Advanced metrics that gauge luckiness?
Hi, so as to keep this short and sweet I'll withhold most of the nature of my ulterior motives (except to mention that I'm trying with small amounts of effort to convince myself and others that I'd be thrilled with the long-term implications of OSU firing Tressel and hiring Dantonio--unlikely though all of that may be).
I simply want to know--point blank--whether any statistical metrics ("advanced" or otherwise) are respected for their ability to assess a football team's "luckiness" (e.g., whether Team A received a disproportionate number of "lucky bounces", perhaps at "critical times", relative to the field of Teams B through triple-Z). And, if they do exist, I'd like to know which they are and why they are.
Thanks in advance for the help.
I'm not sure about basketball, but I know in football you can usually determine how "lucky" a team is by whether or not they have a particularly good or bad TO ration compared to historical averages and the circumstances. So while a bad TO margin is somewhat expected with a young offense, Minny a couple of years ago had an amazing TO margin and went to a bowl game. The next couple of years they regressed to the mean and they were horrible.
I remember the Lions in 07 started out 6-2 and everyone was talking Super Bowl. Those 8 games, their turnover margin was like plus 15, maybe as high as 20. Unheard of type numbers.Teams were throwing the ball to untalented players who were playing centerfield, dropping snaps at the goal line. etc.
The second half it came back to normal: 1-7. The next year more of the normal, 0-16. You take away the luck element and that franchise could have been 2-30 over a two year period very easily. They had no players.
October 27th, 2011 at 12:27 AM ^
No. To think that you can gauge "luck" is a classic logical fallacy.
you meaure luck by looking at the difference between what the advanced metrics would predict the outcome would be and what the actual outcome was.
March 11th, 2011 at 10:31 PM ^
By that logic, advanced metrics predict the outcome of games perfectly except for the influence of luck - which is far and away the deciding factor in each and every game. I think not.
March 12th, 2011 at 10:45 AM ^
Why do you think not? This makes perfect sense to me:
Result = Measurable Skill + Unmeasurable Skill + Luck
What advanced metrics do is attempt to ensure that Measurable Skill >> Unmeasurable Skill. As the "Unmeasurable Skill" variable approaches zero, luck is the only thing left that causes variation in the result.
Michigan's luck is +.009 which ranks 159 in the country. They're slightly lucky but everything has been correct as KenPom predictions for the most part.
Ok, I guess it depends on your definition of luck.
Well, I heard this from the BTN postgame show: Michigan is now 6-5 in games decided by single-digits, compared to 1-8 (!) last year. It seems like we were very, very unlucky last season.
March 12th, 2011 at 12:45 AM ^
Or just mentally soft. We didn't have great leadership last year or a player who could run the offense and the team, so it's logical that we'd lose close games. Off of the top of my head, the OSU one was due to luck, but I can't think of many times where we just weren't outplayed down the stretch. Illinois is that team this year, they just can't close. I hate harping on intangible things (at least to the outside observer) but dang, Illinois sems like a dysfunctional team that isn't playing to its potential to me.
lol I love how I got negged for saying that measuring luck is a fallacy. If you are measuring luck, you aren't, it's something else. Go look up the definition of luck.
There was an excellent diary about this last year. Try a search. I'm on my phone now and if I get around to it I will look for it when I get in. I think The Mathlete (where are you btw??) was the aurhor. The answer was yes, and Northwestern was consistently 'lucky' and Oklahoma had the 'unluckiest' year.
Someone please give me an example of what luck is pertaining to a sporting event. Perhaps I am misunderstanding the question.
Good Luck: Your team won
Bad Luck: Your team lost
Here's the best definition of "luck" I've ever heard: Luck goes to the prepared.
Or, Bill Russell used to say this when he was working NBA games on CBS: The harder you work, the luckier you get.
I get where you are coming from, though, Jim: TRying to sound intelligent (even on this stuffy board) and discussing "luck"--it's an oxymoron.
March 12th, 2011 at 10:04 AM ^
Those posts most well align with what I am searching for:
That I'm effectively trying to find a spreadsheet of teams, rank, and numbers that **suggest** something about their luckiness. Id est, that suggests something about their apparent record relative to their mathematically conceived coulda/shoulda/woulda record (by the way, that asterisk-sandwich-as-a-highlighter-&-qualifier is for you, JLahey--with whom I agree apropos the nature of "luck" [But still: I'm trying to be practical here; so let's save the etymology and philosophy for later, if you will]).
As far as the archives and MGoSearchEngines are concerned, I'm not finding anything. Hopefully, this doesn't mean that it's actually there in spite of my tries, that I suck at feeding search engines efficient roots, and that I'm about to get embarrassed. But maybe it does. And I can accept that.
