Tag: PWR forecast

Weekly UND PWR outlook

In this idle weekend, #2 UND (PairWise Rankings) is most likely to keep its #2 ranking, but if they move it will probably be down to #3.

UND_Feb9

 

That .7% chance of UND taking #1 is largely the likelihood of Alaska-Anchorage sweeping Minnesota State (KRACH gives Anchorage a ranking of 53.43 and Minnesota State a ranking of 613.98, thus predicting Alaska-Anchorage has an 8% chance of winning each game, or about a .6% chance of winning both).

To see the threat of falling to #3, UND need only look at current #3 Boston University. UND currently wins the comparison 2-0, but on a razor thin .5856-.5845 RPI advantage. Even a single BU win (and a few other things going right) could be enough for BU to overtake UND.

End of regular season outlook

The regular season outlook has improved to the point that UND is nearly a lock for an at-large bid. Only an epic collapse (e.g. 2 or fewer wins in the final 8 games) combined with a disastrous conference tournament could knock UND out.

UND_endofseason

Can UND take over #1 this weekend?

It’s a bit of a longshot, but if everything goes well for UND this weekend, they could climb to #1 in the PairWise Rankings. Incumbent Minnesota State controls it’s own destiny — a sweep guarantees they stay atop the heap for one more week. A split, however, opens the door for UND.

mankato

If UND sweeps, they can take #1 with a suitably bad performance by Mankato. It’s remotely possible, though not likely, for North Dakota to take #1 even with a split.

NorthDakota

Finally, it’s remotely possible (about 2%) for Nebraska-Omaha to take the #1 spot this weekend (a scenario that would almost certainly require a sweep of UND), and extremely remotely possible (about .2%) for Boston University to take #1.

A fork in the road for UND’s NCAA tournament chances

My last full season prediction for UND, in February 25’s No surprises in UND’s PWR Outlook, suggested that UND needed 2 of its 4 remaining games to stay positioned for an at-large NCAA berth. UND closed the season with two splits and landed at a PWR ranking of #10.

Because UND did just barely enough to stay above the bubble, it can’t rest on its laurels. Elimination this round (in either 2 or 3 games) would push UND back onto the bubble, most likely toward the bottom.

That wouldn’t spell the end of UND’s season by any stretch: UND could get lucky and land in the upper #11-13 range based on other teams’ outcomes, or most conference tournaments could be won by higher ranked teams allowing #14 or even #15 in at-large, or unexpected future developments could raise UND’s PWR ranking even after UND has stopped playing.

But, far safer would be for UND to come out of this weekend with two wins. That still wouldn’t quite assure UND an at-large bid (as low as #13 or #14 after two wins this weekend would be possible with some bad luck), but they would be on the inside hoping not to fall instead of on the outside hoping to rise.

Methodology

Each forecast is based on at least one million monte carlo simulations of the games in the described period. For each simulation, the PairWise Ranking (PWR) is calculated and the results tallied. The probabilities presented in the forecasts are the share of simulations in which a particular outcome occurred.

The outcome of each game in each simulation is determined by random draw, with the probability of victory for each team set by their relative KRACH ratings. So, if the simulation set included a contest between team A with KRACH 300 and team B with KRACH 100, team A will win the game in very close to 75% of the simulations. I don’t simulate ties or home ice advantage.

Resources

No surprises in UND’s PWR outlook

Three weeks ago in UND’s PWR outlook with 10 games to go, I predicted that UND needed to win 7 of its remaining 10 regular season games “to be in a comfortable position for an at large bid going into the conference tournaments”.

UND proceeded to win 5 of its next 6 games, and has just 4 remaining in the regular season. The forecast remains remarkably consistent—UND now seems to need to win just 2 more games to go into the conference tournament in position for an at-large bid. Even just 1 more win would put UND on the bubble, with the conference tournament providing ample opportunity to rise off the bubble.

undendofseason

This week’s outlook

#10 UND is positioned to stay about the same with a split, most likely climb to #5-#7 with a sweep, or fall to #12-#14 if swept. Note that coming out of the weekend at #4 is possible, though it would take a lot of help.

undoneweek

Methodology

Each forecast is based on at least one million monte carlo simulations of the games in the described period. For each simulation, the PairWise Ranking (PWR) is calculated and the results tallied. The probabilities presented in the forecasts are the share of simulations in which a particular outcome occurred.

