[HBR Blog] Success Comes From Better Data, Not Better Analysis

I'm pretty sure Chris Jent "didn't read" either.

 

MOREY IS WRONG, WE NEED DWITE HOWARD!!!!1

 

but without good analysts you don't know what is "good data". it's always yin and yang.

 

The best analyst in the world can't do squat with bad data.

 

I don't disagree. No one is critcizing good analysts, just saying that every team has access to them.

 

This is pretty close to the discussion Durvasa and I had last year about the gathering of the data, and whether what was available to the public gave anything worth reading, I don't think it is a good indicator, there is so much more data you can gather, but it takes human eyes to gather it....very smallish details that make a large difference.

A matter of inches on a closeout, or rotation, someone has to gather that data, in an accurate manner and create an algorithim that helps make it relevent.

Very good read, BTW.

DD

 

Whatever data sets and data advantage you have, hoard it and make it proprietary so only YOU have it, since good analysts are easy to get now.

Everyone wants the "open source" and for the data to be available to the public. But that goes against having a competitive advantage.

Morey is NOT saying to be STINGY with data. I'm saying if its about winning games then its best to keep any advantages close to the vest.

 

You know, people who are like dancing and don't like reading.

Didn't you see the post I was responding to?

 

The Cavs also beat the Heat and the Lakers in the regular season. Any team can beat any team over a long 82 game season.

To win in the NBA or any sports, you simply need a lot of talent, great teamwork/chemistry, and a little bit of luck. That's the been the case from the beginning.

I appreciate good data and analysis, but sometimes the "eye test" is all you really need.

If a coach wants to make his player more accountable for his bad defense, show him a clip of his poor defense. There's no need to show him a stat of not challenging shots.

If a coach wants to criticize a player for his knack of bad shots, show him a clip of that after the game. That's what Phil Jackson does after some games versus Kobe, and Kobe can start out the next game passing for the first half of the game.

Sometimes coaches/others get so caught up in statistical analysis (and a great example is Spoelstra of the Heat) that they depend on them so much for their gameplan, and not have any creativity. They become so clueless DURING a game on how to counter when a team does something that isn't expected (like a zone for an entire series).

The game of sports is so awesome is b/c of the element of the "unexpected." And teams with great talent/players are able to do those unbelievable things, going beyond what stats can prove.

I mean, you really can't tell me Kobe is one of the clutchest players in the game by looking at every clutch stat there is. But to most fans, he's one of the clutchest player to ever play the game, and the most clutch of our generation.

 

1. Non one said talent doesn't matter. But the data can help get an edge in identifying the best talent.

2. No one is saying you can't show a tape of plays to players to motivate them.

3. Nothing about statisitcs limits creativity. And actually, statistics will be used to enhance creativity. Coaches keep doing the same old things, stuck repeating the old coaching methods because they are "right" way to do things. Advanced stats may show that doing unconventional, counter-intuitive things may actually help in winning.

 

Smoke screen.

 

Good post.

Let me say first, that I truly understand the "eye test" you talked about, the thing is, every single thing we see, can be transformed in data.

I will give you the Kobe example.

What do the "Internet warriors" (and unfortunately some media) use to evaluate a players clutchness???

FG% in the last x mins, right???

But how about putting this data in context, with... more data....

How many of those game winning shots were buzzer beaters???
How many shots were uncontested???
How many shots were heavily (more then one player) contested???
How many were bail out shots???
Was it a quick shot or a result of collective/isolation play (rushing shot against fundamental basketball)

Lets imagine that we have all of this data, it still doesn´t mean anything, because we still don´t know the same numbers for all the others clutch players from this generation.

And that is why what Morey said is absolutely true. The problem nowadays in gathering THE BEST data, is simple, it needs even MORE data to contextualize the one you have, and to get that endless data you need endless people gathering it.

That is why evaluating guys coming out of college is so complex, because even if you have all the right data about what the player did in college, just the fact that all the variables around him will change, you can´t put your data into context, so your prediction about his transition to the NBA can´t be based on numbers alone.

Don´t be fooled with the FG% in "clutch time", if you ask Morey in private, he will tell you why Kobe was/is a great clutch player, and why he would want the ball in his hands, and he will probably use other data that is not usually used to justify it.

And one last thing, there is no doubt that you need talent to win, no one is saying better data wins championships, contextualized data gives you better probability to sign/trade/draft the best player available

Basically, it is the data that Morey (and the "analytical faces" around him) has access to , that makes the large majority of the people on here think that if D12 (or any other superstar) signs with us, there is no doubt that Morey will get the pieces around him in place, DM is one of the best at incremental value/talent in the league, problem is, he was playing the game with a broken foundation to begin with.

