Some interesting tweets from an advanced stat guy

[InstiGATOR1]

I received this series of tweets yesterday, but did not have time to post them:

Jeff (BPredict) ‏@BPredict
No, it means they're underrated. RT @robdauster: After losing to Tennessee last night, Florida is now 0-4 in close games. Concerning? Yes...

Jeff (BPredict) ‏@BPredict
Ex: 2010-11 Kentucky. They were 1-6 in games decided by <5 points & 3-7 in road games. As an underrated 4 seed, they made the Final Four.

Jeff (BPredict) ‏@BPredict
There simply is no evidence that bad luck in regular season close games carries to the postseason... in any sport. UF is very underrated.

The UK 2010-11 data are quite interesting.

[corpgator]

Pomeroy has a stat specifically for "luck." He calls it cyclical with teams that generally have bad luck one year having good luck the next. The thing is, though, Florida may have the lowest luck rating over the 12 seasons he has kept the stat. We've only been in the top 100 once the entire time.

That would suggest a trend to me. I'm going to run the data and calculate the average for every team soon.

[corpgator]

Ok, I ran the data and we have the 21st worst luck over the 12 year period Pomeroy has tracked the stat. In other words, we're the 21st worst team out of 321 teams at winning close games.

I threw out every team that hadn't played all 12 seasons.

By the way, Georgia Tech and Miami are 1-2 in that stat, but GTech is first by a mile, averaging 290th luckiest team over the 12 season to Miami's 250th. Our average is 221st.

Surprisingly no team is better than 69th on average. So some teams are really good at losing close games, but no team is really good at winning them.

[Jonas]

Corp, do you think that is indicative that Billy's scheme isn't as effective in a close game?

I wish the stats also gave us the relative strength of our opponents when we are in a close game. It is entirely possible that the stats are skewed to us losing close games more than average, because Billy's system is so good, we are able to play teams close that are actually much better than us. They still result in losses, but maybe when we were expected to lose by 10, we only lose by 5.

Maybe I'm grasping for straws, but I'm trying to reconcile Billy's great overall record and tournament record with his subpar record in close games. You would think to do well in the tourney, you need to be able to win close games against opponents that are just as good as you.

[InstiGATOR1]

Originally Posted by Jonas

It is entirely possible that the stats are skewed to us losing close games more than average, because Billy's system is so good, we are able to play teams close that are actually much better than us. They still result in losses, but maybe when we were expected to lose by 10, we only lose by 5.

I have thought about the same thing and think this could be true for this team. I do not think it has been consistently true in the past.

[InstiGATOR1]

Originally Posted by corpgator

Ok, I ran the data and we have the 21st worst luck over the 12 year period Pomeroy has tracked the stat. In other words, we're the 21st worst team out of 321 teams at winning close games.

I threw out every team that hadn't played all 12 seasons.

By the way, Georgia Tech and Miami are 1-2 in that stat, but GTech is first by a mile, averaging 290th luckiest team over the 12 season to Miami's 250th. Our average is 221st.

Surprisingly no team is better than 69th on average. So some teams are really good at losing close games, but no team is really good at winning them.

Have you thought about running the numbers a bit more and writing this up as study? You could entitle it something like is "Is Ken Pomeroy Luck Luck." Because if you are right and luck is not random over seasons, then Pomeroy is measuring something else.

[GatorRade]

Originally Posted by corpgator

Ok, I ran the data and we have the 21st worst luck over the 12 year period Pomeroy has tracked the stat. In other words, we're the 21st worst team out of 321 teams at winning close games.

I threw out every team that hadn't played all 12 seasons.

By the way, Georgia Tech and Miami are 1-2 in that stat, but GTech is first by a mile, averaging 290th luckiest team over the 12 season to Miami's 250th. Our average is 221st.

Surprisingly no team is better than 69th on average. So some teams are really good at losing close games, but no team is really good at winning them.

This actually sounds like evidence against your hypothesis to me. Some team had to be the lowest. That close grouping of all the teams is consistent with Pomeroy's views.

[g8rboy]

our problem is we win by blowout too much; and lose the close ones.

[GothamGator]

You need to look at the average luck scores over all the seasons and then check the variance over all the teams. You can also check how many teams are consistently "lucky" vs. consistently "unlucky". The odds that a team is unlucky all 12 years is only 1 in 4096, so if it is really luck, you shouldn't see any of those. The odds that a team is unlucky 11 of 12 years is 1 in 341, so you should get one of those, but unlikely to have many more than that.

[Colin]

We have only lost when playing teams on their home courts, but there's no home court advantage in the NCAA tourney.

If we're healthy come tourney time, we will be one of the toughest outs in the tourney. Not worried about the UT loss, considering our depth issues in that game, and that it was on their home court.

[corpgator]

I'm not a stats guy, with my knowledge of stats being only from a stats I class, so it will take me a while to do anything more in depth. A cursory glance at my list shows no one with good luck or bad luck over all 12 seasons. I was wrong, and we did crack the top 100 twice at 83rd and 95th, and had positive luck 3 times.

It wouldn't be that hard to run luck against SOS to see how they correlate.

I saw 3 teams with bad luck 11 out of 12 seasons and 1 team with bad luck 10 out of 12, and neutral luck 0.00 once.

[GatorRade]

Originally Posted by g8rboy

our problem is we win by blowout too much; and lose the close ones.

I'm not sure this is a "problem". Would you prefer to win close ones and lose by blowouts? Basically, it just means that our team's outcomes are likely to be between +30 and -5, which is obviously much superior to +5 to -30.

[g8rboy]

Originally Posted by GatorRade

I'm not sure this is a "problem". Would you prefer to win close ones and lose by blowouts? Basically, it just means that our team's outcomes are likely to be between +30 and -5, which is obviously much superior to +5 to -30.

sorry. i was being facetious.

[GatorRade]

Originally Posted by g8rboy

sorry. i was being facetious.

Sorry, I missed that, since your facetious comment is so close to actual comments on this board.

[g8rboy]

Originally Posted by GatorRade

Sorry, I missed that, since your facetious comment is so close to actual comments on this board.

this is true. i am generally not negative, nor get upset, since i am actually not playing the games.

[WESGATORS]

You can post all the stats you want, some people have preconceived notions that simply cannot be refuted.

Go GATORS!
,WESGATORS

[WESGATORS]

6 of the past 10 national champions were ranked below #120 in luck in their championship years. The lowest two were North Carolina (#185, 2009) and FLORIDA (#186, 2006). No champion has been ranked in the top 50 in luck since 2003 when Syracuse won it (luck ranking = #18).

Go GATORS!
,WESGATORS

[gator2109]

Winning a game on the road simply shows that you are a stronger team that is capable of overcoming the advantage the home team gets from their own court and crowd. There is no bearing on the post season at all. Don't understand the obsession with lamenting over close road losses and the fact that we haven't won the few tight games we have had this season and the last two tournaments.

[gtr2x]

Never heard of a "luck stat" before. That's a new one on me and I like stats. Seems pretty subjective tho.

[corpgator]

Originally Posted by gtr2x

Never heard of a "luck stat" before. That's a new one on me and I like stats. Seems pretty subjective tho.

Only the name is subjective. What it measures is performance in close games. Since looking at all of the data collectively points to close games coming down to chance and not skill, Pomeroy wanted to measure how teams performed in close games over a given season.

He said the trend he found was that teams that were "unlucky" one season were generally "lucky" the next.