KenPom Ratings: Arizona State at No. 34?

[Spanarkel]

Since KenPom is routinely cited on DBR, can someone please explain how KenPom currently ranks Arizona State 34th, vs. AP ranking of 5(with 5 first place votes)and USA Today ranking of 6th? Is there no place for the eye test anymore?

(sorry if this question is better addressed in a current thread, please feel free to move)

[uh_no]

preseason bias

[MChambers]

Since KenPom is routinely cited on DBR, can someone please explain how KenPom currently ranks Arizona State 34th, vs. AP ranking of 5(with 5 first place votes)and USA Today ranking of 6th? Is there no place for the eye test anymore?

(sorry if this question is better addressed in a current thread, please feel free to move)

KenPom is entirely based on statistics, not the eye test. If you don't like it, you can rely on the polls. Personally, I think the polls have overrated ASU, although the Sun Devils do have two more players joining the rotation in the next few weeks, and they could make ASU a lot better.

[CDu]

To expand upon what uh_no said, Pomeroy includes a preseason factor every year to allow for rankings early in the season. His approach depends upon data to rank teams' performances, and early in the season the data just doesn't exist. The preseason factor is based on a team's returning talent and their historical performance. As the season goes on, the preseason factor is weighted less and less and eventually not at all.

As such, Arizona State is being deflated by their past performance. That will go away. As long as Arizona State keeps winning, they'll keep rising.

[sagegrouse]

preseason bias


KenPom is entirely based on statistics, not the eye test. If you don't like it, you can rely on the polls. Personally, I think the polls have overrated ASU, although the Sun Devils do have two more players joining the rotation in the next few weeks, and they could make ASU a lot better.

I believe Pomeroy's approach does not weight current year results 100 percent until early to mid-January. So, what else figures in? Some preseason assessment, including heuristics (quantitative judgment).

[UrinalCake]

Seems like every year there’s a team or two where the KenPom numbers don’t line up. Last year it was Wichita State, if I remember correctly they were a borderline top-15 team in the human polls yet their KenPom was really low due to playing in a weak conference, so they got something like a 10 seed.

I don’t have another way to explain why AZ State is so low, they have really good wins against a relatively hard schedule.

[uh_no]

I believe Pomeroy's approach does not weight current year results 100 percent until early to mid-January. So, what else figures in? Some preseason assessment, including heuristics (quantitative judgment).

past 9 years performance, returning players, top 30 recruits, transfers, players coming off injury, penalty for coaching change, coach's history (longer it is, the more impact it has....for instance anywhere tony bennett coaches is going to play slow tempo)

https://kenpom.com/blog/2016-preseason-ratings/

is the best description he has, I think. though he's added transfers since then

[CDu]

Seems like every year there’s a team or two where the KenPom numbers don’t line up. Last year it was Wichita State, if I remember correctly they were a borderline top-15 team in the human polls yet their KenPom was really low due to playing in a weak conference, so they got something like a 10 seed.

I don’t have another way to explain why AZ State is so low, they have really good wins against a relatively hard schedule.

Wichita State was #8 in Pomeroy by the end of the year. So I don't think it was their Pomeroy ranking that dragged them down to a 10 seed. More likely it was that the committee didn't buy them given their W/L record in a mid-major. Their RPI at tourney time was #32, which is the more likely reason that they got the 10 seed. The committee still leans fairly heavily on RPI, and not Pomeroy.

[MChambers]

I believe Pomeroy's approach does not weight current year results 100 percent until early to mid-January. So, what else figures in? Some preseason assessment, including heuristics (quantitative judgment).

But nothing as subjective as the eye test.

[Wahoo2000]

Wichita State was #8 in Pomeroy by the end of the year. So I don't think it was their Pomeroy ranking that dragged them down to a 10 seed. More likely it was that the committee didn't buy them given their W/L record in a mid-major. Their RPI at tourney time was #32, which is the more likely reason that they got the 10 seed. The committee still leans fairly heavily on RPI, and not Pomeroy.

The RPI is such an archaic metric. I don't think the committee members should even receive that info.

Also - In my perfect world, the committee wouldn't even get team names on their sheets, or conference affiliation. Just raw numbers - records, advanced stats, etc. Make the bracket that way first THEN reveal the names and make the fixes to avoid rematches as necessary, and last to adjust for geographical purposes. There just needs to be a way to limit the "tv appeal" as a factor for picking the tournament as much as possible. I know that's not technically supposed to be a part of it, but........ come on.

