I am the ignoramus and this is Part II of a discussion I had with the Steve Perrin about advanced basketball statistics. If you missed Part I go here. In it we covered Plus/Minus, On/Off Value and the delightful and potentially life-changing Adjusted Plus/Minus... (with guest appearances by Thomas Bayes, Shane Battier, and Franz Kafka.)
Steve Perrin started Clips Nation in 2006. He's not only the foremost expert on the Los Angeles Clippers, he's also a smart guy and he knows a lot about stats. Let's dive right in...
The Ignoramus: Okay, Steve, I think I have to give in to the idea that this is not a result-oriented search and just go with the flow. So, next, let's talk about Effective Field Goal Percentage (eFG) and True Shooting Percentage (TSP)... what do you think of those numbers?
SP: I'm glad you asked about those specifically. When people talk about advanced stats, they invariably start taking sides: the true believers think that the data is the only story that matters and the conventional wisdom is all wrong, while traditionalists tend to think that all advanced stats are at least deeply flawed and more like pure mumbo-jumbo and you just have to watch the games to know what's going on. I'm here to tell you that eFG and TSP are two stats that we can all agree about. In fact, I hesitate to even call them "advanced" stats since that term has so much baggage. To me, they're just better stats, but they're grouped under the "advanced" heading on Basketball-Reference.com so I guess that's what they are.
TI: I'm liking this so far. I like it a lot. You're giving me hope. May I assume both have something to do with how often the ball goes through the hoop?
SP: You may - or more specifically, how many points go on the scoreboard. Effective field goal percentage, also called adjusted field goal percentage (on some sites like ESPN.com) is simply a more accurate way of measuring a player's field goal shooting, taking into consideration the higher payoff on a three point field goal. It's easiest to explain eFG with an example: if Blake Griffin takes six two-point shots and makes three of them, his FG% is 50%; if Eric Gordon takes 6 three point shots and makes 2 of them, his FG% is 33%. But the goal is to outscore your opponent, and each player scored 6 points on their 6 shots; eFG simply accounts for the extra points contributed from made three pointers to come up with a single field goal percentage value that accurately reflects the scoring that resulted from the shots taken. In our simple example, both Griffin and Gordon would have an eFG of 50%.
TI: Beautiful. Really beautiful. An advanced stat I can really grab onto. What about TSP?
SP: TSP takes this a step further. eFG accounts for 2 point field goals and 3 point field goals, but there's a third way to score in basketball: free throws. Ultimately, a shooting percentage number is looking for one thing: how many points did the player score, and how many possessions did he use to score them. If a player is fouled in the act of shooting and misses the shot, that shot is not recorded in the box score as a field goal attempt and instead the player goes to the free throw line for 2 (or 3) shots. But even though it's not a field goal attempt in the box score, from a possession standpoint it acts the same - our team had the ball, and now we don't (barring an offensive rebound of a miss). So TSP figures that for every free throw a player takes, that's about the same as .44 of a field goal.
TI: Why .44? Where'd that number come from? Was it Thomas Bayes again? That guy's determined to crush my mood—
SP: --If every trip to the line was for 2 shots, then it would be .5 - two free throw attempts is one possession of the basketball just as one field goal attempt is one possession. But there are a few more free throws tossed in over the course of a game: and-ones, technical free throws and three free throws after a foul on a three point attempt all skew the 1 to 2 ratio of possessions to free throws. .44 is the number to use to account for those extra free throws based on historical NBA data. It's a constant used in every TSP calculation (and other possession stats), and as such it will not be completely accurate for every player - a player who finishes a higher than normal percentage of and-one plays for instance is being short changed by the .44 value - but it provides a good approximation.
TI: So, I guess I should just accept .44 and move on, right?
SP: Pretty much. Now, you'll sometimes see a third, related metric used: Points per shot, or PPS. PPS is just what it sounds like: it's the number of points scored per field goal attempt. Ignoring free throws, a player with a eFG of 50% would score one point for each shot, or a PPS of 1. PPS therefore has the virtue of being really simple to interpret at a glance - anything less than 1 is not good, and the higher above 1 the better. However, PPS is not nearly as informative as TSP—
TI: --In your opinion!
SP: --Since PPS does not consider FT attempts, it fails to recognize the lost possessions on those free throws. Consider Corey Maggette—
TI: Corey Maggette?! You mean... this guy?!
SP: Maggette was the NBA league leader in PPS for a couple of seasons because of his ability to get to the free throw line - but 20 points on 8 field goal attempts and 16 free throws is not as good as 20 points on 12 shots and no free throw attempts even though PPS would be much better for the first example (2.5) than for the second (1.67).
TI: Wait, wait! 20 points on 8 and 16 is better than 20 on 12 and zero because it takes fewer trips down the floor, right? So, fooey on PPS?
SP: I never use PPS, because it is an incomplete metric, and because TSP is simply better. I will sometimes use eFG% just to isolate on field goal percentage, and not penalize a player for not getting to the line, or worse for being a poor free throw shooter. DeAndre Jordan is one of the few players in the league who has a better eFG than his TSP.
TI: I'm glad you brought up DeAndre Jordan, because I follow him on Twitter.
SP: Yeah, well, because of the .44 factor, if you shoot 88% from the line, it's like making 100% of your shots in calculating TSP. But Jordan shoots so well from the field (68.6% on mostly dunks) and so poorly from the line (45.2%) that his TSP actually drops down to 64.8% when you take free throws into account.
TI: Oh. Okay. Well, I'm exhausted. I think that's enough for today.
We'll continue our
lesson discussion in the next installment when we study "estimated" stats. Hopefully we'll also get to "ratings", and the "single number" stats... though I'm not making any promises (Clipper Steve really likes to yammer). By the way, Wins Produced and PER promises to be a doozy, as Perrin takes on Hollinger and that Berri guy. Steel cage match, fight to the death... fasten your seatbelts indeed!