Author Topic: kenpom 30, no tourney  (Read 41076 times)

0 Members and 1 Guest are viewing this topic.

Online pissclams

  • Global Moderator
  • Pak'r Élitaire
  • *****
  • Posts: 46344
  • (worst non-premium poster at goEMAW.com)
    • View Profile
Re: kenpom 30, no tourney
« Reply #325 on: March 07, 2019, 03:15:16 PM »
but which team has the higher grit factor?


Cheesy Mustache QB might make an appearance.

New warning: Don't get in a fight with someone who doesn't even need to bother to buy ink.

Offline mocat

  • Pak'r Élitaire
  • ****
  • Posts: 39042
    • View Profile
Re: kenpom 30, no tourney
« Reply #326 on: March 07, 2019, 03:16:30 PM »
our offense is ranked #106 in kenpom (worst of kp top 25 teams). the second worst offense in the kp top 25 is wisconsin (#43).

 :sdeek:

Offline K-S-U-Wildcats!

  • Pak'r Élitaire
  • ****
  • Posts: 10040
    • View Profile
Re: kenpom 30, no tourney
« Reply #327 on: March 07, 2019, 03:29:56 PM »
Serious question: I don’t “get” stuff like Ken Pom, RPI, etc. Iowa State is a complete dumpster fire but even after last night’s loss is sitting at 18 per KP. We’re at 25. Texas, which barely has a winning record and is in the bottom half of the conference, sits at 26. What does that mean? Is KP measuring something different that is maybe meaningful in a different way?

you're focusing too much on W-L records. it's more a predictive ranking based on offensive & defensive efficiencies. basically ISU should have a better record than they do (they rank 304 out of 353 in "luck"), and we should have a worse record than we do (#28 luckiest team).

you can look through ISU's schedule and see a ton of blowout victories and close losses

you can look through our schedule and see hardly any blowout victories, and double digit losses to marquette, texas, a&m, iowa st, and kansas

I appreciate this explanation. It makes sense - to a point. But over the course of an entire season you have to ask whether “luck” is really the factor, right? After playing about 30 games, including an 18 game conference schedule where everyone plays each other twice, I don’t think anyone can seriously contend that K-State, Texas, and Iowa State are comparable teams - we just lucky. Can they?

The Cats could still flub this weekend or the B12T and if that’s the case then I guess the models bear out, but assuming the Cats beat OU and at least make it to the semis - it’s gonna feel awfully shafty to be seeded comparably to ISU and Texas based on a predictive model like NET.
I've said it before and I'll say it again, K-State fans could have beheaded the entire KU team at midcourt, and K-State fans would be celebrating it this morning.  They are the ISIS of Big 12 fanbases.

Offline mocat

  • Pak'r Élitaire
  • ****
  • Posts: 39042
    • View Profile
Re: kenpom 30, no tourney
« Reply #328 on: March 07, 2019, 03:47:33 PM »
we're kind of the basketball equivalent of an otherwise mediocre baseball team that for whatever reason is really good at hitting with runners in scoring position.

Offline 8manpick

  • Pak'r Élitaire
  • ****
  • Posts: 19129
  • A top quartile binger, poster, and friend
    • View Profile
Re: kenpom 30, no tourney
« Reply #329 on: March 07, 2019, 03:49:26 PM »
How much do those efficiencies get skewed by a couple of games against, say, Tulsa and aTm? I used to feel pretty comfortable with these rating systems but this year throws me off. Is there any consideration for throwing out outliers?
:adios:

Offline 8manpick

  • Pak'r Élitaire
  • ****
  • Posts: 19129
  • A top quartile binger, poster, and friend
    • View Profile
Re: kenpom 30, no tourney
« Reply #330 on: March 07, 2019, 03:51:39 PM »
we're kind of the basketball equivalent of an otherwise mediocre baseball team that for whatever reason is really good at hitting with runners in scoring position.
Are we the 2014 Royals?
:adios:

Offline mocat

  • Pak'r Élitaire
  • ****
  • Posts: 39042
    • View Profile
Re: kenpom 30, no tourney
« Reply #331 on: March 07, 2019, 03:52:13 PM »
we're kind of the basketball equivalent of an otherwise mediocre baseball team that for whatever reason is really good at hitting with runners in scoring position.
Are we the 2014 Royals?

what if the 2018 cats were the 2014 royals  :surprised:

Offline K-S-U-Wildcats!

