clock menu more-arrow no yes

Filed under:

WRNL Exclusive: A Conversation With Ken Pomeroy

New, 31 comments

Why do fans get so upset with analytics?

Fall brings a lot of wonderful things. College football, cool weather, orange leaves and of course the return of college basketball. With the return of college basketball comes the return of rankings, bracketology and upsets -- all things that college basketball fans have learned to love and enjoy.

One of the names you will hear early and often this season is Ken Pomeroy. Writers on this site and others will refer to his rankings and statistics as a way of defending hypotheses or just for use as a point of reference. Within many communities, the “KenPom” rankings have become a bit polarizing, as fans often don’t like analytics when they indicate their team isn’t any good.

We asked (very politely of course) if Pomeroy would take a few minutes to chat and give us some background on his site and system before the season started. We talked about several topics, but unfortunately there was no answer as to why Steve Prohm insists on using his timeouts after made baskets.

Enjoy!

On how he got into basketball analytics...

Ken Pomeroy started his career in basketball analytics as most people would suspect: as a meteorologist for the government. Wait, what? As a hobby, Ken began pulling data and trying to formulate a system that determined the strength of college basketball teams. Four years later, Ken has quit that gig and is now fortunate enough to be doing his “hobby” full-time.

Biggest misconception about his rankings...

Unlike the RPI and human polls, a loss doesn’t necessarily move you down and a win doesn’t necessarily move you up. The true value of the ranking system is that it tries to put an accurate value on how strong each team is. Ken Pomeroy’s rankings are designed to be more predictive and determine how good a team is, not necessarily how good of resume they have.

“In the RPI, if you beat the 30th ranked team at home by one point, that is viewed as a quality win. In my system, if you beat the 30th ranked team at home by one point you are probably valued as being the 45th best team in the country. If you were ranked as the 20th best team before the game and beat the 30th ranked team by one point at home, you will probably move down, which confuses some people.”

Most beneficial way for a casual fan to use KenPom data...

Looking beyond a team’s raw rating can give a lot of useful information. If you really want to know a team’s strengths and weaknesses and what type of style they play, there are a lot of useful pieces beyond looking at just their ranking.

The biggest key is to try and remove yourself from fandom and getting upset with the rankings and look at the data with as unbiased of a view as possible. Ken also mentioned that the player data can be some of the most useful data for a casual fan who wants to know how well players are performing beyond the stats they see on their television screen.

Pomeroy also acknowledged that a true casual fan probably wouldn’t be spending their free time digging into the data that much. So, don’t try talking to the fan in the upper concourse who yells “Go! Go! Score! Shoot!” about offensive efficiency.

What percentage of college teams are using KenPom in their program?

There are 351 teams. Of those, there are roughly 300 that are using the data at some point in a given week. There are roughly 100-150 teams that have someone designated to keep tabs on the data throughout the season. Of those teams, there are probably around 50 that legitimately know what they are doing with the data and are getting a value from using it.

In short, pretty much everyone is using the site, but there is only a small amount of “power users” that are really getting an advantage from using the site.

We didn’t have the heart to ask him if UAB was in that “power user” group.

On consulting with Fred Hoiberg and the staff when he was at Iowa State...

There is information not on Ken Pomeroy’s site that is derived from play-by-play data. This is the kind of information Pomeroy used in his consulting work for several teams, including the Houston Rockets and even Iowa State at one point.

Ken primarily worked through TJ Otzelberger when he was an assistant in Ames. Pomeroy would forward the data onto the staff with additional commentary before each game.

With the staff change, Ken was unsure whether he would be doing much consulting work going forward for Iowa State. Except with us at WRNL, of course.

Do and/or should the NCAA selection committee use your data?

Inside the “war room” of the selection committee, their system is almost entirely governed by the RPI. They have begun to facilitate the use of more metrics, including Ken Pomeroy’s system.

If Ken was starting the selection process from scratch, he would include the use of 8 to 10 different systems. Most systems have an outlier or two. The use of more systems would minimize the effect of the outliers and give a more accurate composite view of specific teams.

Obviously he thinks his system would be of great use in this setting, but also acknowledged that there are other good systems that should be included as well.

Thoughts on other metric systems such as ESPN’s BPI...

From an outside view, Ken thought this particular index was a good system. By the end of the season, most good systems will offer similar results because there is so much data available.

