.

My take after a quick look at today's Corsi/Fenwick numbers is still that Cornell (~12th on aggregate possession metrics) was a solid top 15 but probably not a top 10. That's not bad for a rebuilding year, and getting that big incoming class an up close look at a title contender is a real positive even if the game itself was pretty rough.

Cornell had a below average SOS. I think last I looked it was like 34th, but I can't seem to find it now on CHN. I'm not sure how you account for that exactly, but if a team that's eg. 16th in possession metrics has the 5th hardest SOS, they should be ahead of Cornell. @sezenack is an RPI fan who maintains a blog and I think he was tracking SOS-adjusted possession stats. @ursusminor, perhaps you can ping him? Would be interesting to see.

I'm just seeing this now. Here's what the numbers were before the NCAA tournament if you're still curious https://docs.google.com/spreadsheets/d/1gbNvoRrsMyXcuE3uk6Bnou_LkBzQvXG6rUxzMYHh9J8/edit?usp=sharing

Quote from: Trotsky on April 03, 2026, 02:50:19 PM

Quote from: ursusminor on April 03, 2026, 02:27:39 PMThis makes me even happier that the currently known recruit list for 2026 is IMHO significantly better than Cornell's.

We look forward to their transfers.

No, I am not that much of a jerk.

Gonna join in to say RPI doesn't have anyone worth poaching from last year's team LOL. Regardless, I've been told that our only departures this year aren't leaving because they want to...

Quote from: BearLover on November 19, 2025, 08:43:07 PMhttps://www.rpifieldhouse.com/p/ecac-thoughts-so-far?utm_campaign=post&utm_medium=web

More great content from sezenack. Love any piece that makes a real attempt to use underlying metrics to evaluate players or teams. I particularly liked his model of Corsi adjusted for SOS. You can find possession metrics on the internet, but they're not very meaningful  unless adjusted for the huge disparities in college teams' SOS.

Agreed on pumping the breaks on Dartmouth. I think though he is too high on Cornell and the ECAC in general. We'll see what happens, but Cornell has looked shaky so far and if they're indeed the second best team in the league (which sounds about right), then this doesn't look to me like a 3-bid league. Also don't really get the point about Union screwing itself with its schedule-the whole point of the NPI/Pairwise is to adjust for this.

I think being too optimistic on the ECAC is a fair critique of mine. You guys definitely see more of Cornell than I do; I mostly just surf around the various games when I'm not watching RPI.

The point with Union is that they have no shot of getting an at large bid because their schedule is too weak, but I think they could be in the conversation with a more challenging schedule. They have largely dominating their weak schedule, and I think they're capable of beating some good teams that would get them higher in the NPI. But they scheduled weakly, so their margin is too slim to realistically get an at large bid. Hope that explains it a little better

Posted my yearly ECAC preview this morning. I have Cornell 3rd, not as much scoring depth as past years but love the defense and Cournoyer in the net. Hope you all enjoy!

https://www.rpifieldhouse.com/p/ecac-preview-2025-26

Quote from: TrotskyTBRW Predictions:

Code Select Expand


 1  Quinnipiac
 2  [color=#b31313]Cornell[/color]
 3  Harvard
 4  Clarkson
 5  Colgate
 6  St. Lawrence
 7  Yale
 8  Union
 9  RPI
10  Princeton
11  Brown
    Dartmouth


I'll declare Keith Allain the second coming of Christ if Yale finishes 7th in the league

Quote from: BearLover

Quote from: sezenack

Quote from: BearLover

Quote from: ursusminorAn RPI fan and his computer program's ECAC predictions. https://thefieldhouse.substack.com/p/ecac-2023-24-preview?utm_source=profile&utm_medium=reader2

Cornell is #2, and Harvard is #7 below #4 Princeton.

I appreciate that someone is trying to quantify these things rather than making prediction based on feel, but the fact this model ranks Harvard's returning offense as 12th in the league evidences a clear flaw in its methodology, in my opinion. I would guess the issue is that Harvard's top players last year were SO good that they got all the PP time and played a crazy amount of minutes, especially when Harvard needed a goal. So yeah, nearly all the offensive production came from those players, who are gone now. But that doesn't mean the players who played behind them, and will now take on a larger role, are not good players—it just means they didn't add much offensively last year, because they didn't need to.

And though I wish it were true, I just don't see Cornell as an elite team this season. We could compete for an NCAA spot, but if that happens it is most likely as one of the final at-large bids. We have a ton of freshmen, no one who has played more than two years for Cornell (and no one aside from Seger who has played more than two years of college hockey at all). We are still going up against much older and more experienced teams whenever we play a non-Ivy. I like our trajectory but don't expect us to be elite until the fifth-year eligibility rule finally goes away in a couple seasons.

ursusminor told me he posted this here, and I saw this and wanted to reply.

Harvard's 12th ranked returning offense isn't an opinion; it's a fact. They return 122 points and played 34 games, so they return 3.59 points per game, which is last in the league. It's an entirely objective metric. I disagree that it's a flaw in the methodology and think it's much more of a symptom of the lack of college hockey data available to the public (mainly time on ice).