  • Pak'r Élitaire
  • ****
  • Posts: 10040
    • View Profile
Re: kenpom 30, no tourney
« Reply #332 on: March 07, 2019, 04:05:23 PM »
we're kind of the basketball equivalent of an otherwise mediocre baseball team that for whatever reason is really good at hitting with runners in scoring position.

So our stats don't look good except for the only stat that matters.

I'm a big proponent of predictive computer models. We use them in lending to great effect. Statistically they perform better than manual underwriting. So maybe a few years of the tournament will support the NET (although the advantages from higher seeding will make this tricky to evaluate).

But if there is a significant disparity between NET and conference record/standings, the committee ought to lean more heavily toward the record. Otherwise the regular season becomes purely a numbers game that devalues results. This is why the NCAA ditched the RPI, which currently ranks shitty ol' KU #1 because of their scheduling schemes. That's asinine. At this point Iowa State and Texas don't deserve a seed commensurate with K-State, and vice versa.
I've said it before and I'll say it again, K-State fans could have beheaded the entire KU team at midcourt, and K-State fans would be celebrating it this morning.  They are the ISIS of Big 12 fanbases.

Offline mocat

  • Pak'r Élitaire
  • ****
  • Posts: 39042
    • View Profile
Re: kenpom 30, no tourney
« Reply #333 on: March 07, 2019, 04:07:58 PM »
yeah but hitting w/ RISP is extremely hard to predict and salvy popped out to kung fu panda  :frown:

Offline catastrophe

  • Pak'r Élitaire
  • ****
  • Posts: 15100
    • View Profile
Re: kenpom 30, no tourney
« Reply #334 on: March 07, 2019, 07:13:49 PM »
“Luck” is a really funny thing to include in a computer algorithm.

“Hey Ken, this ranking system seems weird. Some of the top teams consistently struggle to beat even middle teams.”

“Oh well that’s because their luck factor is very low. Need to pump those luck numbers up and they’d be at the top of the conference.”

Offline catastrophe

  • Pak'r Élitaire
  • ****
  • Posts: 15100
    • View Profile
Re: kenpom 30, no tourney
« Reply #335 on: March 07, 2019, 07:20:09 PM »
What they should do is change “luck” to “heart” and count it as a major asset.

Offline mocat

  • Pak'r Élitaire
  • ****
  • Posts: 39042
    • View Profile
Re: kenpom 30, no tourney
« Reply #336 on: March 07, 2019, 08:07:46 PM »


“Luck” is a really funny thing to include in a computer algorithm.

“Hey Ken, this ranking system seems weird. Some of the top teams consistently struggle to beat even middle teams.”

“Oh well that’s because their luck factor is very low. Need to pump those luck numbers up and they’d be at the top of the conference.”



No it's not a factor in their ranking, it's just like a leftover number to quantify how much a team's efficiency doesn't match their w-l %

Offline catastrophe

  • Pak'r Élitaire
  • ****
  • Posts: 15100
    • View Profile
kenpom 30, no tourney
« Reply #337 on: March 07, 2019, 08:46:06 PM »
I get that it’s not a part of the ranking formula. It’s just funny to me that they act like it’s another variable in a statistical model instead of calling it what it basically is: a rate of error.

Offline catastrophe

  • Pak'r Élitaire
  • ****
  • Posts: 15100
    • View Profile
Re: kenpom 30, no tourney
« Reply #338 on: March 07, 2019, 08:49:12 PM »
It’s like a meteorologist saying they have a mathematical system to predict the weather and every time it’s wrong they just call it “luck” instead of thinking, hmm maybe I need to rethink how I’m weighing these variables.

Online wetwillie

  • goEMAW Poster of the WEEK
  • Pak'r Élitaire
  • *****
  • Posts: 30237
    • View Profile
Re: kenpom 30, no tourney
« Reply #339 on: March 07, 2019, 08:59:02 PM »
Maybe find another thread to ramble in?
When the bullets are flying, that's when I'm at my best

Offline catastrophe

  • Pak'r Élitaire
  • ****
  • Posts: 15100
    • View Profile
Re: kenpom 30, no tourney
« Reply #340 on: March 07, 2019, 09:22:24 PM »
I’ll stop once there is another actual KP30NT post ITT.

