[Paint Touches] How accurate are KenPom's preseason rankings?

[The Sultan]

You wrote "in regards to calling a team #26," whatever that means.
What I'm asking is whether the way they tabulate points for the top 25 is different than the way they tabulate points for the others receiving votes.

No, but that's not the point. TAMU's example above, in a response to you, is the point.

[brewcity77]

You wrote "in regards to calling a team #26," whatever that means.
What I'm asking is whether the way they tabulate points for the top 25 is different than the way they tabulate points for the others receiving votes.

What it means is that you cannot call a team #26 based on a 25-team poll. It's an invalid response. TAMU noted this earlier, but here's the reason why. If one voter put DePaul #25 and no voters had Marquette on a ballot, your argument would be that DePaul is ahead of Marquette. But if the poll went to 26, and all of the voters would have put Marquette #26 and none of the voters would have put DePaul on their ballot, the points for Marquette would put them ahead of DePaul.

There is no #26. There is no #27. There is no numbered team beyond #30. There are other teams receiving votes, but only the top-25 are ranked because the system does not recognize teams beyond #25, so any total not reaching the top-25 isn't part of the poll result because the mathematical structure doesn't allow it.

[Pakuni]

No, but that's not the point. TAMU's example above, in a response to you, is the point.

Right, right. The very highly likely scenario in which the 26th best team in the country receives zero votes but 100th best DePaul gets one.
But what if, say, Xavier receives four 23rd place votes, three 24th votes and one 25th vote, giving them three fewer points than 25th ranked Iowa.
Is Xavier the same as a team that received no votes?

[withoutbias]

There are definitely some dweebs on Scoop.

We don't need to be data scientists when it comes to college basketball AP and Coaches polls.

[Uncle Rico]

There are definitely some dweebs on Scoop.

We don't need to be data scientists when it comes to college basketball AP and Coaches polls.

Without Bias is an interesting choice

[The Sultan]

Right, right. The very highly likely scenario in which the 26th best team in the country receives zero votes but 100th best DePaul gets one.
But what if, say, Xavier receives four 23rd place votes, three 24th votes and one 25th vote, giving them three fewer points than 25th ranked Iowa.
Is Xavier the same as a team that received no votes?

It was an exaggerated scenario no doubt, but the point of it is accurate.

[The Sultan]

There are definitely some dweebs on Scoop.

We don't need to be data scientists when it comes to college basketball AP and Coaches polls.

So in this topic, you have called people "number nerds" and "dweebs." Outside of the fact that this would last gotten you laughs in seventh grade, I am not understanding why you even bother participating in the discussion.

[TAMU, Knower of Ball]

You say this in defense of an article that examined only one conference, and a small one at that.

I didn't say that in defense of any article. I said that to counter your faulty argument that because AP voters can predict the top 25 about as accurately as KenPom that is somehow generalizable to ranking the entire 363 team division.

These two thoughts are contradictory.
In reality, IUPUI does not have as much a chance as Kentucky to be highly rated, because their ratings (for now) is based on what happened last year. Teams are largely bound by the previous season's results.
Which is fun to ponder, when you consider how different Kentucky's roster looks from one year to the next.

Yes they do. IUPUI just isn't as good as Kentucky nor recruits as well as Kentucky. They had every opportunity last season to put up as good of numbers as Kentucky or have as good of a recruitment cycle as Kentucky. They didn't so the algorithm correctly assigns them to be lower.

Also just to clarify in case you aren't aware, KenPom's preseason rankings are based on player stats. So last season when KenPom ranked MU in the preseason, it accounted for Moresell, Kuath, Kolek, and Omax' stats from their previous stops. It also accounts for the incoming freshmen and the fact that players departed. Yes there are things like coaching changes or how taking on new role could impact player performance that are not captured by the algorithm and these are things that are inherently unpredictable.