Ken’s issue with the BPI is that ESPN has the ability to blind people with the system as if it is the one and only credible system out there.

At least they haven’t credited Tim Tebow with creating the BPI. Not that we know of anyway...

On evaluating a team’s 3-point defense...

In recent years, many of us fans have roasted the Cyclones after particular games for their 3-point defense. Most of the criticism is based on the percentage the opponent shoots from long range. For example, “I can’t believe we let Buffalo shoot 50% from three on us. Our 3-point defense is SO bad!”

Pomeroy has a different perspective.

“Most players aren’t going to shoot a 3-point shot with someone right in their face. The defense doesn’t have a lot of control over whether that shot goes in. They do have a lot of control over whether the opponent takes the shot.”

Pomeroy assesses a team’s ability to defend the 3-point shot based on how many their opponent takes in any given game. In short, if you can keep the opponent from shooting them, that is really the only thing you can control.

The instance in which his theory wouldn’t necessarily hold true is for teams with a really strong interior defense. In that case, shooting a contested 3-point shot might be better than the alternative of a contested 2-point shot.

In a study last summer, Pomeroy found that the 3-point success was about 80 percent in the hands of the offense, 19 percent in the hands of the defense, and whether or not you are using an Under Armour ball accounts for the other 1 percent.

Kidding — we were just making up that last part about the Under Armour basketballs.

What changes would you make to college basketball?

Pomeroy said the first thing he would get rid of would be more timeouts. In his opinion, the game is more fun to watch with fewer interruptions and less coach involvement. Rule changes that help that happen are on the top of Pomeroy’s list. Some of the other specific changes he would like to see are:

  • Reducing the shot clock to 24 seconds.
  • Move to quarters instead of halves.
  • Extend the 3-point line to at least international distance.

Would a change to a 24 second shot clock result in fewer upsets?

A fan (myself included) might think that with more possessions, the stronger team will have an even bigger advantage going forward. Pomeroy isn’t convinced.

“There would be more possessions, but we saw last year with the reduced shot clock there weren’t any fewer upsets. If anything we saw more upsets. The shorter shot clock will benefit the teams with more experience. More possessions would favor the team with more talent, but I think that would be offset by the team with more experience.”

What are your thoughts on per 40 minute statistics?

The popular response to extrapolating a player’s statistics to calculate their averages for an entire basketball game is often “they don’t play 40 minutes, so those statistics are pointless”. I disagree from a variety of angles on this topic, so it was interesting to hear Ken’s response.

“I think they are pretty useful, especially when comparing players within a team. When you are comparing per 40 statistics, you are ignoring pace and looking at per minute instead of per possession.”

Ken went on to explain it wouldn’t be fair to compare a player from Iowa State and a player from Virginia based on their 40 minute statistics. There are certain statistics that trend pretty well from part-time minutes to 40 minute statistics, such as rebounding, scoring and shooting.

Not every stat tailors well to per 40 minute stats. For example, blocked shots wouldn’t translate well because a player may not be as aggressive going after shots if they played more minutes, because they are would have to worry about foul trouble.

On college teams utilizing 2-for-1 opportunities at the end of half and game...

Ken indicated it was basically a no-brainer to take the opportunity when you could. Over the long run, taking advantage of the 2-for-1 opportunity was worth an extra quarter to a third of a point.

With the college game playing two halves instead of four quarters, the number of opportunities are limited. Taking advantage of a 2-for-1 opportunity at the end of a half increases your chances of winning by 1%. Though this seems rather small, it is still an opportunity worth taking advantage of.

How much time do you spend on the site and rankings during the season?

Obviously Ken doesn’t input all the data himself throughout the season, but we were curious to know how much time he spent per week maintaining the site.

“During the season the system is all automated. My only role during the season is to make sure the data that goes in isn’t erroneous in some way. Occasionally there is a box score that needs tweaking. If the data is good, the system should work all on its own.”

Other tidbits

Iowa State currently slots in at #26 on his preseason rankings. Remember, this isn’t a poll like the AP or USA Today put out. Ken isn’t “voting” on where these teams end up. Early in the season, a large portion of the ranking is based on last year and it will adjust moving forward. Stay patient and remember that the numbers aren’t biased.

You can catch Ken Pomeroy’s work at kenpom.com. You can also follow him on Twitter at @KenPomeroy.

We appreciate Ken taking the time to chat with us as we prepare for another exciting season of Iowa State basketball.