jimdahl

I'm finding that they make it in slightly more scenarios in which they win the first game than the second. It's not huge, though (like a 10% difference).

I wrote a little more about what UND needs here: A more in-depth look at the at-large chances for teams on the bubble

Yeah, UND advances in about 1% of the scenarios in which they lose then tie. You'd think I'd have learned to be more careful with my language. While my notes actually say "UND can't advance with two losses" I have occasionally misspoken and said that UND needs a win. To derive the 1%, as you can see from the table here, UND advances about .5% of the time they get 0 wins. Since there are only two zero win scenarios (lose then lose, or lose then tie), we can deduce that UND makes it about 1% of the time that they lose then tie.

There are people who are sure, but as you might imagine there are just reasons they can't say so too loudly. In terms of possibilities.... A (somewhat) plain English explanation and Raw data on PWR/tournament possibilities

Looks like UND leaves the weekend #13 with a win or #17 with a loss. Would be hard, but not impossible, to climb into an at large bid from #17 without playing anymore games.

First post Congratulations ladies!

Yep, falling a bit with the win should be no surprise. If you look at the chart I posted, UND stands a decent chance of falling even with two wins. We're still in much better shape than the near catastrophic outcome we would have seen with zero wins.

Yeah, the "win 2" curve includes both of those scenarios. If they were broken out independently, they'd look about as you expect -- the "win 2 of 2" would look pretty much like the left part of the existing curve, while the "win 2 of 3" would overlap a lot but fill in some of the gap between the "win 2" and "win 1". Back when I was just doing this for UND by hand I used to break those, but now that I've automated the whole thing for all teams I just haven't accounted for those best of threes yet, since it really only comes up one week a year.

A fork in the road for UND's NCAA tournament chances It couldn't be simpler -- win this weekend puts UND above the bubble and hoping not to fall, lose this weekend puts UND below the bubble hoping for some luck. It's possible to win and then drop out (an immediate loss in the next round plus some bad luck), and it's possible to get eliminated this weekend and make it in. For info on other teams, see Final regular season PWR forecast (collegehockeyranked)

Yep, glitch with the feed from the Herald that a handful of recruiting links from February came through as dated March 3. Because they're older, the Herald charges for access to them.

I remember -- in-state tuition from 92-96 I paid something like $2000-$2500. There was ample parking, you just had to walk (e.g. usually from the lot North of Memorial Stadium), except for my classes at CAS (now Odegard, apparently) where parking was plentiful.

In the long run, yes. I don't actually see it affecting the ranking today (though I'm not really well set up to do one offs like this, so its possible I messed up).

It didn't take much digging to find out that the teams just ahead of UND in the PWR really paved the path for UND last weekend: Cornell (got swept) Vermont (got swept) Duluth (got swept) Colgate (two ties) Providence (lost and tied) KRACH had put the combined odds of enough of that happening to let UND jump to #12 around 8%.

There is definitely still beer involved.

Yep, and most surprisingly the things that I would have claimed would have helped them the most (Denver and Maine getting swept) did not occur. Not much to say other than that the 8% chance events do occur sometimes (I suppose about 1 out of 12 times, right?) I'm curious whether this "over achieving" means the 12 spot is particularly precarious, and should have that info within a day or so. If I get a chance I might try to look into why UND ended up at the upper end of its likely outcomes this week, but it wasn't immediately obvious.

The NCAA announcement said this: There are two legitimate ways to interpret that description of the weightings: apply them to the win%, or apply them to all three components (the entire game). USCHO applied the weights to just the win% (based on comments they made on their forum, primarily because that's what basketball does), whereas CHN applied them to all three components. I decided I preferred the CHN implementation, so adopted it. I think it's a very good sign that USCHO has come to agree. I don't think they would make this change, and thus reveal they were wrong, unless they had pretty good evidence that the CHN implementation was correct. So, I think we can now have pretty good confidence that this PWR matches what the committee will do.

In PWR news, USCHO seems to have changed their PWR tables to match those on this site and CHN, so all three sources should now be identical (except for the occasional data differences, particularly while game results are coming in). Nice to have that controversy cleared up, so there's no confusion which are the right numbers.

Thanks -- fixed. It's those pesky shootout wins.

While UND certainly needs to win to move on to the NCAA tournament, I think this overstates it a little. I think an at-large bid is well within reach if the boys win more than they lose.

UND still with a little more downside potential than upside. Still a very good chance of an at-large bid if UND wins more than it loses. As previously mentioned, this is using the CHN implementation (for USCHO readers, their PWR is a bit different).

CHN just finished theirs a week or two ago. But, while they were working on it they still had the old table posted, so I was warning people that their posted table was wrong. Now that they've finished it, their's seems more correct to me. In short, in trying to do it myself before USCHO or CHN did it, I found out how basketball applied the weights and did that (even though it seemed subpar). When USCHO came out with their's about a month ago, they did the same thing I had done, so I assumed it must be correct (even though it had proven pretty quirky as I was trying to implement it). Once CHN did it the other way and declared they were sure they were right, I now think USCHO just did what I did -- found out how basketball does it and assumed that must be how hockey does. But, basketball's implementation doesn't work well with the dropped games. So, I'm going to start moving entirely to the CHN calculation (though will maintain the ability to run USCHO's if we get more information in the future).

Yeah, the conference playoffs can definitely help. But, keep in mind that if you don't win it all you get at least one loss. So, if an at-large bid matters for you, you're probably adding something like 3-1 or 4-1 at best. For teams around 15 in PWR, that can be enough to push you in; but generally not if you're down around 20. (At least under the old PWR, the new one seems a little easier to climb from 20 to 13-14). For what it's worth, I like CHN's PWR slightly better than USCHO's (objective, not because of how UND performs). I've spent the last week working on duplicating it, so now should be able to run CHN's or both going forward.

UND PWR outlook with 10 games to go The "end of regular season" chart moved exactly how it should have for a split since the last one. The "one week" chart shows a lot of downside potential (which fits with the other chart's claim that UND needs to go 7/10 just to keep up with its current ranking or climb slightly). Note that I'm still using the USCHO formula, because it's the one I had working first. Still working on the CHN formula.

As I posted on collegehockeyranked, there's some new uncertainty in how to implement this year's revised RPI/PWR. CHN has published their tables for the first time, noting that they're different from everyone else's, but that they're sure they're right. I've been in communication with both them and USCHO, and I'm not sure there's going to be a quick resolution to this. So, for now, we probably shouldn't be treating anyone's PWR tables as quite so much a "fact" as we did in past years. For now, SiouxSports' tables match USCHO, though I'll definitely watch both carefully and try to do some analysis as to how different the two techniques really are.

Enjoy!