These two thoughts are contradictory.
It's also not something we know. Saying KenPom missing by 30 spots out of 363 is just extrapolating from the small and not representative sample size of a single 11-team conference. Until someone puts in the work to look at all those teams - or at least a larger, more representative group of teams - we have no idea how accurate he is.
And I'll take a pass on doing that work for now.

You don't know it but that doesn't mean others don't. Pre-fatherhood I used to track this kind of stuff because I'm a nerd and there's varying levels of merit to all of the major algorithms. They've also been embraced by the majority of coaches, talking heads, and Vegas (and even AP voters) most of whom know a little more than us scoopers on this topic. All the major algorithms have their strengths and weaknesses but are all somewhat predicative. You can also read thier methodologies and see that the numbers are sound.

A funny part of all of this is that many AP voters use KenPom when making their preseason picks. There's at least one of them who just copy and pastes the KenPom top 25 every week. So at least a portion of the AP poll's accuracy is due to voters using KenPom.

Honestly, I don't think we're that far apart. You've acknowledged that KenPom's preseason rankings are somewhat predictive of KenPom's postseason rankings and the postseason rankings are great. So that means that the preseason rankings have some value, we just disagree about how much. Personally, I don't think they are worth much beyond having a neutral preseason evaluation

[brewcity77]

Right, right. The very highly likely scenario in which the 26th best team in the country receives zero votes but 100th best DePaul gets one.
But what if, say, Xavier receives four 23rd place votes, three 24th votes and one 25th vote, giving them three fewer points than 25th ranked Iowa.
Is Xavier the same as a team that received no votes?

In the top-25? Yes. There's 1-25, also receiving votes, and everyone else.

And while the team at 26 would probably be reliably 26-28 if the ballot was longer, it's the 26-35 range where people are just getting random votes. The more teams that get votes, the less reliable anyone beyond 25 is.

[Dr. Blackheart]

Is Scoop University offering college credits for reading this thread?

[Skatastrophy]

Is Scoop University offering college credits for reading this thread?

You need some CEUs? I'll get you some CEUs.

[Newsdreams]

Oh Lord, let it be 11/7 so we can MOPE about not winning by more than 30 pts. instead of MOPPING about analytics, I have a headache

[MU82]

Oh Lord, let it be 11/7 so we can MOPE about not winning by more than 30 pts. instead of MOPPING about analytics, I have a headache

[PointWarrior]

Bold assumption, at least one Mope is thinking we could lose to Radford.

Oh Lord, let it be 11/7 so we can MOPE about not winning by more than 30 pts. instead of MOPPING about analytics, I have a headache

[Its DJOver]

Bold assumption, at least one Mope is thinking we could lose to Radford.

Of course we could lose to Radford.  The game's played on the court, not in a computer, once the ball is tipped anything can happen.  Isn't that the whole reason behind your disdain for kenpom?

I don't think anyone realistically thinks we will lose to Radford, but could we? Sure.

[The Equalizer]

In the top-25? Yes. There's 1-25, also receiving votes, and everyone else.

And while the team at 26 would probably be reliably 26-28 if the ballot was longer, it's the 26-35 range where people are just getting random votes. The more teams that get votes, the less reliable anyone beyond 25 is.

Take an example of a runoff election--a situation where you have 7 candidates on the ballot, voters get one vote, and if nobody gets 50%, the top two vote-getters move to a runoff election.

Under your logic, there can't be a second-place finisher--because voters have only one vote. You can only determine 1st place.  After that, using your logic, it is impossible rank 2nd through 7th. Pretty silly.

Since we can easily rank all candidates from 1st to 7th place with just one vote, under the same logic we can extend this to 25 votes in a pool of 360 candidates, and rank teams beyond 25th.

Its perfectly valid to rank teams also receiving votes in the delining order in which they received them, and it's going to be fairly accurate. The team receiving the 26th most votes in the current system is ranked 26th under the way the voting was conducted.  The team receiving the 27 most votes in the current system is ranked 27th.  And so on.