[wobatus]

past 9 years performance, returning players, top 30 recruits, transfers, players coming off injury, penalty for coaching change, coach's history (longer it is, the more impact it has...for instance anywhere tony bennett coaches is going to play slow tempo)

https://kenpom.com/blog/2016-preseason-ratings/

is the best description he has, I think. though he's added transfers since then

I like to check T-Rank as well as kenpom, and that has Arizona State 19th. Also a number based system but sometimes a little better. They usually match up fairly closely.


http://www.barttorvik.com/

[uh_no]

I like to check T-Rank as well as kenpom, and that has Arizona State 19th. Also a number based system but sometimes a little better. They usually match up fairly closely.


http://www.barttorvik.com/

wasn't aware of them. that site layout is eye-burning though.

[wobatus]

wasn't aware of them. that site layout is eye-burning though.

Indeed. You can go to the individual team pages and that's free, unlike kenpom. Run by Bart Torvik, who is on twitter. Not to start a war or anything, but just went to go find his feed and found this:

https://twitter.com/totally_t_bomb/status/940972605871460352

[uh_no]

Indeed. You can go to the individual team pages and that's free, unlike kenpom. Run by Bart Torvik, who is on twitter. Not to start a war or anything, but just went to go find his feed and found this:

https://twitter.com/totally_t_bomb/status/940972605871460352

Not sure that'd be war starting....good tournament teams will generally have to be flexible...to be able to make adjustments to prevent a team from doing the same thing every time down the floor. Past few years, we haven't been able to make any adjustment to stop dribble penetration. This year it seems to be threes.

You don't have to be perfect, but you have to be able to do something...and against BC, we couldn't do anything.

This team right now would be susceptible to a hot shooting team, but most teams are, including MSU. It's just that it took an all american to beat MSU, but it took a couple of <?????> from BC to beat us.

This isn't the team that we'll go into march with, though. I still think there is a lot of room for improvement....I think the things that need fixing are more fixable than perhaps I have thought in years past. Guarding the three is a bit more about spacing and length than the technique required for guarding the dribble. IMO anyway.

[godins]

wasn't aware of them. that site layout is eye-burning though.

As if KenPom's layout is any better? I like the use of "hot" and "cold" color scales to quickly show teams near the top/bottom of the distribution curve for a given statistic. Plus, it's free and free is nice.

[Olympic Fan]

The RPI is such an archaic metric. I don't think the committee members should even receive that info.


You might not like the RPI (and I have problems with it), but all of us have to understand that RPI is the committee's favorite tool. It's used more to determine good wins and bad losses, than raw ranking. But when the committee looks at a teams' top 50 record or top 100 record, they are talking about RPI.

So like it or not, we need to pay attention to the RPI -- more attention than to Pomeroy, Sagarin or any other ranking system.

PS: Duke ranks higher in the RPI at the moment than in Pomeroy, Sagarin or either of the major polls. As of this morning, Duke is No. 2 behind Texas A&M:

http://m.espn.com/ncb/rpi

[uh_no]

As if KenPom's layout is any better? I like the use of "hot" and "cold" color scales to quickly show teams near the top/bottom of the distribution curve for a given statistic. Plus, it's free and free is nice.

- the horizontal dividers simply stop halfway across.
- a serif font at that size and density is overwhelming (there's a reason sans fonts like arial and helvetica are ubiquitous in digital settings).
- The vertical spacing is a bit much. It breaks up the visual appeal of the table, IMO
- there are no vertical dividers in the table
- dense tables will often break things up by alternating background colors (see kenpom). Obviously that is more difficult when you are adding color to display data, but there are several un-colored columns in the graph. this technique could be applied there

I think the font thing is probably the worst part, followed by the lack of dividers...but they all combine to make a very visually un-appealing table. Adding colors on top of the already iffy layout choices make the whole thing overwhelming

Not that this has any bearing on the work he's doing...which I will surely look into

[UrinalCake]

Also - In my perfect world, the committee wouldn't even get team names on their sheets, or conference affiliation. Just raw numbers - records, advanced stats, etc.

