A Discussion on Analytics – Casual Fan vs Expert

There’s always this immediate scoffing towards analytics. I’m not even referring to the stuff that takes these massive formulas to make such as usage percentage, win shares, etc. (We’ll get to those at some point.) Many of the analytics scoffed at are things such as points per possession on ISO plays or the fact that long mid range 2s are simply inefficient. Simple thing first lets talk about the concept most associate with analytics and that’s the mid range jumpers bit. First of all it’s pretty much common sense that the farther away from the basket you are the less likely you are to hit a shot. I don’t care how many half court shots you practice on you will never hit a better percentage of shots from half court (we’re pretending the extra point for the 3 point line doesn’t exist) than you would right next to the basket. It’s common sense that’s backed up by clear evidence proving that to be true. Farther away you get from the basket more likely you are to miss. Then there’s this chart here.

  
The chart is just showing the more efficient shots in the game and as you can see from the chart the longer 2s are less efficient. The extra point given to the 3 point line is why the 3 point shot is a more efficient shot than a long range 2. This chart is why the mid range game has died a bit. Now many outsiders like Charles Barkley for an exaggerated example will say things like “The analytic people are stupid because they refuse to take any mid range shots ” or something of that nature. Now no good coach that uses analytics heavily like Gregg Popovich for example outlaws the use of the mid range game. A diversified attack usually yields the best results if you can keep the defense guessing. The mid range area still is useful and should not be thrown aside no intelligent basketball mind believes that. They just acknowledge you can’t be a top tier team relying on that to build your offense around.
Another thing people seem to hate is other things that show efficiency that are not field goal percentage. Field goal percentage is an incomplete stat. For example if you shot 50% on 2 pointers and another person shot 34% on 3s what is more efficient? Well 34 multiplied by 3 gets you 102 and 50 multiplied by 2 gets you 100. That sort of thing has to be accounted for. Field goal percentage does not value a 3 point field goal attempt for the extra point it provides so it becomes in a way incomplete. There are other types of efficiency stats that provide that value of the 3 to it one is called effective field goal percentage and the other is true shooting percentage. Effective field goal percentage is very simple. It just adds the value of the 3 pointer to its equation. Nothing complicated. True shooting percentage on the other hand many may see me use consistently on Twitter, encompasses all of efficiency. This includes free throws which many have issue with. Without confusing the hell out of you the basics of the true shooting percentage stat shows the value of 2s and 3s, while additionally valuing free throws. Many have issue with this but if you get 30 points and all 30 points are off of 15 possessions in which you were sent to the free throw line is that not an efficient scorer? Essentially, the player in that situation only took up 15 possessions in which the ball ended up in their hands and hit all of their free throws. The point of scoring is to put the ball in the hoop and unless the free throws without my knowledge are taken away at the end of games it is important for them to be valued when it comes to scoring efficiency.
Now another thing I noticed in response to a thread of tweets I had about Kawhi Leonard and Paul George, showing who scores better in a certain play type, many seemed to have issue or a misunderstanding with the idea of points per possession. For example, Kawhi Leonard averages 0.99 points per possession on isolation plays. This is simply showing how efficient Kawhi is in this paticular type of play. I don’t even consider this to be analytics because it’s just a more detailed number showing Kawhi’s efficiency on a specific play. The reason people use points per possession rather than the field goal or effective field goal percentage is because that alone does not encompass his entire efficiency on those plays. Shooting percentages do not include whether Kawhi gets a turnover, foul, etc. That’s where the points per possession comes into play because if you have a turnover on that type of play that should be shown in your efficiency in that type of play.

Now the complicated formulas and such for things like usage percentage and so on. The main problem I’ve seen with these is the immediate dismissal of them and people blatantly using them wrong for example. Here this guy is using usage percentage to compare two players’ passing ability. 

So, the labeled purpose of usage percentage is to show how much a player is used in a sense of how much possessions end with them and before we had player tracking technology people used it as a way to show the player has the ball in their hands this much. Only issue with that in his comparison is that usage percentage has absolutely nothing to do with passing. Literally no part of passing is included in the number so using it in comparison of two players’ passing ability shows ignorance of what is actually in the formula. You cannot say Player A is a better passer because they averaged 14 assists with a usage percentage of 20%. There’s no correlation between those two things so it is entirely irrelevant. 

Win shares, VORP, and things of that nature are useful in the right context of what they are. Win shares is an attempt show how many “wins” the player is bringing to his team. There’s no issue with the number as long as it’s recognized that it directly correlates with wins. If player x’s team has 70 wins and 10 win shares compared to player y’s 50 win teams with 10 win shares then it’s in favor of player Y with the context of what the numbers mean behind it.
PER is a big one that people either love or hate even though it’s purpose is simple. PER is an accumulation of all box score stats put into one single number to show who has the better stat line. Nothing more nothing less.

Then my favorite phenomenon going on currently is the hatred towards per 36 minutes and per 100 possession numbers. Many look at per 36 and per 100 possession numbers as hypotheticals of sorts when that is not the interpretation intended for it. There’s not a hypothetical at all behind it. The per 36 measure is a way to show production per minute. 36 minutes is used as the independent variable because it allows us to have a better grasp of what that production is. The same goes for per 100 possessions just instead taking pace completely out of the picture. For example there is this tweet. 

Many take it as a hypothetical meaning that if the two players played 100 possessions this would be their numbers which is entirely not the purpose. This is what the players average per 100 possessions. It’s not a hypothetical because it is factual and the purpose of this tweet for me personally was to show that Kawhi and Paul George are essentially scoring at the same volume if you take the pace of both teams out of the equation. 

Another issue I’ve seen with the perception of per 36 minutes or per 100 possessions is people using things as examples to make them void when the sample sizes are vastly different. For example there has been a comparison done per 36 minutes of Marreese Speights and Nikola Vucevic. The numbers come out to this

Nikola Vucevic per 36 minutes 20.9 points 10.2 rebounds 1.2 blocks

Marreese Speights per 36 minutes 22.2 points 10.3 rebounds 1.5 blocks

The issue with this comparison is that the sample sizes completely make this irrelevant. Speights averages 11 minutes a game while Vucevic averages 31. People ignore that vital component but it would go the same for per game numbers. Brandon Jennings his rookie year in his first 10 games averaged 25.2 points 4.6 rebounds 5.9 assists on shooting splits of 48%FG 52%3PT 75%FT and a true shooting percentage of 57.9%. If he got injured for the entire season after that he would’ve been 12th in assists, 3rd in rebounds by point guards, and 6th in scoring. Would the fact that he only played 10 games not matter then? Case in point sample sizes matter so be aware of that. 

In the end numbers are just numbers. They give us evidence of some narrative and are useful as they are facts. Not acknowledging is refusing to acknowledge facts. Numbers are not everything and understanding the game on an eye test level is still by far the most important thing. I’m not here to tell you that numbers trump all else. I’m here to bring understanding of things that seem foreign to so many.  We cannot ignore them but they should be taken into account. The analytics evolution is taking over basketball so it’s important to understand the direction in which the NBA is going. 

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s