PWR Rank vs. KRACH Rank

[DeltaOne81]

That's far too small of a sample to matter as  much as you are implying it should, Ken.

By that rational, we're equals with Dartmouth this year. Is that true? Or, Mich St is distinctly better than us, right?

What about that Minn is 0-2-1 against UAA, and Wisconsin of 4-0-0 against them? You have a 4 game sample directly between the two, that's a 7 game sample with only one intermediate. Bigger sample? Wisconsin 15-4-1 to Minn's 12-7-1 for Common Opponents.

In their conference? Wisconsin 15-6-1, Minn 13-10-1.

Obviously none of these pictures tell the whole story, but neither does the head to head. They're just one part of the picture. You can weight things however you want to personally, but to pull up the H2H record, and say that's the only thing that matters, is certainly not going to be a consensus opinion.

[jtwcornell91]

[Q]Ken '70 Wrote:

MN beat WI 3 of 4 this year = reality
KRACH says WI is better than MN = unreality

Got it?

[/q]

RPI took 3 out of 4 points from Brown this year = reality
ECAC standings say Brown is better than RPI = unreality

::rolleyes::

[French Rage]

[Q]jtwcornell91 Wrote:

Common opponents and head-to-head results are a more direct way to compare teams than games against the rest of the NCAA, but they deal with a much smaller sample.  How you perform in a few games can override your performance for the whole rest of the season.  And the problem with the TUC criterion is that it compares straight winning percentage, and it's possible to have very different schedule strengths within the subset of TUCs.  Just look at the PWCs from 1999, when Quinnipiac had the 5th best record in the nation vs Teams Under Consideration:

[/q]

Random question, but back then didnt they used to determine who was a TUC based on over .500 record instead of over .500 RPI?

[jtwcornell91]

[Q]French Rage Wrote:

 [Q2]jtwcornell91 Wrote:

Common opponents and head-to-head results are a more direct way to compare teams than games against the rest of the NCAA, but they deal with a much smaller sample.  How you perform in a few games can override your performance for the whole rest of the season.  And the problem with the TUC criterion is that it compares straight winning percentage, and it's possible to have very different schedule strengths within the subset of TUCs.  Just look at the PWCs from 1999, when Quinnipiac had the 5th best record in the nation vs Teams Under Consideration:

[/Q]
Random question, but back then didnt they used to determine who was a TUC based on over .500 record instead of over .500 RPI?[/q]

Yes, so it was more extreme then, since it was easy for MAAC teams to become TUCs. Although it turns out that both of the other MAAC teams with winning records also had RPIs above .500.  (Although only UConn and not Holy Cross would have with the 25-50-25 RPI formula.)

[Ken \'70]

[Q]DeltaOne81 Wrote:

 They're just one part of the picture. You can weight things however you want to personally, but to pull up the H2H record, and say that's the only thing that matters, is certainly not going to be a consensus opinion.[/q]


[aside]I don't think you're are paying attention. The substring you're responding to, started by me, proposes that PWR use KRACH instead of RPI for the SOS component and that KRACH-ifying TUC makes sense.  I can only suppose, therefore, that your complaint above is that KRACH be used instead of PWR for ranking.  If not, then be careful where you put your replies because you're taking the discussion out the context which clarifies it.[/aside]

Of course H2H is only part of the picture, that's my point exactly - it's part of the picture. Just like KRACH, or RPI, is part of the picture.

But KRACH by itself doesn't value H2H, it blends those games in with all the rest when comparing two teams - that's why it's not dealing properly with reality.

PWR gives extra weight, appropriately, to what happened on the ice between two teams in coming up with it's relative ranking of the two.  And to what actually happened when those two teams played the same opponents.  That's not relevant to a ranking system?

It seems that your (plural) real issue isn't just that KRACH's mechanics are better than RPI's or TUC's or COP's or H2H's or some combination thereof, but that the philosophy of an overall ranking by a method of individual comparisons is fundamentally wrong.  Is that it?  

P.S.  The consensus opinion among the practitioners who count, the NCAA and its selection committee, is that H2H, COP etc are very important.

[DeltaOne81]

I guess I did take it out of context. I have read this thread, but I can't necessarily go back through the whole thing each time.

I'm more on the side of a KRACH-ified PWR. Not sure if we should KRACH-ify components or just replace RPI with KRACH. But something along those lines. Really, I was responded to the fact that you seemed to be saying that Wisconsin is better than Minn, and right now there's no way I can agree with that. If that wasn't your point, then I'll try to read more context and maybe you try to give more context

[KeithK]

I'm not convinced that head to head results are necessarily more meaningful than other games in determining which team of two teams is better.  It's a question of sample sizes.  One to four games (which is typically what you have) is small compared to the whole 30 game schedule (which is a small sample to begin with but it's what we've got).  Just because Yale might get lucky and tie Wisconsin one night (or beat 'em) doesn't tell us anywhere near as much about their relative strengths as the rest of their schedules.  Wisconsin is clearly better, and would be even if the Elis had somehow managed to beat the Badgers twice.  Obviously in the Yale/Wisconsin case the weight most people would intuitively apply to the rest of the schedule makes the result of the overall comparison obvious.  But in the Minnesota/Wisconsin comparison it's not quite so obvious.

I think we can all agree that the transitive rule doesn't apply to sports (A beats B, B beats C, therefore A is better than C).  But the reason it doesn't apply is because the results of any one game don't tell you enough about the relative strengths of the two teams involved.   This fact holds true when evaluating teams in a ranking system using H2H vs. entire schedules.

The only truly fair way to judge relative strength of a number of teams in a completely absolute fashion would be to play a round robin tournament between all the teams (identical schedules) with each team playing the others a very large number of times (sample size).  Oh, and you'd have to play it in a short period of time so temporal effects are eliminated... Unfortunately I don't think that's really consistent with the whole student athlete (or human athlete) thing.

[Ken \'70]

[Q]jtwcornell91 Wrote:

 [Q2]Ken '70 Wrote:

MN beat WI 3 of 4 this year = reality
KRACH says WI is better than MN = unreality

Got it?

[/Q]
RPI took 3 out of 4 points from Brown this year = reality
ECAC standings say Brown is better than RPI = unreality[/q]


Hey, you're getting it, congratulations!

(In a world with just RPI and Brown, RPI is better.  In world with all ECAC teams, Brown is doing better.  In a world with more leagues and teams, deciding who is better is best accomplished by considering those facts plus how each did against the best teams each played and how they did against the common teams they played.  Plus some formula that can't begin to yield results until dummy values are put in and which may need another dummy value to prevent a team like Cornell '70 from having a rating of infinity, which would make things a tad sticky.  See, I'm getting it too!)

[RichH]

Any thread where JTW is said to be "getting it" regarding this stuff is approaching a fine level of comedy, in my book.  ::popcorn::

[Josh '99]

Kinda seems like Ken is going in against a Sicilian when death is on the line.   ::laugh::

[Ken \'70]

[Q]DeltaOne81 Wrote:

 I guess I did take it out of context. I have read this thread, but I can't necessarily go back through the whole thing each time.

I'm more on the side of a KRACH-ified PWR. Not sure if we should KRACH-ify components or just replace RPI with KRACH. But something along those lines. Really, I was responded to the fact that you seemed to be saying that Wisconsin is better than Minn, and right now there's no way I can agree with that. If that wasn't your point, then I'll try to read more context and maybe you try to give more context [/q]

Fair enough.  Thanks.

And I like a KRACH-ified PWR too.