I’ve had that same thought - take away whatever prejudices the committee members have based on the name on the jersey and treat everyone based solely off their body of work. In practice this would be difficult to do, because the people could mostly figure out which team they’re looking at based on the numbers. So you’d have to sequester the committee members for the entire season, not let them watch anything, then present the numbers to them on Selection Sunday. But it’s still a good idea in theory.

[uh_no]

I’ve had that same thought - take away whatever prejudices the committee members have based on the name on the jersey and treat everyone based solely off their body of work. In practice this would be difficult to do, because the people could mostly figure out which team they’re looking at based on the numbers. So you’d have to sequester the committee members for the entire season, not let them watch anything, then present the numbers to them on Selection Sunday. But it’s still a good idea in theory.

There is also a problem of recusal. If you figure out which team is which, you obviously have a major conflict of interest. You also can't recuse yourself without everyone else knowing which team is which.

In practice everyone would know who's who anyway.

Whatever criteria they use are going to be arbitrary and up for debate. We get something to talk about. And if we get screwed or benefit every once in a while? At least it gave us something to rabble about!

[ipatent]

Generally I prefer KenPom, but ASU is up to #13 in Sagarin. (http://sagarin.com/sports/cbsend.htm)

[fidel]

- the horizontal dividers simply stop halfway across.
- a serif font at that size and density is overwhelming (there's a reason sans fonts like arial and helvetica are ubiquitous in digital settings).
- The vertical spacing is a bit much. It breaks up the visual appeal of the table, IMO
- there are no vertical dividers in the table
- dense tables will often break things up by alternating background colors (see kenpom). Obviously that is more difficult when you are adding color to display data, but there are several un-colored columns in the graph. this technique could be applied there

I think the font thing is probably the worst part, followed by the lack of dividers...but they all combine to make a very visually un-appealing table. Adding colors on top of the already iffy layout choices make the whole thing overwhelming

Not that this has any bearing on the work he's doing...which I will surely look into

You must spread pitchforks around, or something like that.

[22JumpShots]

I read someone explain the polls like this...and it has helped me to cope with their madness.

**Pre-Season Polls are based on what teams are "supposed to be".

**The AP Poll throughout the season is basically who is on a "hot streak".

Neither numbers in either poll truly mean jack, as any team can lose any night, and the opposite.

So AZ St. is on a tremendous hot streak. Doesn't mean they are the 6th best in the country.

[ChillinDuke]

Not sure that'd be war starting...good tournament teams will generally have to be flexible...to be able to make adjustments to prevent a team from doing the same thing every time down the floor. Past few years, we haven't been able to make any adjustment to stop dribble penetration. This year it seems to be threes.

You don't have to be perfect, but you have to be able to do something...and against BC, we couldn't do anything.

This team right now would be susceptible to a hot shooting team, but most teams are, including MSU. It's just that it took an all american to beat MSU, but it took a couple of <?????> from BC to beat us.

This isn't the team that we'll go into march with, though. I still think there is a lot of room for improvement...I think the things that need fixing are more fixable than perhaps I have thought in years past. Guarding the three is a bit more about spacing and length than the technique required for guarding the dribble. IMO anyway.

I'm not exactly sure what to fill in the <?????> above, but one option is "really good players."

They aren't on any draft boards that I've seen at this point in time, but Jerome Robinson and Ky Bowman are excellent players.

Jerome is listed on ESPN as a 6'6" junior averaging 17.2 points, 3.8 boards, 3.1 assists (2.5 turnovers), and ~1 steal per game, on 45% FG, 48% from 3, and 80% from the line. That's darn solid, if not great. All of his shooting %s are up year-over-year, and the crazy part is his actual averages are down across the board since last year. If he gets those more or less in line with last year, that's exceptional.

Yet Jerome Robinson isn't even the best player on his own team. Ky Bowman is listed at 6'1" and is a sophomore averaging 16.8 points, 6.9 boards, 5.8 assists (2.6 turnovers), ~1 steal, and half a block per game, on 43% FG, 32% from 3, and 81% from the line. And his FG% and 3P% are considerably down from his freshman season when he shot 49% and 45%, respectively. If those percentages move back up near his performance last year, watch out.

I wouldn't be overly fussed calling them a better backcourt than even our own (Grayson/Trevon).

