In Sam Hinkie, the 76ers hired a guy who believes in advanced statistics, which ruffled some feathers from fans and analysts of the more traditional mindset.
But what exactly makes one an analytic?
If Doug Collins says before the season that he wants his team to take more shots from three point range and the free throw line, he's a traditional coach, correct? But if Sam Hinkie says he wants to improve our shot distribution and take more shots with higher expected values, he's an analytic?
If a coach says that his team gave up too many offensive rebounds, he watches the game. But if a GM says that the 32% offensive rebounding rate we gave up to our opponent was too high, he doesn't watch the game?
If a radio host says that, yeah, <X player> might only shoot 42% from the field, but he gets to the line a lot, so that helps offset his inefficiency, he watches the game. But if a blogger says that, with a true shooting percentage of 56% he's very efficient for a high usage rate player, that blogger doesn't watch the game?
If a baseball analyst says "yeah, he only got 47 RBI's last season, but he was batting in the 8-hole and didn't get many opportunities to knock runs in", he maintains his "uses his eyes" credibility. If TheGoodPhight says that Carlos Ruiz's RBI% is really high, then they should get out of their moms basement.
There's this divide that exists between people who use traditional statistics and those who use more advanced statistics. To be honest, that's partly due to the way we describe advanced statistics. At times, we do a poor job of explaining that these concepts, more or less, represent data and context that traditional analysts are already using.
Most people who follow basketball realize that a free throw doesn't go into the field goal percentage calculation and that, when comparing two players, if one player with a 42% field goal percentage gets to the line 8 times per game and another who shoots 42% only gets there 2 times per game that, field goal percentage can't be the only number used to measure efficiency. Analysts who use traditional statistics already have their own formula for how to figure out a players worth. They'll cite points per game, rebounds per game, assists per game, field goal percentage, free throw attempts, and will happily cite all of these figures when writing a column or talking on the radio.
Step one in bridging the divide is doing a better job of explaining exactly what these statistics represent.
We watch the games too
One of the first things skeptics will say is that "I prefer to watch the games". The insinuation, of course, is that people who use these fancy numbers do not.
As somebody who also works as a college basketball scout, this has always been a subject that has been near and dear to my heart. This simply could not be further from the truth.
When you see me with bloodshot eyes in the morning, it's because I was watching tape on Synergy Sports the night before scouting a prospect, not because I was organizing my spreadsheet.
This is perhaps further perpetuated by Hollywood, with the memorable scene in Moneyball where the Billy Beane character claims to not watch the games. It was, of course, a Hollywood invention. Also a Hollywood invention was the Peter Brand character, an overweight Jonah Hill portrayed as a baseball outsider. Of course, in real life, Paul DePodesta (the person who the Peter Brand character was based off of) started his career as an advanced scout for the Cleveand Indians, not a number cruncher. He was not the one who introduced sabremetrics to Billy Beane, and Billy Bean did watch games. But Hollywood was very desirous to perpetuate the computer-geek-who-knew-nothing-about-baseball myth, because that was the interesting story.
The reality is that there is nothing even remotely representing an either-or in this debate. There is not one (successful) advanced statistic proponent who does not spend copious amounts of time watching the sport that they love. Sam Hinkie watches more basketball than any person who is out there criticizing him.
As Spike Eskin noted, this is largely a straw-man argument. "I watch the game!" Great. So do we.
The difference is what metrics the two sides value. And make no mistake about it, traditional "watch the game" guys reference stats quite a bit as well.
I hear this mentioned a lot. "Statistics lie!"
EXACTLY. Now we finally agree on something.
Ironically, it's the people who look at advanced statistics who turn the most critical eye towards statistics. We absolutely believe that statistics lie. It's that belief that statistics are not very representative of what is happening on the basketball court that leads us to try to find better statistics to use.
And that's the ultimate goal of advanced statistics. Find something that is more representative of what is happening on the basketball court.
While people have been using rebounds per game as a way to measure how productive a player is on the glass for decades, there are deficiencies in that metric. How many minutes per game did a player play? How many rebounds were available to be collected? How did his team do on the glass while he was on the court?
Rather than throwing our hands up when we run into one of these situations where a statistic has gaping holes in its usefulness we try to find a better way to measure it. Context matters.
Statistics are better at telling you the what than the why
Statistics, both traditional and advanced, do a better job of telling you "what" than "why".
Some of the statistics I find the most useful, besides ones that look to correct glaring problems in traditional stats (fg%, rebounds per game, assists per game, etc) are the ones that tell you what. They're perhaps more appropriately named "situational" statistics rather than advanced statistics.
What I'm talking about here is the ability to find out that X player shot X% on corner three's, or shot x% and scored X points per possession used when shooting a dribble jumper off a pick and roll, or shot X% when making a jump shot over his right shoulder but 10% less when shooting over his left shoulder.
These aren't particularly advanced calculations, but they're invaluable in the data they provide. In fact, they're the exact type of things scouts will note when watching a player.
Where the statistics struggle at times is figuring out the why. Can Jrue Holiday's efficiency off the pick and roll be improved with a big man who will dive to the hoop and take the defensive attention away from Holiday? Why do two players have a negative adjusted +/- when on the court together? What of their skill sets don't mesh well?
It gets even harder when predicting future player development. Many advanced metrics will tell you the effectiveness of a 19 year old player, but what happens when that player develops a jump shot? How much will that open up the rest of his game? There weren't advanced statistics that would have showed Jrue Holiday would go from 30.7% at the shortened collegiate 3 point line to 39% from the NBA three point in one year. In order to predict that, you had to watch his form -- his balance, mechanics, follow through -- not look at the numbers.
The goal of advanced statistics is not to replace watching the game. The goal of advanced statistics is simply to find statistics that are more representative of what we're trying to measure. To recognize the flaws in current statistics and figure out ways to make them more telling.
Anybody you talk to who has worked with Sam Hinkie in the past will tell you how smart he is, not just about numbers but also about the game of basketball. People whom I respect very much, such as Zach Lowe, will tell you that he is among the smartest basketball people he has talked to. If you're putting down this hire just because he believes in advanced statistics, I urge you to take a long hard look at what advanced statistics are and what they're trying to accomplish, and stop making straw-man arguments such as using different statistics than you do means he doesn't watch the game.