Thursday, December 16, 2010

Total team luck points through week 14

by Bruce D
(Team luck points = (bad luck points)-(good luck points), so negative numbers are the luckiest)
For a more in-depth explanation of what "luck" points are, go to a previous post here.
Luck is tracked to better analyze a team's true ability and to help in predicting results of upcoming games where some may not know what portion of a team's record and points performance was due to just luck.

8 comments:

Neil said...

I cannot replicate these numbers with the information given in your original post, what weight do you give to each 'luck' category?

Bruce D. said...

Neil,

The points for(+) the unlucky team, are the same amount of points against(-) the lucky team.


punts blocked=3
interceptions=2.5
fumbles lost=2.5
field goal miss/block=2.5
punt returns for a TD=4.5
ko returns for a TD=4.5

The data is gathered from yahoo sports, which allows sorting by these criteria, and provides the number of these types of plays.

Bruce D. said...

Another note to the above post,

The reason for lucky teams to be negative is, as stated here

http://community.advancednflstats.com/2010/11/randomness-and-luck.html

"In all these cases the "good luck" team is given a negative value, and the "bad luck" team is given a positive one. This is to point out what is more likely to happen in the future relative to current team performance, and not what has happened in the past"

Neil said...

Thanks for the response, Bruce. What would you think of treating all fumble recovery as luck? Right now your model only considers it to be lucky if the defense recovers. Wouldn't it make more sense to treat every fumble recovery by either offense or defense as worth half of a possession (1.25 pts)? This would basically treat every fumble recovery as a coin flip.
If this is a stat-availability issue, the team stats on nfl.com have an easier view of fumbles than yahoo does.

Bruce D. said...

Neil,

My(maybe wrong) thoughts are that fumbles that are not recovered by the opponent is just like a tackle, and the only way to determine "luck" is if its lost.

If you lose more than you get from your opponents, you're more "unlucky".

Over-all fumbles lost should even out, and if they don't, the difference should be luck.

This analysis is not based on what should happen, its based on what did happen. I have a hard time wrapping my brain about that many times.

Bruce

Bruce D. said...

Neil and others,

Don't mean to double post, but thoughts come as they may.

"Luck" seems to be the missing ingredient in the predicivity of the NFL. Since the nature of "luck" is assumed to be random, its nature is to end up even(to the mean?, average?).

So can we assume "lucky" teams can continue to be so lucky?

So can we assume "unlucky" teams can continue to be so unlucky7

If we don't assume that, then have to assume:

"Lucky" teams can't be as lucky.

"Unlucky" teams can't be as unlucky.

We all know what "assume" means(at least military trained), so take it as you will.

Bruce D.

Neil said...

I was thinking that fumbles are either dropped/muffed or forced and that recovery is luck--you are proposing that fumbles are always luck, so they only matter when possession changes. I can definitely see both sides: I see muffed punts of an oblivious blocker's backside as mostly luck, but most sack/fumbles as mostly the result of skill. In both of those cases, I think recovery is pretty close to entirely luck. I guess I could check to see if one model is more predictive than the other before arguing too much more about it. Thanks again for sharing the specifics of your model.

Bruce D. said...

Neil,

It seems that all fumbles are due to luck, but if a team loses more than their opponents, they're more "unlucky".

It should even out, and if not, it should be luck.

Zero lost fumbles is the same as 100 lost and 100 gained. this analysis is what is the actual over-all result.

A fumble recovered by the fumbling team is like a tackle, no?

Bruce

Post a Comment

Note: Only a member of this blog may post a comment.