SEhockeyDAD wrote:As has been noted, its still early for things to shake out. Right now, KRACH has Austin ranked higher than Rochester Mayo despite the Mayo win over Austin. It should change after Austin plays some more significant opponents, I believe.
you can't base a whole season on one game... maybe Austin had an off game but saying a team is bad because they lose one game is poor judgement
mayo probably is the better team but they're also AA...
I didn't say Austin is a bad team, nor did I indicate that the rankings should be based solely on the Mayo-Austin game. My point was that its too early for KRACH rankings to be really accurate and that the Austin ranking (23rd overall) is an indication of that.
Although I indicated that it should be different with more significant opponents for Austin (because of their current strength of schedule),I'm now unsure how much their SOS can change. Currently Austin's SOS is ranked 107th, but of their remaining games, only Blaine and another game against Mayo can raise their SOS. The rest of their schedule has teams KRACH currently ranks between 43 and 107.
Missing (Red impact computer rankings, black impact stats only):
12/30/2006 Blizzard (WI) Moose Lake Rebel Invitational
12/30/2006 Dickinson (ND) Lake of the Woods Rebel Invitational 12/29/2006 East Grand Forks Roseau Roseau Girls Holiday Classic
12/28/2006 Park Rapids Detroit Lakes
12/27/2006 Totino-Grace Winona Winhawks Rochester
Note that some would still argue that this is VERY preliminary for the current season (even with nearly 1000 games in for the nearly 125 teams). Some believe that there isn't enough data (games) yet to make this valid at all. Also - some believe that there are still issues with teams with weak "Strength-of-Schedule" (SOS) being ranked too high - especially class A teams - and subsequently some would argue that the rankings by class may be more accurate, but not with the classes mixed together. Also, some believe that the SOS of some teams in general - even within their own class - may not yet be totally accounted for in their ranking at this point.
As to the column headings... I should have a key at top, but don't (good idea for future though):
Column:
1 - RK - RANK OVERALL IN STATE (OUT OF 124 TEAMS)
2 - PV - PREVIOUS RANK OVERALL IN STATE
3 - CL - CLASS
4 & 5 - Let's Play Hockey = Associated Press Poll Current & Previous Ranking
6 - RK - RANK IN CLASS
7 - PV - PREVIOUS RANK IN CLASS
8 - SC - SECTION #
9 - RK - RANK IN SECTION
10 - PV - PREVIOUS RANK IN SECTION
11 - TEAM
12 - RATING - This is the value that the algorithm determines based on record and opponents
13 - SOS RK - Ranking (lowest to highest) of average of opponents' rank - first calculate a rating, then rank in that order, then add up all the ranks of opponents and divide by number of opponents
14 - SOS RATING - average of opponents' rank as described above
15 - GP - Games Played - NOTE - the remaining columns count all games played/stats even though we don't count the games against non-MN opponents in the RATING consideration
16 - W - Wins
17 - L - Losses
18 - T - Ties
19 - PTS - Points (2 for each W, 1 for each Tie)
20 - GF - Goals For
21 - AVG - GF Average
22 - GA - Goals Allowed/Against
23 - AVG - GA Average
24 - PCT - Winning %
An "Average" team is considered a rating of 10.000. Any team above this 10.000 value is better than "Average" in the state (for all teams in both classes altogether). Below is worse. But, KRACH also tells us by how much...
If you want to know Team A's odds of beating Team B, you simply take the number for A and divide it by Team B's. Also easy to see the probability of A beating B (A/A '+ B).
i.e. Using the 2003-4 data below, the odds of SSP beating EP is 829.495/161.646 = 5.13 approx. 5 to 1 or the probability of EP beating SSP is (161.646/(161.646 '+ 829.495)) = approx. .163 or a 16.3% probability.
RANK RATING TEAM (LPH FINAL RANK)
1 887.113 BENILDE ST MARGARETS (1A)
2 873.438 ELK RIVER (1AA)
3 829.495 S ST PAUL (5AA)
4 387.613 N ST PAUL (2AA)
5 294.115 ANOKA (4AA)
6 235.504 HASTINGS (10AA)
7 213.948 BLAKE (3A)
8 188.409 CLOQUET
9 181.472 WAYZATA (9AA)
10 161.646 EDEN PRAIRIE (3AA)
cluelessinminnesota wrote:QRF predicted Edina over AHA in December using same data as KRACH. KRACH had AHA over Edina.
QRF predicted Blake over Breck last night KRACH did not.
Still don't know what is more accurate. Just pointing out that KRACH seems to work through the mess of limited data faster.
It will be more interesting later on.
FYI
KRACH says Eden Prairie will beat Edina
QRF says Edina is better than EP
Rankings mean nothing really, but are fun to debate, as too are the methods by which we determine them. QRF makes some dangerous assumptions, whereas KRACH makes teams earn their rankings through their play directly & indirectly through their opponents' play. Getting to seemingly logical rankings quicker using false assumptions is great for speed, but we'll see long term which is more accurate I suppose.