At this point, all you're doing is making perfect the enemy of the good.  The current system is reasonably accurate in ranking teams 26th and beyond, even with the limitation of the 25-place ballot. Stop trying to argue otherwise.

[brewcity77]

Quite simply, if teams receiving votes beyond 25 were considered to be accurately ranked, they would be. But the reality is they are not. There is 1-25 and others receiving votes. Period. People can piss and moan all they want, but the reality is nothing beyond 25 is recognized and asserting otherwise is just your personal fantasy.

[TAMU, Knower of Ball]

Take an example of a runoff election--a situation where you have 7 candidates on the ballot, voters get one vote, and if nobody gets 50%, the top two vote-getters move to a runoff election.

Under your logic, there can't be a second-place finisher--because voters have only one vote. You can only determine 1st place.  After that, using your logic, it is impossible rank 2nd through 7th. Pretty silly.

Since we can easily rank all candidates from 1st to 7th place with just one vote, under the same logic we can extend this to 25 votes in a pool of 360 candidates, and rank teams beyond 25th.

Its perfectly valid to rank teams also receiving votes in the delining order in which they received them, and it's going to be fairly accurate. The team receiving the 26th most votes in the current system is ranked 26th under the way the voting was conducted.  The team receiving the 27 most votes in the current system is ranked 27th.  And so on.

At this point, all you're doing is making perfect the enemy of the good.  The current system is reasonably accurate in ranking teams 26th and beyond, even with the limitation of the 25-place ballot. Stop trying to argue otherwise.

The purpose of an election is different from that of a ranking. An elections goal is to get most top votes. So you can clearly pick a second place winner because the only goal is most top votes and every elector gets a top vote.

A ranking is different. The purpose is not to identify each electors top vote but the gauge the perceived strength of each candidate. As already pointed out, 1 random team getting a 25th place vote from one voter does not necessarily mean that the electors  as a whole think that team is better than another team with zero votes who all the electors agree is just outside the top 25.

All that being said, if someone wants to call the team with most "others receiving votes" the 26th best team, who cares? They aren't actually the 26th ranked team but we all know what they mean.

[THRILLHO]

Quite simply, if teams receiving votes beyond 25 were considered to be accurately ranked, they would be. But the reality is they are not. There is 1-25 and others receiving votes. Period. People can piss and moan all they want, but the reality is nothing beyond 25 is recognized and asserting otherwise is just your personal fantasy.

But that's a decision the polls have made, and this (admittedly stupid) debate seems to be whether there is any theoretical reason that it's more valid to call the team with the 25th most ranking points #25 than to call the team with the 26th most ranking points #26.

While I just called this is a stupid debate, as a connoisseur of the genre I've been giving it some thought. I think the intuition is that, since we only ask voters to rank their Top 25, the amount of information that we have about a team gets worse and worse as we go down the points ranking. At the extremes, if there are 10 teams in others receiving votes, and the last one gets 1, which means literally one voter had that team at position 25, then it is highly unlikely that that team would be ranked #35 if all voters ranked all the teams. Not that they'd definitely be worse, just that we literally have next to no information.  This is essentially TAMU's point as I understand it and it makes sense.

But the other side of the argument says, that's essentially true for all teams near the lower end of the polls -- the #25 team wasn't ranked by every voter, and maybe they are polarizing and a lot of voters would rank them 35. If #25 and #26 have very similar point totals, and if everyone was asked to rank 26 teams maybe their order would switch (if every voter who previously gave points to the "#25" team ranked the "#26" team as 26, and no voters who gave points to the "#26" team but not the "#25" team ranked the "#25" team as 26). Without a complete ranking, it is possible for there to be movement for any team who is not included on every ballot. I get this point too -- 25 is an arbitrary cutoff for being "officially ranked" (though not seeming arbitrary because it corresponds to the number of teams voters are asked to rank).

