• Welcome to OGBoards 10.0, keep in mind that we will be making LOTS of changes to smooth out the experience here and make it as close as possible functionally to the old software, but feel free to drop suggestions or requests in the Tech Support subforum!

Ken Pomery/Tennessee Rank Question (NWT)

This argument is never going to be resolved because the two sides are using different definitions of success. Wins and losses is a different measure than efficiency. And while efficiency may be a good predictor of how a team will do, wins and losses ultimately decide how "successful" a team is during the season.
 
The big thing here is that when a "good" team loses, it's usually going to be close, because teams generally won't be blowing them out. When those same "good" teams win, the wins usually won't be close, because they'll be blowing other teams out. So good teams are usually going to have more close losses than losses of wide margins in the first place, which should drag their close game records, again, down towards .500.

There's another good piece here about Florida a few years ago, and it talks about Duke's close game troubles in 2010 (you know, the year they won the tournament).
 
RJ throwing out "millenial bullshit" is just fantastic.
 
DistrictDeacon said:
This argument is never going to be resolved because the two sides are using different definitions of success. Wins and losses is a different measure than efficiency. And while efficiency may be a good predictor of how a team will do, wins and losses ultimately decide how "successful" a team is during the season.
Click to expand...

I'm a big stats guy but I've got no problem with a team getting favorable seeding for winning a lot of close games or similarly getting docked in their seeding for losing a lot of close games. At the same time, when making predictions I will favor the statistical against the anecdotal because I think in the long run it is more accurate.
 
Yeah, because you know what there's a lot of in the NCAA tournament?

CLOSE FUCKING GAMES!
 
bmoneydeac said:
I'm a big stats guy but I've got no problem with a team getting favorable seeding for winning a lot of close games or similarly getting docked in their seeding for losing a lot of close games. At the same time, when making predictions I will favor the statistical against the anecdotal because I think in the long run it is more accurate.
Click to expand...

Agree 100%.
 
Yeah, basically the NCAA tournament seeds teams based on wins and losses. Correctly so. It is important that this is the case so the games have meaning and are exciting to fans (the consumer). If Wake Forest is down 2 with :01 left in the ACC tournament and Miles Overton takes a 3, Kenpom or Las Vegas or other quantitative ranking systems don't really care if that shot goes in. But is important that the NCAA tournament cares for the excitement of the moment.

If you lose a ton of close game you are going to get trashed by the NCAA tournament but Kenpom and Vegas don't care because they know that losing a ton of close games has no affect on your future ability to win or lose close games. So that's how Tennessee situation happens.
 
BiffTannen said:
Yeah, because you know what there's a lot of in the NCAA tournament?

CLOSE FUCKING GAMES!
Click to expand...

Right, but the results of close games don't have any real predictive value for the future results of close games.

Which is to say, we should expect Tennessee to be about as successful in close games going forward as, say, Florida or Arizona.
 
RedSoxFaithful said:
Right, but the results of close games don't have any real predictive value for the future results of close games.

Which is to say, we should expect Tennessee to be about as successful in close games going forward as, say, Florida or Arizona.
Click to expand...

Really? Then why did Dean Smith win so many close games? If there was a close game with Dean Smith on the bench, which team did you think might win? The team with the better coach and the better players?
 
I'm pretty sure Ron Wellman used advanced statistics in seeding the NCAA tourney. That's what he means when he says, "eye test."

In fact, I bet some of these very same arguments were had amongst the selection committee.

Oh, to be a fly on that wall.
 
bmoneydeac said:
I'm a big stats guy but I've got no problem with a team getting favorable seeding for winning a lot of close games or similarly getting docked in their seeding for losing a lot of close games. At the same time, when making predictions I will favor the statistical against the anecdotal because I think in the long run it is more accurate.
Click to expand...

I agree with this first of all. Also, RJ's head is going to explode when I show these stats for the top four seeds (as seeded by the committee, bringing about its own conflation issues):

In games decided by five points or less, the top 4 seeds (16 teams) are 79-31
In games decided by three points or less, the top 4 seeds (16 teams) are 47-19
In games decided by one point or less, the top 4 seeds (16 teams) are 16-3

Of most interesting note, is the fact that the teams outside of KenPom's top 10 who got a top four seed (Wisconsin, Michigan, Syracuse, UCLA, SDSU, and Iowa State) went 9-1 in games decided by one point or less and a ridiculous 20-4 in one possession games. This of course presents the classic chicken or the egg issue: are they good because they win close games or do we just think they're good because they're winning those close games? Also important to note here is that these are the stats which are going to be most favorable to the "good teams win close games" argument. The tournament ranks and seeds teams on this theory, where good teams who also won close games are going to be seeded above other good teams who lost close games (as BMoney says above, I agree with this theory) but it's been proven to be true by KenPom and others that over the long run, good teams avoid close games they don't win them at a higher rate than anybody else.