Offline mocat

  • Pak'r Élitaire
  • ****
  • Posts: 39042
    • View Profile
Re: kenpom 30, no tourney
« Reply #341 on: March 07, 2019, 09:30:29 PM »
I get that it’s not a part of the ranking formula. It’s just funny to me that they act like it’s another variable in a statistical model instead of calling it what it basically is: a rate of error.
It's not a variable tho

Offline kso_FAN

  • Global Moderator
  • Pak'r Élitaire
  • *****
  • Posts: 29506
    • View Profile
Re: kenpom 30, no tourney
« Reply #342 on: March 08, 2019, 06:58:14 AM »
we're not good at anything offensively except not turning the ball over and not getting our crap swatted

3PT% in Big 12 play though... :D

Offline kso_FAN

  • Global Moderator
  • Pak'r Élitaire
  • *****
  • Posts: 29506
    • View Profile
Re: kenpom 30, no tourney
« Reply #343 on: March 08, 2019, 07:03:41 AM »
I don't think our fans really understand how bad our offense was in the OOC against a pretty meh schedule. We did one really good thing with our scheduling by only including 3 >200 teams. However, we only ended up with 1 <50 team. Then we proceeded to play a bunch of those teams close, got one ok-ish win over Mizzou, got drilled by the only good team we played, and lost to a pair of mediocre (at best) teams. It was a really great schedule for an RPI metric, but no so much for any metric that includes efficiency. I think this team more than made up for it in the league, but the major reason any efficiency metric is in the mid to high 20s is because of our dreadful OOC offense.

Offline catastrophe

  • Pak'r Élitaire
  • ****
  • Posts: 15100
    • View Profile
Re: kenpom 30, no tourney
« Reply #344 on: March 08, 2019, 07:11:02 AM »
I get that it’s not a part of the ranking formula. It’s just funny to me that they act like it’s another variable in a statistical model instead of calling it what it basically is: a rate of error.
It's not a variable tho

A fatalist. I like it.

Offline mocat

  • Pak'r Élitaire
  • ****
  • Posts: 39042
    • View Profile
Re: kenpom 30, no tourney
« Reply #345 on: March 08, 2019, 07:12:37 AM »
I get that it’s not a part of the ranking formula. It’s just funny to me that they act like it’s another variable in a statistical model instead of calling it what it basically is: a rate of error.
It's not a variable tho

A fatalist. I like it.
It's an output not an input

Offline catastrophe

  • Pak'r Élitaire
  • ****
  • Posts: 15100
    • View Profile
Re: kenpom 30, no tourney
« Reply #346 on: March 08, 2019, 07:23:07 AM »
Yes, like a rate of error.

Offline mocat

  • Pak'r Élitaire
  • ****
  • Posts: 39042
    • View Profile
Re: kenpom 30, no tourney
« Reply #347 on: March 08, 2019, 07:50:34 AM »
Yes, like a rate of error.

yes. there are currently 184 teams with positive "luck" and 169 with negative "luck".

i guess if you wanted to come up with a catastrophic (™) rating system, you could just rank the teams based on their W-L % and then you wouldn't need a "luck" factor!  :dunno:

Offline catastrophe

  • Pak'r Élitaire
  • ****
  • Posts: 15100
    • View Profile
kenpom 30, no tourney
« Reply #348 on: March 08, 2019, 08:24:16 AM »
If I were even remotely a stats person I’d devise a system that actually took the “luck” factor into account and self-adjusted to raise the rankings for high luck teams (those consistently outperforming their expectations based on the model) and low luck teams (those consistently underperforming expectations).

To just leave it out there like a :dunno: seems lazy.

Offline mocat

  • Pak'r Élitaire
  • ****
  • Posts: 39042
    • View Profile
Re: kenpom 30, no tourney
« Reply #349 on: March 08, 2019, 08:26:06 AM »
i really don't think you would!