The Secret Life of Stats

[Cyclical]

Good Simmons piece. Something I've been wanting for a while - if we're going to get into individual stats, let's get some meaningful ones. Not holding my breath.

I want to know Wade's percentages on contested, wide-open and clock-saving threes. I want to know how many uncontested jumpers LeBron creates for teammates. I want "mega-assists" (passes that create a layup or a dunk) and "half-assists" (for each made foul shot). I want "unforced turnovers," like in tennis (Tony Allen would be Wilt Chamberlain in this category), and "nitty-gritties" (some combination of charges taken, deflections, balls saved from going out of bounds and rebounds tipped to teammates). I want "Unselds" (a long outlet pass that leads to an assist for a layup or a dunk) and "Russells" (a blocked shot directed to a teammate).

You know who'd fall way short in Russells? Dwight Howard. He slaps everything out of bounds. Congrats, Dwight, you just gave the other team the ball back. Why are you smiling?

http://sports.espn.go.com/espnmag/story?id=4011524

[Spin Move]

I really liked the article, its got a really boyish charm to it, which is a slightly different, less cynical tone then most of Simmons writing.

[MyInsatiableOne]

Funnily enough I liked this column even less than most of Simmons' articles...perhaps because I don't place as much in stats/numbers as a lot of people tend to...

[Banks2Pierce]

I went to that MIT thing. It was really cool. I could've listened to the basketball analytics thing for days. Too bad it was only an hour. It was hilarious to see Cuban glare at Simmons every 30 seconds. He's also dead-on about Morey being dork elvis. Best part of the day comedically was a bunch of the people struggling to get beers down at the reception afterwards.

Statistics are cool and should be in the toolbox of many things used to evaluate NBA players and potential NBA players. They are not the end-all, be-all, but they are up there in the levels of importance. It is nice to have a gm that is open to a ton of different ways to make the team better, whether it be brain typing, statistical analysis, or whatever. Guy is a competitor to the max.

Think I saw your name in the program, bfb. Were you there?

[GuyClinch]

82 games tracks passes that lead to dunks or layups...but I digress.

Problem is even if you had all those statistics it would still be tricky to use them in some meaningful manners. Guys like Simmons and other journalists would just isolate one loathesome statistic for some guy they don't like and go 'see I told you so.' Dwight Howard sucks see how he is the lowest in "Bill Russels" he bats all those balls out of bounds..

We have to keep statistics of EVERYTHING and then looking at one happens to entire team decide the value of each players. It's a daunting challenge.. Problem is guys who have different roles on the team effect the entire teams stats. If your on a unit with a bunch of great rebounders your rebounds will suffer. If your playing on a team with a ball dominating PG your assists will go down and so on..

This is why I would take a skilled basketball observer like DA over any set of statistics. Because mentally the human mind manages to do this trick better then statistics near as I can tell..That being said i think statisics are still broadly valid - they just fall short of an EXPERT observer.

[GreenDreamer]

I place a lot of importance on stats. I find that the whole "My eyes are the best tool" approach is particularly flawed. In the end, a person like that is prone to being like Archie Bunker - listening only to himself, or those who say what he says, and never really understanding anything. I'd much rather be right about something than merely be told that my opinion was right.

A guy like Iverson is a perfect example of this. I used to love the guy too, but the qualitative stats pegged him as a loser a long time ago. His moves to two different teams have only served to reinforce that, with the last trade being the most conclusive. Billups has long been a darling of the advanced stats, and Denvers play since his arrival, along with that of Detroit since his departure, has only served to give validity to the approach of these systems.

My personal favorite stats are net plus/minus, oncourt/offcourt plus minus, and win shares. You have to use a bit of sense when looking at these, but they can be very helpful. Plus/minus might not be a true indication of how absolutely good a player is, but it is an excellent indication of how good that player is for his team.

When your eyes are telling you the same thing as the stats, then all is good. When the two are different, then an intelligent person will look more closely. Unfortunately the stupid far outnumber the smart in this world. Mark Twain put the ratio at 9:1, stupid to smart, but he was a bit cranky and may have been overstating it a bit.

[GuyClinch]

A guy like Iverson is a perfect example of this. I used to love the guy too, but the qualitative stats pegged him as a loser a long time ago. His moves to two different teams have only served to reinforce that, with the last trade being the most conclusive. Billups has long been a darling of the advanced stats, and Denvers play since his arrival, along with that of Detroit since his departure, has only served to give validity to the approach of these systems.

Nah you pegged him as a loser and looked for stats to support that. AI went to the finals without another all-star and in fact beat a very tough Laker team at the time in one game. For AI its about having the right kind of players to surround his special kind of talent with..

That being said he has slipped in recent years of course..

[threrf23]

I've started a few statistically inclined threads on this board. Obviously, I am a fan of statistics. I would have been interested to attend the MIT conference.

Anyways, I am a fan of statistics, but if I could cite one difference with the statistical community, it would be the same difference I have with most economists (and more generally the subject of economics). You just can't, or IMO shouldn't, take something that is clearly not an exact science and treat it as if it were an exact science.

I know the objection here is that nobody in the statistical community actually views basketball as an exact science, they are just aiming to analyze data to support other existing facets of player/team/etc evaluation. But that's besides the point. I am talking about the objectives of their statistical analysis and the way they treat the subject when seeking to meet those objectives.

Its like, engineers tend to be inherently programmed to create the perfect product from a functional/capability-wise standpoint, and in their quests to create the perfect product (which usually doesn't exist) they have a well established tendency to overlook usability issues that can be key to marketability. I'm used to dealing with inventors/entrepreneurs, and they tend to be instinctively trained/encouraged to be as opportunistic as possible. They'll often go after mass market success (which is often not achievable) and in the process its easy for them to overlook small niche markets which can be profitably targeted if done right. When statisticians begin using their computer software to find the perfect (or as close to perfect) formula which often doesn't exist, its easy for them to overlook some of the objectively solid and meaningful info which can be extracted from available data.

Now, to be honest, that paragraph above just sounds good, and I'm not sure its really like that. But its something to that extent. I do have to say, also, unless I misinterpret Bill Simmons, I actually think he hits this head on in his article. That impresses me.