An Analytical Approach to the Rockets 2013-2014 season

First, unless Harden's replacement has a history of zero WS, you can't ignoring replacement player's WS if Harden is hurt. Other players' usage rates will also go up (Lin and Bev are left with bulk of ball handling duty). Higher usage rate = higher WS. It's not as simple as just deducting Harden's WS if he is injured.

Second, not sure if DRTG or ORTG are the liable platform to project the future results. Those ratings are as accurate or, as in this case, as biased as the person who invented them. Dean Oliver derived those ratings by using a set of complex formulas taken straight from the box scores. His calculations favor players with high eFG and if you go through the list, you will find players with high eFG's usually have the highest ORTG. Harden's ORTG took a considerable dip from OKC but we all know he is a much better offensive player, which doesn't reflect his value by looking at his ORTG's. GSmith had a huge difference in ORTG and DRTG, but in actuality his on court +/- was only a +2. In short, those ratings failed the conventional wisdom test.

Third, 'A' for effort and passion.

 

I can't tell if this is like, satire, or if jtr is really that full of himself.

How is this "the most realistic appraisal" of the Rockets' season? You have no idea how well they're going to play - Howard changes the entire dynamic of our offense.

 

I think you misunderstand what ORTG is. It is not a measurement for how "good" a player is in offense. It's a measurement of their offensive efficiency. Harden's efficiency went down last year (objectively speaking) while his usage went up.

 

I'm sorry JCDenton but your post was among the most nasty I have read here. You should be truly ashamed of yourself for that. What the OP is engaged in is analytic speculation about the upcoming season. Nothing whatsoever wrong with that and in fact it's a nice distraction at this time of the year when threads are largely devoid of fun facts. Since it's the upcoming season there is nothing to do but speculate at this point. My suggestion for you JCDenton would be abstain from threads that annoy you so much and chill out.

 

JTR writes very interesting, well-thought posts based on excellent analysis and research. But, yeah, he really is that full of himself.

 

yeah, how about actually reading his post before you reply, it would go a long way towards making you look like you actually have a brain.

JCDenton: If anyone here is getting banned anytime soon, it wont be jtr, it will be you. As it is, we consider you a joke...a bad one, but a joke nonetheless. How bout you stick to your lameass threads and leave the real posters alone.

 

LOL i did not know JCDenton was still around

 

This statement is presumptuous and false.

And people complain about me making predictions??

 

If you base your calculations on advanced statistics from basketball-reference, it's difficult to have accurate projections. What basketball reference is doing is distributing team's Pythagorean wins in accordance to Dean Oliver's formulas. Players with higher efficiency will naturally be awarded with more wins, which is somewhat ignorant and ignoring the impact a player has while he is on the court. Players like Battier will receive the short end of the stick.

There are actually "stats whores" in this league whose foremost importance is their own box scores, or during contract year, they know how to pad up their WS or PER. Hickson, Blatche, D Lee, Calderon just to name a few. Those will slip through the crack of his formula.

My suggestion is to use RAPM to make future projections. Some used -3.5 to -2.5 value as WARP (Wins Above Replacement Player)
If you are really interested in advanced statistics, here is APBRmetric forum. And here is a blogger (link) who used RAPM to project team wins, individual win contribution, and explanation on WARP.

 

I like in depth analysis with numbers. But it takes time to generate, so you don't see such posts frequently. But the fact you don't see it frequently doesn't mean you're the only one doing it, and I think it might understandably irk people to hear that.

I myself haven't posted anything like that because I'm still playing with numbers, and I haven't had anything interesting to add. I've become far more interested recently in ways to graph data rather than direct statistical analysis because it seems far easier to explain... much like your recent team graph. The human brain is really good at recognizing and processing patterns laid out visually.

The thing I have been experimenting with that I think has the most potential is turning data into an animated gif so you can see change as you vary a parameter. But, as always, it is hard to isolate things with the wimpy data available to amateurs.

 

Wins depends on efficiency and possession usage. If a player is very efficient on offense but has a very low usage, then he could be given less offensive win shares compared to a player who is less efficient but uses more possessions.

The Dean Oliver / Basketball-Reference model says that a player's contribution to his team can be viewed as a product of how efficient the player is and what percentage of his team's possessions he is responsible for. There's nothing inherently wrong with that approach; its just that, being based on the box score, there is a limitation to how precisely we can capture the player's "efficiency" and "usage" and assign credit accordingly.

 

Constructing the .gif turns out to be easy. There are even sites on the web that take a sequence of images and spit out the .gif for you. You just need to generate a sequence of images that are all the same size and plug them in.

The hard part is coming up with the sequence of images. I've generated some interesting ones, but the data is too granular and bounces around too much for them to be usable at this time.

 

I think the issue with historical data is that very small changes can make big impacts. Coaching for instance can dramatically affect the +/-. A team can "gel" meaning just being better at playing with each other - can have an impact that individual stats can't measure.

I think this analysis establishes interesting baselines, and it's nice to compare actual results to what historical predicts, but I suppose there's not really any accurate way to predict the future anyway, so heck, I'll give the dude 5 stars.

 

RAPM does not short shrift Battier. His RAPM in 2005-06 was +4.2, for example, and his RAPM in Houston ranged from +1.3 to +2.1. A -0.3 actually sounds pretty right for what Battier is in 2013. He is older and slower now and not what he was back in the days. This happens all the time with players in their 30s.