BC's is one of the best backcourts in the ACC.

Why are some treating this team as if its the same team from 2015-'16 that went 0-18 in conference in Jim Christian's second year? This ain't them.

- Chillin

[CDu]

If Bowman continues to pass as well this year as he has so far, he will be on draft boards. That kid can ball.

[gep]

You might not like the RPI (and I have problems with it), but all of us have to understand that RPI is the committee's favorite tool. It's used more to determine good wins and bad losses, than raw ranking. But when the committee looks at a teams' top 50 record or top 100 record, they are talking about RPI.

So like it or not, we need to pay attention to the RPI -- more attention than to Pomeroy, Sagarin or any other ranking system.

PS: Duke ranks higher in the RPI at the moment than in Pomeroy, Sagarin or either of the major polls. As of this morning, Duke is No. 2 behind Texas A&M:

http://m.espn.com/ncb/rpi

That espn link for RPI seems to be a bit old... Duke is listed as 11-0.

I got this link from the NCAA website... with games through Dec 12.

http://www.ncaa.com/rankings/basketball-men/d1/ncaa-mens-basketball-rpi

Duke is #1, Arizona State is #2... in RPI :confused:

[BandAlum83]

That espn link for RPI seems to be a bit old... Duke is listed as 11-0.

I got this link from the NCAA website... with games through Dec 12.

http://www.ncaa.com/rankings/basketball-men/d1/ncaa-mens-basketball-rpi

Duke is #1, Arizona State is #2... in RPI :confused:

I don't know when the time period covered is, but how did Boston College drop from #74 to #81? They beat Columbia by 15 on the 12th.

[vick]

I don't know when the time period covered is, but how did Boston College drop from #74 to #81? They beat Columbia by 15 on the 12th.

RPI doesn't account for margin of victory, and Columbia is terrible, so this makes sense.

[BandAlum83]

RPI doesn't account for margin of victory, and Columbia is terrible, so this makes sense.

So winning against a bad team lowers your RPI?

[Olympic Fan]

So winning against a bad team lowers your RPI?

It can, yes ... and losing against a very good team can help your RPU.

[English]

So winning against a bad team lowers your RPI?

Frequently, even playing against a very bad team will lower your RPI, yes...and it's the gift that keeps on giving because, as the bad team continues to play badly and its RPI falls, your RPI falls as a measure of Strength of Schedule. That why you'll see that playing and beating many bad-but-not-egregiously-bad teams (e.g., 10-0 against RPI 100-200) is often better for one's RPI than beating top teams and bottom-feeder teams (4-0 against top-50 and 6-0 against 250+).

[Bluedog]

So winning against a bad team lowers your RPI?

Yep...RPI formula = 25% team winning percentage (WP), 50% opponents' average winning percentage (OWP), and 25% opponents' opponents' average winning percentage (OOWP)...So, yeah, your opponents' WP counts for MORE than your own WP. Does NOT matter who you beat and lost just total record. If you beat the #2 and #20 teams in country and lost to the #100 team in country your RPI is the same as if you beat the #20 and #100 teams in country, but lost to #2.

[Kedsy]

So winning against a bad team lowers your RPI?

Basically, the RPI is a combination of three variables: (a) your team's winning percentage; (b) the aggregate winning percentage of your opponents; and (c) the aggregate winning percentage of your opponents' opponents. Basically, (a) constitutes 25% of the RPI; (b) constitutes 50%; and (c) constitutes 25%. Or at least that's the way it used to be. It's possible they may have tweaked it since I last examined it, but any changes probably didn't alter the basic framework.

With the above, you can probably see having Columbia (1-10 record) as one of your opponents (category (b), 50% of your RPI) would hurt you more than the win would help your record in category (a).

You can also probably see how easy it is to game the RPI -- just play a bunch of games against bad teams in worse conferences. For example, right now Pomeroy's #252 team (Western Illinois) is 7-2. Playing them and getting the almost automatic win would help you in both (a) and (b), constituting 75% of your RPI. Sure, their poor schedule would hurt you in category (c), but that's just 25% of your grade, so overall a game against a team like that would help your RPI.

I hope you can also discern how stupid it is when the talking heads cite a team's RPI and then back it up with "schedule strength" (which for the RPI is the same as category (b)). It's double-counting of the most egregious sort.