🏀 KuBuckets Archive

According to Ken Pomeroy, KU is ranked behind Baylor, Oklahoma and Wichita State. FWIW, the consensus/average of all the computer polls shows:

• 8 - KU
• 10 Baylor
• 12 _Wichita State
• 13 - Oklahoma
• 15 - ISU

KU just won the Conference outright, has a better overall and conference record than Baylor and OU (3-0 against those 2 teams), has the #1 strength of schedule...how can Pomeroy be so consistently wrong and so off the consensus when it comes to KU? By the way, he also has Texas ahead of West Virginia.

!KenPom.PNG ↗

Looking at the numbers, our LUCK against the Top 25 on the List is Off the Chart! In KU, we call it "PHOG"! RCJH!

@Shanghai_RCJH

The LUCK number seems to be one Pomeroy created just for KU...

The problems with models is that they reflect the ability to capture the past. Sometimes they continue to work and sometimes they don't. Maybe Kenpom has reached the end of his road.

@sfbahawk

Models are designed to "model and predict." They are weak at the beginning of the season when there is not enough data but at this stage of the season they have reached what is called the "steady state" and should be pretty good. The interesting part is that Pomeroy's model is at odds with most every other model out there, which begs the question is Pomeroy right and all the other models wrong or is it the other way around? One way to evaluate a model is by what is commonly called the "reasonableness test" in which the question is asked do the results look reasonable? If the answer is no, then there is a problem with the model. Obviously Pomeroy's results do not seem reasonable and the majority of other models verify this fact, which logically implies that the Pomeroy model is flawed.

Trying to understand these numbers here. Both Baylor and Oklahoma are ahead of us in the rankings of Nostradumbass but both have worse records.

We (KU) are in the top20 in AdjO and AdjD. BU and OU have one of those stats but not the other. That's not it then.

Well maybe it's the AdjT... OU is ranked higher but Baylor is ranked below most high schools in this stat! That can't be it...

Hmmm, next number is luck. Has to be Luck that hurts us. We seem to be really lucky. Or really unlucky. Is this gauging good or bad luck here? And I'd think good luck is good but maybe it's bad to have good luck?

As for the rest of his trig and calc manure... all other stats we are ahead of the both teams, is that good or bad? I guess bad.

Makes sense. Both teams are better!

Lol anyone remember this in one of JBs post?

http://cjonline.com/blog-post/jesse-newell/2015-01-05/kureview-jayhawks-are-underdog-win-big-12#comment-1186251 ↗

(Sorry, I don't know how to make it into a link)

I would have hoped we would put this Ken Pomeroy guy out of minds. For goodness sake, the guy put a statistical category for "luck". How can one honestly quantify a thing like luck?

Did it count for WVU when Ellis missed the layup as luck? Or did they see Staten throw him off with a great effort? Or what about Johnson for KSU hitting those three's when his percentage is terrible for the season? Or was that on KUs terrible Defense that game?

It is literally impossible to attempt to calculate luck. It is simply something for Pomeroy to cover his arse when his prediction don't match up with the outcome.

Rest easy Rock Chalk brothers and sisters. Pomeroy's rating is for pseudo-intellectuals that can't measure what posters comment on everyday on this board like JB's "bad ball", Self's stubborness, toughness, emotion, and heart. I take everyone's view on this website more seriously than anything Mr. Pomeroy has to "calculate".

Sorry! Rant over. RCJH!

If he has a luck category my bet is that is he put it there so he could fudge the final outcome.

@JayHawkFanToo & @Kip_McSmithers Guess Ken is taking one of my favorite sayings deeply to heart!? "I'd rather be lucky than smart !!" Pretty clear he's trying the same game plan. LMAO as we see how it plays out for him.

@sfbahawk Yeah, that luck category is a red flag if I ever saw one. Fact, IMO it's downright comical...

If you have a category in your model for "luck"... shouldn't you have a category for "bull snit?"

