The big issue with the use of a analytics is to translate a mountain of data into something usable in a real-time environment by the coaching staff.
Sixers coach Doug Collins is among the still-large contingent of heavy skeptics, preferring to manage the game and the players his own way to get the best out of them. Though analytics scream from the hilltops that 3-pointers, layups and dunks are the most efficient scoring opportunities for an offense, coaches like Collins have to apply that to their current roster of players. Just how efficient can your offense be if your wings can't MAKE the 3-pointers, and your bigs can't beat the defense to the rim?
Another example: data analysis tells you that a player A has a really poor shooting percentage in a certain area of the court. Easy solution: tell player A to stop taking that shot.
Sounds easy, but it might not be so easy in real life. A coach has to play five guys at the same time, and get them to think about the best way for the whole team to score every time. A coach also has to teach his players to react to the defense, get the ball to the right person and allow that person to make a play.
Sure, the offense can be designed away from player A taking that terrible shot, but what if the defense knows their best chance to get a stop is to entice player A into taking that shot? What if player A is only taking it when the offensive possession is down to Plan C and there's only 5 seconds left on the clock?
Alternately, it matters not that you design an offensive play to generate an open three or a cutter to the basket if the players you have can't finish the play a majority of the time.
Having said all that, I have to admit I'm a stat geek. I love data. The more the better.
But what I love more than anything else is to put that data into usable context, and to communicate that data to laymen in a way that they can use and understand it. I've made an entire career and garnered many promotions over the years thanks to that unique skill. Bosses love it. Users love it. Geeks love it.
So I can understand when the ESPN Truehoop guys wrap up the MIT Sloan session with one collective outcome: the problem is no longer about gathering data. The problem now is about translating that data into a usable manner by coaches who have to apply it in real time.
It's about having someone on your staff who can understand the mountain of data created off of 360-degree cameras that record every second of every play. It's about turning that data into information, and turning that information into applicable solutions.
It's about developing a working relationship between the stat geeks and the coaches, to work through all the scenarios of option A, option B and option C of plays, both on offense and defense to get the best possible results from the current roster. It's about providing the coaches with something they can use every day to help them make the right decisions.
It's about providing the front office with the missing skillset that the current roster needs. Given the collective skillsets of the guys under contract for next season, which ones are unnecessarily duplicated and which ones are truly missing? And, if that skillset was inserted into the mix, what else is lost because of the player they replaced?
And, to what degree are we comfortable with this analysis? Building a roster isn't just about getting the most skilled players. It's also about finding a good mix of skills, that make the whole greater than the sum of the parts.
From the Suns point of view, if you insert a P.J. Tucker type (all defense, little offense, best at o-rebounding and putbacks) into the lineup in place of a Jared Dudley type (average defense, limited offense, best at open threes and midrange jumpers), you know on the surface what you gain.
Is one guy better than the other? Depends on the situation: who are the other 4 guys on the floor? Who's on the floor for the other team? What's the score? What were the other 9 guys doing while Tucker and Dudley were going their thing?
All these things, and more, come into play when making decisions. Data can help here. Data can analyze all the situations in which each player has played, whose been on the floor and such. It's not just +/-. It's everything that happened around them. If a team can convert that data into some simple conclusions, a coach can use it when making the decision.
The NBA world has solved the data and the information part. The bottleneck is still in the translation to the players, coaches and basketball front office folks.
Make it simple. Make it usable.
Valley of the Suns talks about a new tool to assess players' health and endurance
ESPN's day two wrap
ESPN's day one wrap
Correction from the other day:
The Phoenix Suns have sent representatives to the MIT Sloan Sports Analytics Conference for years.
Former Assistant GM David Griffin attended in 2009 (and possibly 2008), while former GM Steve Kerr attended in 2010. Per another former staff member, the Suns sent at least one representative in 2012.
Now they are going "all in" with 6 representatives, covering both the business side and the basketball side.