Playing to talent or analytics?

[Risk Addict]

I’m was curious what the board thought about analytics vs taking a more nuanced team building approach. I work in a field that is very evidence based. There are best practice guidelines that we follow to optimize outcomes. I get the concept of using large data sets and odds to drive decision making. But there are also a lot of times where you have a unique situation with multiple variables that the data did not really evaluate.

Should we instead be using player level analytics rather than league wide analytics?

If Demar shoots 33% from the 3 that’s 1 points per shot and if he shoots 50% from 2 that’s 1 point per shot. Easy. But what if we have Drummond in the game that has an offensive rebound rate of 25% on 2pt shots and 10% on 3pt shots. That should change decision making.

I feel like those types of game play decisions were obvious before and now there seems to be this blanket mantra of shoot the 3 ( unless your pps difference is significantly better at 2s like Drummond).

Same with drawing fouls. I’ve notice that since Zach has been making a concerted effort to draw fouls, his efficiency on drives has plummeted. His turnovers on those drives has increased. He complains about no calls rather than running back in defense. Compared to coby who is often times avoiding contact (like Zach did in the past). I feel like the personal aspect has to be considered when coaching a guy to avoid or try to draw fouls. There can’t be a cookie cutter game play for everyone.

Lastly, the analytic darlings are now getting paid based on these data. I think teams who are creative with their roster and get talent at a discount and allow them to play their game has a real advantage now. It also makes them harder to guard. If 90% of teams are running the same schemes around macro analytics, it makes it harder to anticipate what the 10% using micro analytics are gonna do in certain sets.

From a spectator standpoint, I also feel like the league has gotten boring. You used to see different looks from different teams. You have run and gun transition teams, high 3 point volume teams, teams that work out of low post, pick and roll…. You got a new style of play each night. Now the games all blur together. It’s just boring and redundant to me, even if the shots are falling.

[MikeDC]

I’m was curious what the board thought about analytics vs taking a more nuanced team building approach. I work in a field that is very evidence based. There are best practice guidelines that we follow to optimize outcomes. I get the concept of using large data sets and odds to drive decision making. But there are also a lot of times where you have a unique situation with multiple variables that the data did not really evaluate.

Should we instead be using player level analytics rather than league wide analytics?

If Demar shoots 33% from the 3 that’s 1 points per shot and if he shoots 50% from 2 that’s 1 point per shot. Easy. But what if we have Drummond in the game that has an offensive rebound rate of 25% on 2pt shots and 10% on 3pt shots. That should change decision making.

I feel like those types of game play decisions were obvious before and now there seems to be this blanket mantra...

Do you have any evidence to back that up?

Seriously though, the NBA (and every other league) have always been copycats and following fads. Even before they had much data to back them up.

With more data, it doesn't really change. Analytics will show an opportunity or possible opportunity, and people will try it out. Over time, the more successful that new strategy is, the more people will do it. And the more effort people will put into figuring out how to defeat it.

[ChettheJet]

To me as team president, if you don't trust the VP, GM, pro and college scouting departments and other suits to look at players and say, this guy can handle the ball, this guy has a smooth stroke, this guy is a beast on the boards then you go with analytics and numbers, of everybody's stats, their vertical, their time through the slalom or whatever.

Is it analytics that keeps everybody saying that they don't want Zach Lavine, or do they not see the player he is when he's on the floor and they use his contact and injuries to ignore seeing him blow past defenders right to the basket or that he hits so many contested 3's? Do analytics show you where Luka can score 70 points?

I would rather bet my job on looking at players and how they play basketball to decide who I want and who is going to play well with the other 10 guys we've got. Ask yourself would Wes Unseld or Dave Cowens even get a look if the NBA had analytics 40 years ago? 6-7 and 6-8 centers in an era when 6-11 was about the minimum that they played against. I don't know, would analytics see Hall of Famers when rating them before they showed what they could do? they would both probably be sent home because they weren't big enough to be stretch 4's and they had no range.