The outcome of each game in each simulation is determined by random draw, with the probability of victory for each team set by their relative KRACH ratings. So, if the simulation set included a contest between team A with KRACH 300 and team B with KRACH 100, team A will win the game in very close to 75% of the simulations. I don’t simulate ties or home ice advantage.

Resources

UND continues to position itself for NCAA tournament

A sweep last week put #12 UND in good position for the NCAA tournament. UND’s rise of 5 positions was on the upper end of expectations, as many opponents near them in the RPI faltered:

  • Cornell (got swept)
  • Vermont (got swept)
  • Duluth (got swept)
  • Colgate (two ties)
  • Providence (lost and tied)

The remaining season outlook is pretty much what you’d expect — last week’s, but with everything shifted by the two wins UND earned since then. UND now needs to win about half its remaining games to be in at-large position going into the conference tournament.

endofseason

One week outlook

A pretty quiet week for UND. A split most likely results in staying around #12-#14, a sweep would facilitate a climb to about #8-#11, while getting swept would lead to a fall to about #16-#19. Of course, as we saw last week those under 10% chance scenarios can occur (and should about one of ten times!)

oneweek

PWR formula note

The uncertainty around the PWR formula has apparently been resolved, with USCHO adopting the CHN formula last weekend. (See PWR formula uncertainty resolved). So, these forecasts now use the same formula as both USCHO and CHN.

Methodology

Each forecast is based on at least one million monte carlo simulations of the games in the described period. For each simulation, the PairWise Ranking (PWR) is calculated and the results tallied. The probabilities presented in the forecasts are the share of simulations in which a particular outcome occurred.

The outcome of each game in each simulation is determined by random draw, with the probability of victory for each team set by their relative KRACH ratings. So, if the simulation set included a contest between team A with KRACH 300 and team B with KRACH 100, team A will win the game in very close to 75% of the simulations. I don’t simulate ties or home ice advantage.

Resources

UND’s PWR ranking outlook

#17 North Dakota fell a couple spots with a split last weekend (the fall was consistent with my prediction for what would happen in a split), and continues to face more downside than upside.

A split this weekend would most likely drop UND a couple more spots, while getting swept could cause UND to plummet into the mid-20s. A sweep would do a little better than making up the ground lost last weekend, likely landing UND in the 13-15 range.

UNDoneweek

PWR calculation details

Readers of this blog probably already know that the RPI and PWR formulas for hockey changed this year. As I mentioned in a previous column—Dueling PWRs—there are currently two different interpretations of the new formulas. USCHO and CollegeHockeyNews implement the home/away weightings for RPI a bit differently.

After this post, unless otherwise stated this blog uses the CHN implementation. UND fares slightly better under the CHN implementation.

UND PWR forecast using the CHN PWR implementation

UNDendofseason

UND PWR forecast using the USCHO PWR implementation

UNDendofseasonUSCHO

I will continue to monitor both and highlight any significant differences in future posts.

Methodology

Each forecast is based on at least one million monte carlo simulations of the games in the described period. For each simulation, the PairWise Ranking (PWR) is calculated and the results tallied. The probabilities presented in the forecasts are the share of simulations in which a particular outcome occurred.

The outcome of each game in each simulation is determined by random draw, with the probability of victory for each team set by their relative KRACH ratings. So, if the simulation set included a contest between team A with KRACH 300 and team B with KRACH 100, team A will win the game in very close to 75% of the simulations. I don’t simulate ties or home ice advantage.

Resources

UND PWR outlook with 10 games to go

This week I’ll take another look at UND’s chances at making the NCAA tournament at large. I do that by forecasting UND’s PWR ranking, which mimics the NCAA tournament selection process.

