Joe Johnson and Advanced Stats

Advanced statistics fans seem to fall into one of two categories. Those that discover evidence of what they already believe about a particular player, and then trumpet that stat, and those that just love them as a form of abstract geekdom, a way of exploring the "virtual" of a game we love. Those that hate them either seem to dismissively not look at them closely enough, or just don't like what they say about a player they cheer for and see with their own eyes.

The Advanced stat blogger The City came up with a very interesting way of measuring scoring greatness. Basically he uncovered a frontier line that exists between TS% - an efficiency stat that is supposed to measure the how efficiently you score - and USG% - a volume stat that expresses how much of the offense you take up. The reason why this relationship is supposed to be telling is that it is argued that there is a law of diminishing returns in scoring, ie at a certain point the more you shoot the less likely it is to score, in the broadest sense. You can see some of the detailed investigative work to check if this is true here at Count The Basket.

The result of The City's greatest scoring graph looks like this, a distribution of all of the great scoring (TS% vs USG%) seasons since the inception of the 3 pt shot:

via i180.photobucket.com

He calculates individual scoring greatness as the distance from the frontier line - you can see the red perpendicular lines in the graphic above. While I think there are probably better or perhaps slightly more interesting ways of creating a scoring picture the data is really fascinating. He restricted the data in calculating his frontier line to TS%>55% and USG%>25% which left out I think some very potent seasons, but when questioned he sent me a full list of the 1,000 greatest scoring seasons since the 3 pt shot with the minimum restriction taken off, and it is fair to say that at the very least this list seems to be a wide net capturing the preponderance of scoring proficiency in the league, even though it may contain a skew towards big men who have higher shooting percentages and may not have not best weighted offensive impact. At the very least we can say if you are an NBA scorer and aren't on the list at all your game has historical weaknesses or the entire statistical enterprise needs to be questioned.

What is amazing about this list is that Joe Johnson, for all his accolades as being one of the better scorers of his generation (6 straight All Star Games) actually made the list of the 1,000 best scoring seasons only ONCE. Wow... In that one season (2006-7) Joe Johnson was 25, playing for Atlanta, and had a USG% of 28.30 as well as a TS% of 55.80. It was the 459th greatest scoring season on the list. You can see some of the company that the lone Joe Johnson great season keeps here:

via i180.photobucket.com

Glancing over his stats it seems that the central thing holding him off the list was that is TS% was never something to write home about. We all know he was/is a high volume scorer, but perhaps what this data brings out is just how inefficient he has been. In fact this appears even historically bad if this data-picture has substance to it. You can read The City's description of the application of the scoring greatness frontier line in this post. Personally I feel PER (no pun) should be incorporated into the data - how about a z-axis - which may have given Joe Johnson's very good 2009-10 season more weight, but the facts remain what they are: Joe Johnson's TS% to USG% ratio is very poor for an elite NBA scorer.

If you want to sort and play with the data yourself just download the spreadsheet to Excel.

Even if you chaff at the statistical characterization of Joe Johnson above, there is one more harbinger against him in terms of his being a Net...and that is his age. An additional analysis that The City did with his frontier line was an Aging Curve attempting to capture just when players tend to decline. Below was the general distribution with the 30th year marked in red:

via i180.photobucket.com

You can read his discussion of the Aging Curve findings here. There are, as noted, exceptions and uniqueness to the shape of Aging Curves, but one can see even from individual charts the trend at just about now is downwards. Joe Johnson is 31.

So what is the story about Joe Johnson? At the very least - if we are to believe even the impression of the advanced stats - he has been a high-volume scorer who has scored at historically low efficiency marks in terms of TS% and he has already entered into a zone of historical decline. Does this mean he is a bad player? No. It means he is extremely high-priced and relatively inefficient as far as elite scorers go, and probably will become more inefficient as we go. He no doubt has lots of intangibles that a vet big-time scorer has, things that will simply win games for the Nets, but there is a picture here to be thought about. In a way the Nets had to make this move in order to leverage themselves into legitimacy - a big franchise defining move coinciding with the Brooklyn rebranding - but if we take the wide view they are vaulting into playoff and merchandising relevance on something of a long bet: that Joe Johnson will not decline as quickly with age as he should, and he is a much better scorer than the TS% story tells.