Game Thread GAME# 63: WIZARDS @ BLAZERS - MARCH 4, 2020 - WEDNESDAY, 7:00, NBCSNW

The data is so noisy that you may lose your hearing if you keep thinking of it as a legitimate sign of anything.

Thats my point. Its beyond noisy.

 

The eye-ball test is about as flawed of a way to validiate anything. Humans are great at confirmation bias which allows them to see things that align with what they already believe and miss (not see) things that go against what they believe to be true.

 

It's noisy because there isn't a strong correlation. If the correlation were strong, it would look real clean and have an obviously progression to it.

 

I really do want to understand your point, so here's my main question I guess:

When do you think it's more valuable to compare Portland's stats to other teams and when would it be less valuable?

 

Dude, you're obviously trying to push a narrative if youre going to sit here and say that visual observation coupled with team-based stats is more flawed than a statistical average of other teams that have a shit ton of noise and an utter lack of proper context.

 

Again, theres a difference between evaluating what style of play is best for the team based on what results in wins for that team, and comparing the quality of a team-based attribute in regards to the rest of the NBA.

 

Its noisy for some of the reasons I brought up, which you conveniently ignored.

 

I'm pushing the narrative that assists don't have a strong correlation to winning because not only is it true, but about 10 years ago I was stunned to find out it was true. Up to that point I had been repeating what I was told by coaches I valued and validating those points by using my confirmation bias only to realize that there were other ways to skin a cat than I was originally willing to consider.

The data really isn't that crazy, I'm not sure why it appears noisy to you.

 

Fair opinion, my friend. I've always been with you that all things being equal, a team that moves the ball early and often is much more enjoyable to watch on a nightly basis.

 

If you're looking at a list of team assists per game you can't really compare one team to another because that stat doesn't factor in pace of play or turnover percentage or anything else.

Portland is last in the league per game this year in:

Assists (a whopping 1.4 per game behind 29th)
Passes Made (no surprise)
Potential Assists (1.3 per game behind 29th)
Assist Points Created (3.4 points per game behind 29th)
Secondary Assists (actually 29th in this one)

Phoenix leads the league in assists but is also 8th in the league in turnovers per game and high turnover % so their assists are different than Portland who doesn't pass anywhere near them but also doesn't turn the ball over nearly as much.

 

Nice post, I like the data as well! If those things meant a ton, you'd assume we'd be last in the NBA standings as well.

Would it make sense that a team that passes more is also more likely to get assists and turnovers? I don't have the answer, just a theory to why there would be no correlation.

 

See this is where I get frustrated with you sometimes. No stat can be taken and matched exactly with the standings page. I'm factoring in a multitude of stats that all show Portland being horrible in. I've also showed that when Portland performs better in these categories over the last two seasons the results are a much higher winning percentage.

Not once have I ever said that Portland needs to lead the league in assists either. More passes could or could not lead to more assists or turnovers. I will say that it looks like Portland's assist to turnover percentages are actually better with more passes, so no in Portland's case it leads to being more efficient.

 

Sorry for frustrating you, not my intention.

We agree Portland has been horrible this year. The cause of them being horrible, we might see differently.

Do you think we get more assists because we're making more shots or because we're making more shots, we get more assists? The causation of assists has always been tricky in small sample sizes.

 

I had time to go back further...

Portland with more than 21 assists:

In 2017-18 was 16-4. That's an .800 winning %.
They were 33-29 in their other games which is a .532 winning %.

2016-17 they were 22-21 with a .512 winning %.
Without they were 19-20 with a .487 winning %.

2015-16 they were 27-10 which is a .730 winning %.
Under that leaves them with a 17-28 record and a .377 winning %.

2014-15 above 27-13 which is a .675 winning %.
Below was 24-18 which is a .571 winning %.

2013-14 above 39-12 a .765 winning %.
That means they were 15-14 below for a .517 winning %

2012-13 above they were 22-22 a .500 winning %.
Below they were 11-27 for a .289 winning %.


That's 8 seasons under Stotts with multiple different types of teams where EVERY single season they had a winning percentage higher if they surpassed 21 assists. In only one of the seasons was it even remotely close and that makes sense considering the team was .500 that season overall. So no matter what the talent level of the team if they got more than 21 assists they outperformed their winning percentage for both the entire regular season and when they didn't get as many assists.

How's that for correlation?

 

That's why I always check my data against both assists and potential assists so I know if it's just missed shots or lack of passing leading to assist opportunities. The data under Stotts strongly suggests the lack of assists are due to a lack of ball movement, not shot percentage.

 

Great research! So what is your conclusion?

One thing that jumps out to me is that Stotts' offense was over your 21 assist number the majority of games in his first 5 seasons, but not in 2 of the 3 three (correct?).

Now how do we figure out if assists lead to more made shots or if more made shots lead to more assists.

 

I like this! Do you have data on the % of possible assisted field goals that are made? That would be super interesting.

Great work!

 

So let me get this straight...

When you present data that shows there is no correlation between assists and team success based on all 30 teams it's indisputable evidence but when I show for a specific team that assist numbers do in fact have a correlation between a higher winning percentage compared to lower assist totals you automatically resort to having to qualify it?

 

I'm sure somewhere out there you could find that data but it's way too much work on my end to win an internet discussion and I simply don't have time to present that for you.

 

The simple answer to your question would be: No

I think you've proven that if the Blazers hit a magic number of 21 assists they are more likely to win. The why and therefore the significance is unanswered.

I'm glad you think that comparing rosters from different years is valid, I agree with you.