However, I do agree with your points about PP time and general time on ice. Harvard does have good talent that played lower in the lineup that will likely step up. It's entirely possible they beat the projection, and I'd say they are the likeliest team to do that of the 12. Anyone expecting them to maintain their level of last season is mistaken though. Looking at the roster, the only player I see surpassing 30 points is Joe Miller. None of the players that remain are at the level of Coronato, Farrell, or Laferriere. None of the defensemen are at the level of Thrun even though Ian Moore is very good. Those are the players they have to replace, and Harvard isn't equipped to do it regardless of ice time. The players they have are not that caliber, where they put up big numbers and get an NHL deal following the season (with the exception of Miller and Moore).

Sorry if my post came off as overly critical. I do appreciate people trying to quantify this stuff, as I said. And I enjoyed reading your findings and write-up. I also certainly didn't mean to imply that you were putting your thumb on the scale. I understand these metrics are objective and that Harvard's returning points per game number is what it is.

However, the fact you chose to base offensive ranking entirely (mostly?) off of returning points per game is not objective. While the inputs themselves may be objective metrics, choosing which inputs to use is necessarily subjective. I don't deny that you are limited in your data and maybe this is the best you can do; but it's very clear if you look past this one statistic that Harvard is once again going to be very talented this season. They still have by far the most draft picks in the ECAC, by the far most players who were in the central scouting rankings, etc. These are imperfect metrics, but they're other metrics that you could consider factoring into your analysis.

By the way, how do you measure incoming class rank? I may have missed that, but I didn't see it in your post. Cornell's incoming class consists of five draft picks, which I believe is easily the most in the ECAC, but its incoming class is ranked #7.

It didn't come off overly critical; it just seemed like you thought returning scoring was a subjective metric so wanted to make sure you knew it wasn't.

All good points and questions here! I'll do my best to address them all.

Offensive ranking isn't based on mostly returning scoring. Offensive projections are done by essentially taking the best 12 forwards and 7 best dmen projected by the model and adding up their offensive projections. If that means it's the entire recruiting class, that's what is used. If the recruits are bad (SLU), it's all returners used.

I agree Harvard will be talented but talent doesn't always equal wins. Quinnipiac had one draft pick last year and won the natty. For all of Harvard's draft picks, they have been finishing behind Quinnipiac and sometimes Cornell and Clarkson as well. I also don't think all draft picks are created equal. Huge difference between someone like Sean Farrell and Zakary Karpa. For what it's worth, I did consider including draft picks as part of it, and I felt like it made the projections worse. There are too many nepotism picks and picks that are clearly just bad like Misha Song that dilutes the value they could add. The really talented players will show it with their play, which the model will factor in.

As for the recruiting questions, I tried to address this a bit in the article, but I wouldn't think of that like a true recruiting class ranking. It's not rating the total talent of the recruiting class. It is strictly based on offensive projections, and it is not total offense but average offensive projection for the incoming skaters. So for Cornell, they get hurt in the rankings due to bringing more defensemen since defensemen get projected for less offense than forwards.

Overall differences from Neutral Zone: only offense not overall talent, averaged to account for size of class (Neutral Zone doesn't go by average), includes transfer portal additions unlike Neutral Zone

I also want to make it clear that the recruiting ranking doesn't directly go into the total projections; it's just cosmetic to add some context to the projections. The overall offensive projections work as I said above by creating a team's best lineup based on the projections, so Cornell's strong recruiting class with mostly defensemen gets factored in properly overall.

Quote from: BearLover

Quote from: ursusminorAn RPI fan and his computer program's ECAC predictions. https://thefieldhouse.substack.com/p/ecac-2023-24-preview?utm_source=profile&utm_medium=reader2

Cornell is #2, and Harvard is #7 below #4 Princeton.

I appreciate that someone is trying to quantify these things rather than making prediction based on feel, but the fact this model ranks Harvard's returning offense as 12th in the league evidences a clear flaw in its methodology, in my opinion. I would guess the issue is that Harvard's top players last year were SO good that they got all the PP time and played a crazy amount of minutes, especially when Harvard needed a goal. So yeah, nearly all the offensive production came from those players, who are gone now. But that doesn't mean the players who played behind them, and will now take on a larger role, are not good players—it just means they didn't add much offensively last year, because they didn't need to.

And though I wish it were true, I just don't see Cornell as an elite team this season. We could compete for an NCAA spot, but if that happens it is most likely as one of the final at-large bids. We have a ton of freshmen, no one who has played more than two years for Cornell (and no one aside from Seger who has played more than two years of college hockey at all). We are still going up against much older and more experienced teams whenever we play a non-Ivy. I like our trajectory but don't expect us to be elite until the fifth-year eligibility rule finally goes away in a couple seasons.

ursusminor told me he posted this here, and I saw this and wanted to reply.

Harvard's 12th ranked returning offense isn't an opinion; it's a fact. They return 122 points and played 34 games, so they return 3.59 points per game, which is last in the league. It's an entirely objective metric. I disagree that it's a flaw in the methodology and think it's much more of a symptom of the lack of college hockey data available to the public (mainly time on ice).

However, I do agree with your points about PP time and general time on ice. Harvard does have good talent that played lower in the lineup that will likely step up. It's entirely possible they beat the projection, and I'd say they are the likeliest team to do that of the 12. Anyone expecting them to maintain their level of last season is mistaken though. Looking at the roster, the only player I see surpassing 30 points is Joe Miller. None of the players that remain are at the level of Coronato, Farrell, or Laferriere. None of the defensemen are at the level of Thrun even though Ian Moore is very good. Those are the players they have to replace, and Harvard isn't equipped to do it regardless of ice time. The players they have are not that caliber, where they put up big numbers and get an NHL deal following the season (with the exception of Miller and Moore).