Sorry for the repetitive bad written book

PeAcE

 

What data can justify Rubio as a top-tier talent by many NBA scouts? Or what data can justify taking HS school players in the lottery when they never play a college game? Data cannot quantify a player's potential, which is how a lot of the superstars/franchise players are taken/made. I mean, it's laughable to think that many people would even think about whether to take Dwight Howard vs Okafor?

Furthermore, there is no stat for athleticism, or ability to create your own shot or pass over a double team, or ability to have ice cold veins when the game is on the line and wanting to have that shot, make or miss. Those are abilities of superstars/all-stars that cannot be quantified by statistics.

And no one is saying data/statistics don't help, or else we wouldn't keep records of them. But there seems to be an over-emphasis on that nowadays.

Sometimes it's best to keep it simple. Get the best players you can get, hope for good chemistry and luck (no injuries), and you'll be in the hunt.

 

Data isn't used for telling the difference between Kobe and Keith Bogans, my eye balls are enough in that case too. It's for telling the difference between Kobe and Wade, Ariza and Lee, Brooks and Lowry.

Also, how do you imagine data is collected, if not through eye balls? That's what I find funny whenever someone brings up the data vs eye ball argument. The only difference between the two is one can be counted on, quite literally, while the other you are often just winging it.

 

There is plenty of data on European players.

As far as HS players, probably not as much, so you just wouldn't use data as much.

Keeping it simple is great, but it doesn't provide you with an edge over the competition. It's like if you're playing the stock market, saying "keep it simple, just buy blue chips" isn't going to give you a trading edge.

BTW, your last paragraph probably summarizes the philosophy a lot of mediocre teams follow...which is why they're mediocre.

 

And again, what data justify Rubio being considered a top-tier talent?

Some of the best players/superstars in today's games ARE directly from HS.

You're comparing a game that a kid can play, to playing the stock market? Lebron James dominating the 5'5 high schoolers on his way to being the #1 pick with probably a mediocre education in high school. Very poor analogy.

To your response to my last paragraph, aren't we a mediocre team?

The mediocre teams don't get the "best" players, they get the best player available--a HUGE difference I might add, so that's why they stay mediocre, ala the Atlanta Hawks with Joe Johnson this past offseason.

So what stat tells you Kobe should get the ball for the last shot?

And the eye ball can you tell Wade/Lee/Lowry were the better player this PAST year.

 

What do you mean what data on Rubio? The same statistics are kept in Europe as here. Hollinger, for example, has his own statistical system to project how European players will do in the NBA.

As far as HS players becoming great...so what? So we should scrap using data? That makes no sense. Using data doesn't mean never using judgment or "eyeballs."

Nope, it's a good analogy. (Analogies aren't meant to be 100% exact). Both picking stocks and picking players are data-intensive efforts. They are both highly competitive with incredibly intelligent people in competition. Every small edge we can get from data will be worthwhile. Saying "Derp...just take the best guys and hope for the best" isn't going to lead to a long- term edge.

Yes...we are a mediocre team, for now, partly because of injuries. But I would say Morey's moves have been well above average thanks in part to the data the Rockets have accumulated.

As far as your last paragraph...a data oriented approach would adivse to NOT sign Joe Johnson. YOUR simple strategy, which was "Get the best players you can get, hope for good chemistry and luck (no injuries), and you'll be in the hunt" is the exact strategy that led to the Hawks signing Joe Johnson.

 

Pull up Rubio's stats this past season in Europe and you know what I'm talking about.

Again, I never say scrap using data. I did say I appreciate good data and analysis. All I'm saying is they are being over-emphasized to judge/rate a player or his potential.

And I point to the example of finding superstars/franchise players. Data CANNOT quantify a player's potential, which is how we judge potential superstars nowadays. And superstars are what separates teams in this league, the contenders from the pretenders. That's the recipe to winning.

All the NBA contenders you can name basically all have one legit all-star/superstar on their team.

You mean to tell me picking Lebron James is a data-intensive effort? Or Dwight Howard? Now finding the prime Shane Battier is a data-intensive effort.

Well, I apologize for saying it wrong when I say "best player you can get." I don't mean best player available.

 

I think the overemphasis is on boards and forums where numbers gurus battle "good ol' eyeball test" guys. Morey has repeatedly stated they don't use data as the be-all end-all answer to the question "what is this player's value?", but when you see some of the arguments on this board, it's like people think he only assigns numbers to people and makes decisions accordingly. Lastly, I don't think there is necessarily an overemphasis on numbers since nobody knows the exact methodology any of these teams use in evaluating a player. Morey has stated he uses data analysis to see if he can prove or disprove what his eyes tell him, but doesn't use it as the only barometer. To what extent he uses it is beyond the knowledge of any poster on this forum (unless they happen to work for Morey in a data analysis role... ).