D’Antoni explains why he doesn’t like calling timeouts

I have a background and a degree in analytics and I want to remind people that analytics is not just plugging things into Google and seeing what happens. It's a way of thinking. Generally, in terms of quantifying things and using algorithms to output effect values that you then have to interpret. Because of that there is always some room for debate and interpretation. However...

It's a little crazy to just assume that Mike and Daryl don't have in-depth conversations about this. Most Clutchfans seem to think that they saw a number spit out by a computer somewhere and then decided to go with that. I mean, maybe. But probably not...

There is a lack of evidence for the anti-MDA crowd in terms of whether the Rockets would have won the games they lost if one more magical timeout had been called. There are times when Mike (or plenty of non-Rockets coaches) call timeouts. Therefore, there is data on that. Running an algorithm (of your choice) on the data should show any effect size or correlation that timeouts have on your chosen statistic (wins, points after timeouts, point differential for the rest of the game, etc.)

I personally don't like that we don't call timeouts sometimes... but it's just silly to have so much anger based purely on assumption. Unless we actually see the data that Daryl has, how can we just assume he's an idiot? I can't.

If that's true, it should be measurable in data. You should either be able to do a regression analysis on the effect of timeouts on plays immediately after or aggregated after the timeout, and also control for home/away, playoffs vs regular season, even opponent, and anything else you can quantify.

Let's say you work in analytics, or as a data scientist. Your job is to let the math give you effect sizes and then try to use your domain knowledge or understanding to try and assign causes. Generally, analytics cannot prove cause, they prove correlation. Likely, based on Mike's comments, Daryl has data suggesting minimal correlation between timeouts and post-timeout results. This is what Mike means when he says this. The question then becomes "why" that is the case.

What you are doing is trying to take narratives and expect data to follow them. It doesn't work that way. You can rework the experiment or the algorithm and try to output different effect sizes or values, and then try to assign reasoning. But you can't set reasoning and then will the data to follow it.

How can we make this assumption? It would be pretty easy for them to split data between post-season and regular season. We've seen other examples of them doing that (e.g. with foul calls). Controlling for major factors is what you learn in high school statistics. Why do we act like PhD guys are so dumb? Because it makes us feel smart.

I'm not sure what this means. See my response to kjayp above. If the data really is showing low correlation between timeouts and post-timeout results, you can't stomp your feet and say "it should show that for ____ reason." If you saw some strong correlation then you could say "OK, this is probably because the Rockets need these timeouts."

Now we throw emotional intelligence in here. Emotional intelligence is a real thing, though it can be very debatable how you define it. General, it's used to gauge a person's ability to create trust and respect from others. The most reliable EI/EQ measurements are going to be from the players, and as far as we can see, they've all been positive. As far as you trying to assign EQ as a factor making timeouts more effective, see my response to kjayp. If the data is showing a certain correlation / effect size, then you try to assign the reasoning. If the data really is showing low correlation between timeouts and post-timeout results, you can't stomp your feet and say "it should show that for ____ reason." If you saw some strong correlation then you could say "OK, this is probably because of EI, it helps the players."

You're going to be upset about this, but.. everything can be measured - at least indirectly. If something can't be measured directly, it can be measured indirectly by controlling for other variables. Physicists use these methods to identify the existence of particles we previously weren't aware of. Climatologists use it to try and tease out the effect of carbon emissions on global warming versus other things.

In this case, MDA is saying Morey had compelling data by measuring timeouts by their direct or indirect effects on other things in-game. Based on their findings, the effect was minimal.

We can't will that effect to be larger by coming up with more reasons why it should be so... we'd have to run some numbers ourselves. Ain't nobody got time for that.

 

Not the same when you have Kobe, Shaq, Pippen and MJ. Also Phil had great tactics to coach and motivate his guys.

 

It’s too bad we don’t have kjay or jw in the front office to educate Morey.

Rockets need to run analytics on message board organization runners vs. the guys whose actual job it is to do this stuff.

 

and nobody cares about your advice

watch portland who got swept. lmao

 

To be fair while Mike mentioned that the data doesn’t bear it out, he didn’t say how the data was collected. We are just basically supposed to take his word for it.

 

Why would Kerr call a timeout? MDA just explained that timeouts don’t work. Lol

 

@JW86 ran the numbers and concluded everyone needed to watch portland

 

I'm not really following what you're saying here...are you saying it was also a dumb strategy for Jackson, but was outweighed by his talented players? Or somehow not calling timeouts with that collection of players is actually a good decision, but for the rockets it's bad?

 

Whatever makes you sleep at night and keep ignoring the point.

Moving on.

 

yeah you made it a point for everyone to watch portland

 

timeout is overrated - MDA

 

Someone tell Dan tony that timeouts do not rollover into the following season......

 

Phil had ways to motivate his guys and MJ nor Kobe actually needed it to work their ass off and get it done. Harden needs it. The motivation in particular is what makes MDA come across as if he’s clueless whereas Phil actually had his whole Zen tactics outside of timeouts. To me it seems like Phil knew MJ nor Kobe or others needed it to right the ship, but maybe he also had this notion that it was useless. In case of the latter then Phil was foolish too.

 

Are you ****ing kidding me. Confirms what we all knew all along

 

If it can be measured, and the data shows that generally timeouts don't affect momentum (let's suppose that is the case), then on what basis are people complaining about MDA not calling timeouts more often to stop the other's team momentum? What is the evidence that this would help more than hurt?

 

Gotcha, appreciate the clarification.

 

This is something I often think about with all the complaining that goes on around here.

Of all sports franchises on the planet, I feel like The Houston Rockets don’t do anything without a statistical deep dive. The amount of reasearch that goes into all the decisions they make has got to be huge, but a lot of posters here would make you think they haven’t even considered other options.

 

Yes, so this is the crux of the argument. We don't know the data and whether the interpretation of the data is correct. But we can't just sit here and say "the data should show _____". It's not so hard to run an analysis. Someone could go get play by play data on all of the playoff games this season and we could start to compare differential for 5 minutes before a timeout versus 5 minutes after, for example. Might be interesting! It would be a lot more interesting to discover something that way than to have more armchair GMs spouting unsubstantiated noise.

 

Incompetent management. I wouldn't give a **** if out star player likes him we won't win a chip with him being coach and capela still on roster.

 

It's funny people are trashing MDA for something Kerr does regularly.

Sometimes having faith in your guys is the wisest thing to do and improves them in the long run.

MDA has flaws. This one is a total myth.