2013-14 KRACH RANKINGS - NOW ON MGHCA WEBSITE

ghshockeyfan · Post by **ghshockeyfan** » Sat Nov 23, 2013 1:03 am

Main website: http://www.bgoski.com/krach

See MN Hockey Hub for rankings and more info.:
http://www.mghca.com/page/show/427504-krach-rankings

sinbin · Post by **sinbin** » Sat Nov 23, 2013 11:27 am

Thanks much! Clearly it's early, but some interesting results, nevertheless.

MNHockeyFan · Post by **MNHockeyFan** » Sat Nov 23, 2013 11:59 am

Wow, in AA the Lake Conference teams take 5 of the top 8 spots!

sinbin · Post by **sinbin** » Sat Nov 23, 2013 1:32 pm

Similar to LPH where the Lake Conference teams occupy 5 of the 7 top spots. Too bad at most 2 of them can make it to State.

ghshockeyfan · Post by **ghshockeyfan** » Sat Nov 23, 2013 11:01 pm

Updated 11/23 - 11PM

allhoc11 · Post by **allhoc11** » Sun Nov 24, 2013 11:05 am

GHS, I know the algorithm is way to complicated for simple folk like me to understand, but I have a question.

When you look at the overall ranks you have two 6-0 teams, one is 30 spots higher in SOS, but 37 points lower in the rating/rank. It was my understanding that Krach weighed teams strength of schedule into the rating, is there something obvious that I'm missing, or is there a benefit in a Krach ranking for heavily outscoring bad teams.

ghshockeyfan · Post by **ghshockeyfan** » Sun Nov 24, 2013 12:17 pm

allhoc11 wrote:GHS, I know the algorithm is way to complicated for simple folk like me to understand, but I have a question.

When you look at the overall ranks you have two 6-0 teams, one is 30 spots higher in SOS, but 37 points lower in the rating/rank. It was my understanding that Krach weighed teams strength of schedule into the rating, is there something obvious that I'm missing, or is there a benefit in a Krach ranking for heavily outscoring bad teams.

No score differenatial is used in this implementation of KRACH.

That aside, and I don't know that I've talked about this in many years, SOS is simply the ranking of the teams you play added up and divided by the number of games to obtain an average.

In an undefeated team comparison situation, then this could be misleading if the team with the best opponent has a wider range of opponents played that could weigh down their average.

My guess is that the highest ranked opponent that the two teams faced and beat belongs to the team that is ranked higher overall - meaning their 30pt differential in rating value comes from the fact that they beat the best team of the two sets of opponents.

So, is this true? ...at a quick glance:

Hopkins is 6-0, has a KRACH rating of 208 #2 overall, and is #78 in SOS (64.5 avg) with this schedule:

North Metro #120
Champlin Park #96
Achiever Academy #13
Eastview #43
New Prague #71
Orono #44

(120 + 96 + 13 + 43 + 71 + 44) = 387/6 = 64.5 avg SOS as mentioned above

East Grand Forks is 6-0, has a KRACH rating of 171 #3 overall, and is #49 in SOS (53.3 avg) with this schedule:

Bemidji #50
Fergus Falls #76
Woodbury #52
Warroad #64
Orono #44
Mound-Westonka #34

(50 + 76 + 52 + 64 + 44 + 34) = 320/6 = 53.3 avg SOS as mentioned above

So, it is as we described above, Hopkins is ranked higher because they beat the best team of the two schedules (Achiever @ #13) - but their SOS avg is worse as they played a wider range of opponent strengh.

ghshockeyfan · Post by **ghshockeyfan** » Sun Nov 24, 2013 12:22 pm

Updated 11/24 - 9AM

ghshockeyfan · Post by **ghshockeyfan** » Tue Nov 26, 2013 12:28 am

Updated 11/25 - 11PM

MNHockeyFan · Post by **MNHockeyFan** » Tue Nov 26, 2013 12:45 am

ghshockeyfan wrote:Updated 11/25 - 11PM

Most Recent Ranking Links:
==========================================
CLASS AA:
http://www.bgoski.com/KRACH_AA.htm

CLASS AA KRACH rankings still showing significant differences from LPH...they should even up considerably in the next few weeks. The question is, in the meantime, which one is the better indicator?

alcloseshaver · Post by **alcloseshaver** » Tue Nov 26, 2013 11:35 am

The Krach is a crock, even with the small amount of data to work with it makes little sense to have for example a team like Rogers ranked ahead of a Maple Grove or an Andover. Wait for more data and use the MSHSL preferred site QRF.

sinbin · Post by **sinbin** » Tue Nov 26, 2013 12:03 pm

Please share the current QRF rankings with us then. Or, share the 2013 QRF rankings compared to final KRACH and final actual results. You could throw LPH rankings into the mix as well. It would be interesting to compare and see which model is "crock-ier". I am agnostic to these models, but in the interest of furthering the development of the predictive analytics discipline, I would be very interested to see how they compare.