Give me some former players going to watch games, sitting in a room watching video and pointing out who can or can't do what and asking if they can learn what they lack.

[MikeDC]

To me as team president, if you don't trust the VP, GM, pro and college scouting departments and other suits to look at players and say, this guy can handle the ball, this guy has a smooth stroke, this guy is a beast on the boards then you go with analytics and numbers, of everybody's stats, their vertical, their time through the slalom or whatever.

I don't understand why anyone would dismiss one or the other. Only an arrogant fool doesn't use every tool at his disposal.

You use both. And when they don't agree, you put it all together and try to figure out why. Anything else is just arrogance.

asking if they can learn what they lack.

This is probably the biggest thing, and it's something that neither players watching film nor numbers on a sheet of paper is going to tell you.

If I were a team president, I'd be bribing people and hiring PIs to dig into these guys' minds and figure out how they work and how they learn and what they're about. Almost everyone comes in having to learn quite a bit. Probably 30-50% of them never seem to. And it's not always for lack of trying. Like, one thing I respect about Zach is he really has worked hard on his game. He's developed his body and his skills. But... if I were a scout watching him at UCLA or as a rookie, there were plenty of things where I'd go, "ok, that kid just did a bunch of stupid ****, but he's got talent". He'll learn not to, like, drive headlong to the basket when he needs 3 points at the end of the game.

But it turns out he never has. Weird.

That was a funny draft. I'd say that analytics would have told you to draft Zach Lavine. Good shooter, can jump out of the gym. Very clear cut analytics case and it dovetailed with what you saw watching him.

He never learned, but he was still the right pick.

The wrong pick, remember, was Dog McDogMut. A guy who John Paxson famously watched when he got snowed in and stuck watching him. That guy had the smooth perfect stroke, and hey, he's a coach's son, so he must have some basketball IQ too. So who cares if analytics point out he actually sucks at everything but shooting and has the agility of a drunken old man. Turns out, he never learned anything either.

[2weekswithpay]

Ideally, you'd be able to use both effectively. Personally, league-wide analytics is the way to go.

When teams run unique offensive/defensive schemes, they usually have a unique offensive/defensive talent. The Jazz used to funnel all drives to Gobert, the Bucks gave open 3s to any middling 3pt shooter, and denied all shots at the rim with Lopez and Giannis, the Warriors' motion offense with Curry, or the Nuggets' mid-low post offense centered around Jokic. Even the Bulls were able to be different when they had both Lonzo Ball and Caruso playing.

Will trying to be different from the rest of the league while being led by talents like Derozan, Lavine, Vuc, and Coby be entertaining?

[Risk Addict]

Will trying to be different from the rest of the league while being led by talents like Derozan, Lavine, Vuc, and Coby be entertaining?

To be honest I wasn’t thinking of the bulls and entertainment, but the league as a whole. Before when you had a warriors game on the schedule, you were excited to see up tempo pace, good shooter ng, live or die by the 3. Then your next game is at the grindhouse in Memphis. Then team ball with good ball movement and set plays in Atlanta, where they over performed their talent. Pick and roll in Utah. To me the various styles were like the identities of the team. Everyone has the same identity now or is trying to. Again I feel like analytics are good but if it restricts your strengths you need to adapt rather than force it to fit.

Celtics championship year Rondo wasn’t taking a ton Of 3s. So if you had that team “adapt to analytics” I bet they would be worse and not win a championship.


I like data. I just think we need to look at individual data more than league wide data in season. For team building and post season exit interview goal setting yeah you can lean more on league wide analytics. But once your roster is set in November and your starter is still shooting 25% from 3P then have to adapt to their current skill set.

[League Circles]

I'm basically an eye test all the way type of guy, which is weird as an engineer. It's not that I don't value analytics, it's that I think the available measurements are pretty poor approximations of what you're actually looking for.

I've always thought that analytics tell you more about role than ability. They can tell you something about consistency, but I'm mostly interested in projecting and acquiring the best 5 man units with the huge presumption that they play as well as they can when it matters (the one thing most champions and contenders share in common).