So here's where I come down on this. To Brew's point, the polls can define their rankings however they want, and they say there are 25 ranked teams and "Others receiving points." Fine, so maybe it would be weird for, say, the ESPN chyron to list the rankings of teams #26... next to their names.

But more importantly, how should we behave on message boards? It's a little bit silly that if someone on a message board refers to the team with the 26th most points as #26, they get called out by pedants immediately saying "There is no #26, there are only 25 ranked teams," and their point not considered further. On the other hand, if you are trying to make a point about a team or coach by referring to their poll ranking being #30, then you probably are reaching.

Anyways, that's what I'm thinking right now but I'm happy to hear if I made any mistakes in thinking about how the rankings work.

[rocky_warrior]

All that being said, if someone wants to call the team with most "others receiving votes" the 26th best team, who cares?

Yet you (and especially Brew) have been taking issue with it for a couple pages now

[TAMU, Knower of Ball]

Yet you (and especially Brew) have been taking issue with it for a couple pages now

I can't speak for brew but I haven't taken issue with calling the top others receiving votes getter as 26. How all of this started was I took issue with Pakuni's assertion that the AP poll goes out to 35-40 teams if you count others receiving votes. That's bad math. Calling the top others receiving votes getter the 26th ranked team, that's just a vocab choice. Online debates about math I enjoy because I'm nerd. Online debates about vocab are dumb (and I acknowledge that I've gotten into my fair share of them here)

[Its DJOver]

While I just called this is a stupid debate, as a connoisseur of the genre I've been giving it some thought. I think the intuition is that, since we only ask voters to rank their Top 25, the amount of information that we have about a team gets worse and worse as we go down the points ranking. At the extremes, if there are 10 teams in others receiving votes, and the last one gets 1, which means literally one voter had that team at position 25, then it is highly unlikely that that team would be ranked #35 if all voters ranked all the teams. Not that they'd definitely be worse, just that we literally have next to no information.  This is essentially TAMU's point as I understand it and it makes sense.

I don't remember which AP voter it is, but I do recall that there was one that would put a different mid-major at 25 every week just so they would show up in the "other receiving votes" category and get a little recognition, and that school every time would have exactly one vote. That voter didn't think that they were worthy of a top 25 ranking, but it was more of a kind of "looking out for the little guy" mindset. 

[brewcity77]

Yet you (and especially Brew) have been taking issue with it for a couple pages now

I'm not a fan of the war on math. Leave that blatant stupidity to the Providence College fans.

[MU82]

For years -- decades actually -- the AP had a top 20. The world's largest news-gathering organization didn't expand the poll to reflect a top 25 until 1989.

It's possible that AP will expand the poll again to top 30 someday. In the meantime, I don't see why it bothers anyone if somebody refers to the first 5 "others receiving votes" teams as #26, #27, #28, #29 and #30 -- just as calling them #21, #22, #23, #24 and #25 before 1989 shouldn't have bothered anybody.

Now, going all the way out to the handful of teams receiving 1 point apiece and calling each of them #41 (or whatever) seems like a stretch. But that's really only my opinion -- just as all the stuff discussed earlier were only the opinions of others.

Polls are for pride and for conversation, anyway. They don't really "mean" anything.

[The Equalizer]

I can't speak for brew but I haven't taken issue with calling the top others receiving votes getter as 26. How all of this started was I took issue with Pakuni's assertion that the AP poll goes out to 35-40 teams if you count others receiving votes. That's bad math.

It's not bad math. All you and Brew have done is argue that if the ballot went to 30 or 35 or 40 teams, the ranking of teams past 25 might be marginally more accurate while scolding everyone about semantics like a couple of Karens. 

Yes, we all know the poll is called the "TOP 25", but data is collected on as many teams as voters believe should appear. It's purely semantics to argue that you can't rank order teams based on the vote totals they've received if that total is 26th or 27th instead of 24th or 25th.