@drgnslayr As a matter of fact, I do...& it's been expanded many times trough the years.

This is right up there with Lunardi's "eye test."

Luck is merely a low-probability outcome. If the outcome is good, it's good luck. If the outcome is bad, it's bad luck. I assume Pomeroy's luck metric reflects good luck. It suggests our team has exceeded expectations, which I suspect many of us would agree with. So this luck metric means unexplained success, some intangible that has not been included in the models predicting success.

Maybe luck reflects "toughness", or some other quality that's hard to quantify. I'd interpret that as a positive. Our team is doing better than predicted. That's good!

Look, KenPom's rankings this year are a head-scratcher. But the silver-lining is that we're not in the cross-hairs they way we have been in the past.

The experts and their rankings are confused, the media are confused, our opponents are confused. the fan base is confused.. maybe even some of the coaching staff.

What does that mean?

Stay tuned. Nobody knows.

......or as Gary Player once said, " the more I practice the luckier I get". I'll take some luck from the 3 point line this afternoon. RCJH

@GBHawk46 those are 3 great Jayhawks on your avi!

@bskeet

Oh so true.... the team with no identity.... creeping up on others this year. Love it!

Ever see the movie "Dead Man?"

Our role in that movie is the Native American named "Nobody."

These things are the problems with computers. They're obviously extremely color blind and unbiased. We've all talked about the ridiculousness of KU being 1 in the RPI. We can also see the ridiculousness of placing KU behind teams we lead by at least two games in the conference.

What becomes a little problematic is that does the committee look at these things? I'd guess they do.

I'm sure someone at Dukebuckets.com is also discussing the same thing as we are. How can they be #8?

@Crimsonorblue22 yep one of my favorite pic's even though 'New York New York' Russ Robinson is my all time player fave'. I just loved his attitude and unselfishness in his game play, I like team players.

@GBHawk46 most Jayhawks play that way!

@Shanghai_RCJH said:

Looking at the numbers, our LUCK against the Top 25 on the List is Off the Chart!

The Final Four, according to LUCK:

• KU
• Duke
• Virginia
• Kentucky

If you match them up:

• KU vs Kentucky --> KU

• Duke vs Virginia --> Duke

• KU vs Duke --> KU

Ahh March, when we can still dream the improbable dream

@bskeet you are feeln it this am!!

@wissoxfan83

There is nothing wrong with computer models, I design engineering computer models and I started creating them when the models design by non-engineers did not pass the reasonableness test, because the original designers really did not understand the engineering part and attempted to explain everything from the mathematical side even when it made no sense from an engineering perspective

Likewise, some of the computer models were developed by mathematicians or weather forecasters (Ken Pomeroy) with knowledge of the math but no knowledge of what they were attempting to model, i.e. basketball. Ken Pomeroy particularly seems to have moved his model away from the mainstream since he went "Hollywood" and his model is now at odds with most every other computer model out there. It is not the models that have problem but Pomeroy's model. He came out with that luck factor (which I call the KU factor, since it seems to be used uniquely to down rank KU) that @Jesse-Newell tried to explain at one time and I have posted many examples of how far off Pomeroy's rankings are, and I have not seen him attempt to justify them any more. Some Pomeroy's numbers, particularly those related to tempo have some merit but his rankings number suck...or blow...or both.

@JayHawkFanToo

And I know nothing much about computers beyond KUBuckets!

@JayHawkFanToo I agree; models are as good as the assumptions they are based on. The hardest part of building a model, according to an old modeling prof of mine, is defining your assumptions and objectives, and making the assumptions explicit. The rest is trivial.

If you disagree with the model output, then you need to go back and look at the assumptions it was built on. I haven't done that, but clearly Pomeroy has weighted something incorrectly. His luck metric shows how far off he is. We don't like his model output because his projections for KU are way off. He's clearly made some false assumptions.