The Basketball Eye-Test vs. Analytics

As the NCAA tournament nears an end (RIP to my bracket and Kentucky) talk of the NBA, it’s upcoming draft and which college players have a skill set that will translate to the NBA is beginning to heat up. The analysis of college players is usually a combination of simple statistics (height, weight, age, shooting percentage, position, etc.), style of play and potential to improve. Still, as of late advanced statistics, some of which consider the pace of play of a college team, true shooting percentage (which factors in three point attempts that typically lower a players field goal percentage), and their Player Efficiency Rating have been used to analyze the impact of prospects on games and how this forecasts NBA success. The recent development of advanced statistics have been embraced by some parties and greatly criticized by others (Charles Barkley has been an outspoken critic). In many ways basketball is now fighting somewhat of an information cascade. For many years prospects have simply been analyzed based off of how decision makers judged their skill sets and measurable traits (vertical leap, height, passing, scoring, defensive ability, etc.), however now decision makers are finding ways to put these measurable traits into numbers which judge how these traits are influencing the game and just how good a player is. However with growing popularity of advanced statistics there is a new crowd being developed which decision makers are investigating on whether it is worth following. The greatest most sensible objection to statistics in basketball is that basketball is a sport which has very few isolated plays and accounting for externalities to these statistics is almost impossible. On the other hand, these stats have done a good job forecasting a player’s ability in enough cases that it has started a debate. In my opinion of course use advanced statistics but they must be contextualized and are probably most effective when judging marginal or “role” players who are not the main piece of the team as it quite easy to see who the great players/super stars are. Advanced statistics can be used to see what players are compatible with great players whose talents are great enough to make up for the weaknesses of their teammates and have a skill set that is great enough to highlight specific talents of other players (LeBron James, Kevin Durant, Steph Curry, James Harden to name a few). Granted there are not enough of these superstar players to go around, so teams devoid of superstars can build great teams using advanced metrics to put a wide variety of players with skill sets that can compliment one another (a la the Atlanta Hawks) in hopes of getting the most out of each player. In short, the best option is to merge both the use of statistics and advanced stats to determine how to put together a good NBA team — here is a link that makes a pretty good argument