The point being that (1) all models will generally improve in accuracy as more data is available and (2) all models will have outliers, particularly when data is scarce.

State Champ 97 · Post by **State Champ 97** » Tue Nov 26, 2013 12:22 pm

QRF will not be available until 12/9. They always wait a while to gain more data to have a more accurate ranking.

royals dad · Post by **royals dad** » Tue Nov 26, 2013 2:24 pm

sinbin wrote:Please share the current QRF rankings with us then. Or, share the 2013 QRF rankings compared to final KRACH and final actual results. You could throw LPH rankings into the mix as well. It would be interesting to compare and see which model is "crock-ier". I am agnostic to these models, but in the interest of furthering the development of the predictive analytics discipline, I would be very interested to see how they compare.

The point being that (1) all models will generally improve in accuracy as more data is available and (2) all models will have outliers, particularly when data is scarce.

You cant really judge them based on results as the section alignment is so heavy in the rankings towards just a few sections. Nothing new it used to always be section 4, now it is 6. When most of the top ten are in 2 of 8 sections and most of the top twenty are in four of the 8 sections your final 8 will not be the top 8 teams in the state.

wolfman · Post by **wolfman** » Tue Nov 26, 2013 3:06 pm

I think the state AA girls tourney should be six teams from section 6. The two other teams should be picked by the dads of section six teams. That way we will have the best eight teams at state and we all want that for the kids.

blondegirlsdad · Post by **blondegirlsdad** » Tue Nov 26, 2013 3:18 pm

wolfman wrote:I think the state AA girls tourney should be six teams from section 6. The two other teams should be picked by the dads of section six teams. That way we will have the best eight teams at state and we all want that for the kids.

I'm sensing some sarcasm here.

And we really could use 7 teams if you must know.

wolfman · Post by **wolfman** » Tue Nov 26, 2013 3:30 pm

BGD im just kidding around.

blondegirlsdad · Post by **blondegirlsdad** » Tue Nov 26, 2013 6:07 pm

Me too.

6 is fine

sinbin · Post by **sinbin** » Tue Nov 26, 2013 9:48 pm

blondegirlsdad wrote:
wolfman wrote:I think the state AA girls tourney should be six teams from section 6. The two other teams should be picked by the dads of section six teams. That way we will have the best eight teams at state and we all want that for the kids.
I'm sensing some sarcasm here.

And we really could use 7 teams if you must know.

6 or 7 might be pushing it. I think everyone would be fine with 5. That seems like a just compromise.

ghshockeyfan · Post by **ghshockeyfan** » Wed Nov 27, 2013 8:29 am

Updated 11/27 - 8AM

Marty · Post by **Marty** » Wed Nov 27, 2013 10:00 am

Way too early for any ranking to have merit other than a few on top that are undefeated.

Strength of schedule impact appears to be too weak ... or my comment above is further validated.

Marty · Post by **Marty** » Wed Nov 27, 2013 10:03 am

4 team state AA tournament.

Minnetonka, BSM, Hill and a play in team.

All games at Ridder.

ghshockeyfan · Post by **ghshockeyfan** » Fri Nov 29, 2013 3:54 pm

Updated 11/29 - 4PM

MinnGirlsHockey · Post by **MinnGirlsHockey** » Fri Nov 29, 2013 4:39 pm

GHS- Final game scores from which site are used? MN Girls Hockey Hub, MGHCA.com, or Minnesota-Scores.net (I believe QRF uses the latter)? Sorry, I probably missed this earlier. I'm still seeing a bunch of discrepancies between the sites (including tons of missing scores on MGHCA.com). Thanks.

ghshockeyfan · Post by **ghshockeyfan** » Fri Nov 29, 2013 4:59 pm

MinnGirlsHockey wrote:GHS- Final game scores from which site are used? MN Girls Hockey Hub, MGHCA.com, or Minnesota-Scores.net (I believe QRF uses the latter)? Sorry, I probably missed this earlier. I'm still seeing a bunch of discrepancies between the sites (including tons of missing scores on MGHCA.com). Thanks.

Currently the HUB is used. This may change once the MGHCA mirrors the HUB as I believe is being worked on for consistency across sites. Theoretically, once the mirroring is complete, then the Minnesota-Scores.net should be in sync too as I believe entering a score in the MGHCA kicks it out to other media outlets like the AP, area newspapers, websites, etc

HighSchoolHockeyForums.com