However, be warned that some dispute has arisen in the college hockey online media world over the proper implementation of the NCAA’s 2014 revisions to the tournament selection process (see Uncertainty around PWR calculation). These forecasts currently assume the USCHO PWR formula.

The end of season outlook

Two weeks ago I predicted UND needed to win 8 of its remaining 12 to be in a comfortable position for an at large bid going into the conference tournaments. UND has split its only two games since then, and the prediction remains stable — UND now needs to win about 7 of its remaining 10.

undendofseason

Of course, if UND falls just a bit short, there’s an opportunity to make up some ground in the conference tournament.

The coming weekend

Though #15 UND experienced the predicted upside potential over the last couple weeks, they now face quite a bit more downside. Though a sweep would most likely result in a climb of a couple PWR spots, anything less would most likely result in a fall. Getting swept would likely even push UND back into the 20s (and with a tough outlook for the remaining season, as mentioned above).

undoneweek

Methodology

Each forecast is based on at least one million monte carlo simulations of the games in the described period. For each simulation, the PairWise Ranking (PWR) is calculated and the results tallied. The probabilities presented in the forecasts are the share of simulations in which a particular outcome occurred.

The outcome of each game in each simulation is determined by random draw, with the probability of victory for each team set by their relative KRACH ratings. So, if the simulation set included a contest between team A with KRACH 300 and team B with KRACH 100, team A will win the game in very close to 75% of the simulations. I don’t simulate ties or home ice advantage.

Resources

UND’s PWR rankings — a look ahead

Last week I gave a quick preview of UND’s PWR chances for the weekend. After a tie and a win, UND landed about where expected at #19. However, UND still has good upside potential.

UND’s one week outlook

A sweep should move UND solidly into “at large” position, while a split would leave UND about where it is (with a slight improvement a bit more likely than a slight fall).

UND’s regular season outlook

The bigger picture, though, is after UND’s phenomenal lossless streak, what do they need to do to finish in charge of their own destiny?

Winning 8 of its remaining 12 would put UND in a decent position to make the playoffs at large, while only winning 6 would likely leave UND in need of some conference tournament success.

More PWR analysis (non-UND teams)

I’m pleased to announce a new web site, College Hockey Ranked, to which I’ll (slowly) be migrating the non-UND hockey ranking information. This will allow SiouxSports to refocus more tightly on being a destination for UND fans, while CollegeHockeyRanked will try to appeal to a more general college hockey audience. UND fans who just want to read about UND won’t have to sift through as much information about other teams (except as it pertains to UND), non-UND fans who want more general analysis won’t have to skip over UND-centric analysis if they just visit that site, and of course anyone interested in both can simply follow both sites.

The immediate impact of this shift is that I’ll start posting my non-UND analysis to articles on that site. At least for this season, though, I’ll probably link between the two a lot.

Methodology

Each forecast is based on at least one million monte carlo simulations of the games in the described period. For each simulation, the PairWise Ranking (PWR) is calculated and the results tallied. The probabilities presented in the forecasts are the share of simulations in which a particular outcome occurred.

The outcome of each game in each simulation is determined by random draw, with the probability of victory for each team set by their relative KRACH ratings. So, if the simulation set included a contest between team A with KRACH 300 and team B with KRACH 100, team A will win the game in very close to 75% of the simulations. I don’t simulate ties or home ice advantage.

Resources

First look at UND’s Pairwise chances

I’ll go into some more depth in future weeks, but here’s a sneak preview of PairWise Ranking prognostication for the season–

This would have been a lot more interesting a month ago when UND seemed at serious risk of a lost season and I could have said something like, “UND needs to win the next 7 just to climb up to 20th”. Here we are at #19 with 14 regular season NCAA games remaining, and UND is definitely back in the hunt.

Here’s how this weekend’s games could affect its PWR:

The first thing that jumped out at me is how huge the spreads are — UND’s median climb for a sweep is 5 positions and fall for getting swept is 7. It appears that UND is just at a dense part of the RPI, but given that the PWR formula is new this year I’ll try to do some more digging.