As you would expect KRACH seems to have gotten much more realistic as the season's progressed, but one obvious quirk is that Breck is still rated 3 places higher than Blake, despite last night's head-to-head matchup where Blake was clearly the dominant team. I know, "one game does not make a season" but this anamoly really stands out.
MNHockeyFan wrote:As you would expect KRACH seems to have gotten much more realistic as the season's progressed, but one obvious quirk is that Breck is still rated 3 places higher than Blake, despite last night's head-to-head matchup where Blake was clearly the dominant team. I know, "one game does not make a season" but this anamoly really stands out.
Right, there are others too like this. Basically, the issue is that Blake has those other losses - yes to very good teams, but they weigh blake down "Krach-wise." Head-to-head competition is the best indicator I agree, and is the first thing used when coaches seed sections for this reason. I'm convinced that by season's end there is no better ranking system than KRACh. Time will tell if that holds, meaning I don't believe another system will better predict section & state results, but, there is always the possibility that QRF or something else could do a better job I suppose this year for the first time.
ghshockeyfan wrote:Head-to-head competition is the best indicator I agree, and is the first thing used when coaches seed sections for this reason. I'm convinced that by season's end there is no better ranking system than KRACh.
Since KRACH is a computer-based system, maybe they should tweak it a bit to give more weighting to head-to-head meetings, especially recent ones.
Missing (Red impact computer rankings, black impact stats only):
GAME DATE AWAY TEAM AWAY SCORE HOME TEAM HOME SCORE NOTES 1/6/2007 East Range Lady Knights Princeton NOTE - These scores are now in, but weren't considered in the latest ranking. Still missing the first two above though... 1/9/2007 Lake of the Woods Warroad
1/9/2007 Winona Winhawks Roch. Century
1/8/2007 International Falls Eveleth
Missing (Red impact computer rankings, black impact stats only):
GAME DATE AWAY TEAM AWAY SCORE HOME TEAM HOME SCORE NOTES
1/12/2007 Balmoral Hall School - Winnipeg, Manitoba (CA) Warroad 1/11/2007 Detroit Lakes Park Rapids
1/6/2007 East Range Lady Knights Princeton
I'm not certian if the 1/11 & 1/6 games were ever even played...
Missing (Red impact computer rankings, black impact stats only):
GAME DATE AWAY TEAM AWAY SCORE HOME TEAM HOME SCORE NOTES 1/13/2007 Orono Mahtomedi
1/13/2007 Redwood Valley Long Prairie-Grey Eagle/Wadena-Deer Cr
1/13/2007 Watertown (SD) Worthington 1/11/2007 Detroit Lakes Park Rapids
1/6/2007 East Range Lady Knights Princeton
Missing (Red impact computer rankings, black impact stats only):
GAME DATE AWAY TEAM AWAY SCORE HOME TEAM HOME SCORE NOTES 1/16/2007 Cambridge-Isanti Rogers
1/16/2007 East Range Lady Knights Park Rapids
1/16/2007 Long Prairie-Grey Eagle/Wadena-Deer Cr Morris/Benson/Hancock
1/13/2007 Redwood Valley Long Prairie-Grey Eagle/Wadena-Deer Cr
1/13/2007 Watertown (SD) Worthington 1/6/2007 East Range Lady Knights Princeton
Missing (Red impact computer rankings, black impact stats only):
GAME DATE AWAY TEAM AWAY SCORE HOME TEAM HOME SCORE NOTES 1/13/2007 Redwood Valley Long Prairie-Grey Eagle/Wadena-Deer Cr
1/13/2007 Watertown (SD) Worthington 1/16/2007 Cambridge-Isanti Rogers
Missing (Red impact computer rankings, black impact stats only):
GAME DATE AWAY TEAM AWAY SCORE HOME TEAM HOME SCORE NOTES
1/19/2007 Beaver Brae - Keewatin/Kenora, Ontario (CA) Warroad 1/19/2007 Rogers Litchfield/Dassel-Cokato
Missing (Red impact computer rankings, black impact stats only):
GAME DATE AWAY TEAM AWAY SCORE HOME TEAM HOME SCORE NOTES
1/20/2007 Beaver Brae - Keewatin/Kenora, Ontario (CA) Lake of the Woods 1/20/2007 Champlin Park Blaine
1/20/2007 Princeton Morris/Benson/Hancock
Missing (Red impact computer rankings, black impact stats only):
GAME DATE AWAY TEAM AWAY SCORE HOME TEAM HOME SCORE NOTES
1/23/2007 Blizzard (WI) Moose Lake 1/23/2007 Buffalo Litchfield/Dassel-Cokato
1/23/2007 Coon Rapids Champlin Park
1/23/2007 Faribault New Ulm
1/23/2007 Long Prairie-Grey Eagle/Wadena-Deer Cr Park Rapids
1/23/2007 St. Louis Park St. Francis/North Branch
1/23/2007 Worthington Redwood Valley