By taking the committee's top 16 teams, it removed most unlucky teams from the pool, thus making it unlikely that the record would be closer to .500. In other words, there is a serious conflation issue which makes it difficult to discern anything from these numbers.
 
BiffTannen said:
Really? Then why did Dean Smith win so many close games? If there was a close game with Dean Smith on the bench, which team did you think might win? The team with the better coach and the better players?
Click to expand...

In any single scenario, yes, you always bet on the better team. This is touched on in the last few paragraphs on the first article I posted:

This is not to say there aren’t teams with the ability to consistently win close games. It’s just that their record in close games tells you almost nothing about that skill. Simply identifying a team that has made a bunch of last-second shots (or seen opponents miss) and declaring them a bunch of winners is accurate from a reporting standpoint, but assigning the members of said team some special amount of intestinal fortitude is, in the vast majority of cases, going overboard. Despite what their coaches may say, their intestines are not all that much different from the team that loses a bunch of close games. If you want to judge a team’s ability in close contests, it’s best to look at their execution. The problem is, this might not be consistent with the outcome of the game.

Of course, this won’t stop people proclaiming that a team that wins close games is truly clutch. And they are going to look brilliant slightly north of 50% of the time after that team’s next close game. You’ll have no choice but to buy in. But that doesn’t mean you’ll be right.
Click to expand...
 
Wakeforest22890 said:
I agree with this first of all. Also, RJ's head is going to explode when I show these stats for the top four seeds (as seeded by the committee, bringing about its own conflation issues):

In games decided by five points or less, the top 4 seeds (16 teams) are 79-31
In games decided by three points or less, the top 4 seeds (16 teams) are 47-19
In games decided by one point or less, the top 4 seeds (16 teams) are 16-3

Of most interesting note, is the fact that the teams outside of KenPom's top 10 who got a top four seed (Wisconsin, Michigan, Syracuse, UCLA, SDSU, and Iowa State) went 9-1 in games decided by one point or less and a ridiculous 20-4 in one possession games. This of course presents the classic chicken or the egg issue: are they good because they win close games or do we just think they're good because they're winning those close games? Also important to note here is that these are the stats which are going to be most favorable to the "good teams win close games" argument. The tournament ranks and seeds teams on this theory, where good teams who also won close games are going to be seeded above other good teams who lost close games (as BMoney says above, I agree with this theory) but it's been proven to be true by KenPom and others that over the long run, good teams avoid close games they don't win them at a higher rate than anybody else.

By taking the committee's top 16 teams, it removed most unlucky teams from the pool, thus making it unlikely that the record would be closer to .500. In other words, there is a serious conflation issue which makes it difficult to discern anything from these numbers.
Click to expand...

LOL. You think rj's going to read that bullshit?

Besides, everybody knows that law school students don't know the first thing about statistics.
 
RedSoxFaithful said:
This article is a pretty good primer on the numbers behind close games.

Basically, KenPom found that a team was no more likely to win "close games" (defined as 1-3 point margins) after winning their first game by 1-3 points than they were after winning their first five games by 1-3 points. It always regressed back to a 50/50 chance.
Click to expand...

Yep...so the numbers presented above mean...well nothing in the context of predictive value. If good teams win close games, then we would expect the top 4 seeds to fare better in close games rather than being able to use KenPom's rankings to get a sense of who is going to advance. I'll take the predictive ability of KenPom's metrics over how teams have fared in close games throughout the season. Which would you use RJ?
 
Wakeforest22890 said:
I'm going to take the top four seeds and see what their overall record is in games 5 points or closer. I don't know what the result is, and it is still admittedly a small sample size, but I would suspect that the 16 teams will be relatively close to .500 in games 5 points or closer. RJ, it seems you would assume that the top 16 teams (according to the selection committee) will be far greater than .500 in close games right?

ETA: Also this is likely to be more in favor of your position because the tournament committee is more likely to seed teams in the top four that have been "lucky" i.e. had good luck over the course of the year in close games.
Click to expand...
Eh, you'd expect the really good teams to be a little better than .500 in close games because of how the distribution would fall. If you are favorites most of the time, close wins are going to be more common than close losses, even if good teams/coaches aren't "better" in close games
 
You must log in or register to reply here.
Back
Top