Again, thanks in advance for the help.
March 12th, 2011 at 10:44 AM ^
The best and simplest way to measure team "luck" is to compare their point differential with their winning percentage. Let's say we have 2 basketball teams--
TEAM A--72.3 points per game, 71.4 points allowed per game, record 19-13.
TEAM B--72.3 points per game, 71.4 points allowed per game, record 15-16.
There are 2 possible explanations for the difference in records--either (1) Team A has been luckier than Team B so far, or (2) winning close games is a skill, and Team A has more of the "winning close games" skill than Team B has.
So first we have to determine which is true, (1) or (2). How would you test whether winning close games is a skill? An obvious approach would be to see if it is repeatable from season to season--in other words, are teams who have winning records in games decided by less than 5 points able to do the same thing again the next season? If the answer is "yes", than winning close games is a repeatable skill and the difference is not due to luck.
I have news for you--the answer in every sport is no, winning close games is not a repeatable skill. Which means of course that winning a large number of close games is a result of luck, and (1) is true--Team A is a "lucky" team in terms of winning games.
So now all that you have to do to measure luck is to determine an expected relationship between points scored, points allowed, and winning percentage. Whatever is left over you may call "luck." Again, luck is, by definition, not directly measurable. It is what is left over after you remove measurable effects. But be careful--there really are unmeasured skills out there, so don't accidentally assume those unmeasured skills are actually luck.
Ok, fair enough. Thank you for the clarification. I am a philosophy grad at a school that is heavily geared toward logic and mathematics so I was being a bit of a dick.
March 11th, 2011 at 11:10 PM ^
I am a big fan of that inclination--in some ways I'm a very amateur philosopher myself. I'm slowly meandering through Kant's Critique of Pure Reason as an interesting coincidence; beyond that, abstraction is my favorite playground, yadda yadda, so on and so forth. I also tend to be very deliberate with my choice of words. This thread and that whole "luck" thing need not apply, of course.
Still, sometimes shit need be gotten done (or questions need be gotten answered), and when that's the case, the philosophers never get to the finish line first...they just most accurately and pain-stakingly articulate the conception of what the finish line is and isn't.
March 11th, 2011 at 11:58 PM ^
October 27th, 2011 at 12:10 AM ^
Though this is 7 months too late, don't do it.
Or if you really went through with it, crawl back to the law school on your hands and knees if you have to.
March 11th, 2011 at 10:27 PM ^
Luckier basketball teams generally have the token ginger at the end of the bench. Little bro has had a number of these red people over the past few 10 years which falls in direct correlation with their strong record over that time. Unconfirmed reports state that they could be leprechauns. Case in point Smotrycz has a little ginger in him and we have been luckier this year than last. Coincidence? I think not.
March 11th, 2011 at 10:34 PM ^
hmmmm not bad at all, can you put this theory into a formal essay and i'll present it to my profs?
March 11th, 2011 at 11:11 PM ^
Rather than write an essay, I will use bulletpoints in honor of the esteemed legal eagle Chris Cicero.
- Wisconsin has the kid with the ginger fro. They had the luckiest shot ever against us.
- Brian Scalabrine has no talent yet has lasted 9 NBA seasons and has played in 4 Finals series, won 1 title, and is on another contender this year despite having no talent.
- Bill Walton served as team leprechaun for the mid-80's Celtics playing few minutes but he undoubtedly caused the famous steal by Bird in the 87 Conference finals.
- State seemingly has a ginger every year on the end of the bench but they must not have one this year which helps explain their poor season.
These are just a few examples, albeit weak, I could put together an essay if I wasnt drinking or didnt have a life.
March 11th, 2011 at 11:04 PM ^
was the first entry after typing 'luck' into mgoblog search function: http://mgoblog.com/diaries/luckiest-teams-2009-updated-michigan-2008
Doesn't get real nitty gritty with stats, using only ppg, but is worth a read. Basically it treats events like fumbles as random events with an expected regression to the mean on a year to year basis, hence the 'luck' aspect. Happy reading.
But the difference between what the advanced metrics predict and what actually happens isn't all luck. I think bad luck in basketball has a tendency to correlate very strongly with bad free throw shooting; certainly the least "lucky" team I've ever seen (2008 Illinois, had three overtime losses and four more by 4 points or less) fit this profile, as they blew at least three late leads (and opportunities to win another game) by bricking free throw after free throw (notably, Pruitt had two free throws with less than 5 seconds left in regulation and a tie game against Indiana and bricked both, then same thing happened in the first OT).