Utah Jazz 2012-13 Preview - Part 4: The Inhumanity and Errors of Statistics

Thanks to the astute decision to bring on Andy Larsen months ago, I assume that I'm not the biggest numbers guy here anymore. After all, I took the bare minimum "maths" classes that I had to in order to get to where I am now. (Which was still a lot, I guess, in retrospect) Still, I continue to rely on numbers to help better frame observations in some sort of internally consistent working model of NBA Basketball.

I love statistics, but sometimes numbers do lie. Or at least, they fib. The numbers attempt to evaluate life, which is both very cold and distant, but also inherently weird. While looking over our team, coaches, and players I found a number of numbers that did not really pass the smell test. And because I don't want to be seen as a guy who dominates a discussion with just their point of view, these inconsistencies hurt all the people on the team. So let's look at a few . . .

• Last season the Utah Jazz averaged 91.4 possessions per game. This was good enough for 12th fastest paced team in the league last year. Do you remember the first game of the season where the Jazz were playing the Lakers who were on their third game in three nights and we didn't run? Do you remember any other part of the season where we'd walk the ball up the court and not take a shot that wasn't with 7 or less seconds on the clock? The Jazz were not a fast paced team last year. We didn't run. We played in the half court because that was the pace that best gave us a chance to win by using our best player at his strength. Then how did the Jazz manage to outpace so many other NBA Teams? Well, it's just the nature of the statistic . . . we got more possessions per game not because we ran like the Suns; we got more possessions because we were #3 in total Offensive rebounds (#2 in ORB%), and #8 in total Steals. The Jazz got the ball a lot, but were still a slow team.
• The offensive rebounding thing is actually a huge thing that skews a lot of our other offensive values. Don't get me wrong, I'm very happy that the Jazz won their battles on the glass. Their opponents only grabbed 1988 defensive rebounds, while the Jazz got 861 offensive rebounds . . . which means that the opponents ONLY got the defensive board 69.7% of the time. As a point of direct comparison, the Jazz got the defensive rebound 2055 times and surrendered only 728 offensive rebounds, which means we took care of business on the defensive end 73.8% of the time. Over the course of the season, the Jazz took care of business on the glass and that had a significant trickle down effect offensively. We just had more chances to score than we should have had.
• The Jazz also played more minutes in the regular season than every other team, except for the (surprise surprise) Atlanta Hawks - who tied the Jazz for 16,165 total minutes this season. That's total minutes times five though, for five players. There were 66 games this season, and each game is supposed to be 48 minutes long. If you multiply that by five (for the five players on the court) that comes out to 15,840 total regular season minutes. The Jazz (and Hawks) had +325 extra player minutes. Divided by five that's +65 total game minutes. And again, a game is only 48 minutes long. The Jazz, because of all the overtimes they played last year, played +1.35 extra games.
• What does this mean? It means that the season statistics for our team (again, particularly on offense) are inflated because they are the statistics for 67.35 games being divided by 66 actual games. As a result, our full season production, while accurate, appears inflated compared to what we're actually capable of doing in a 48 minute game. Sure, all the other teams also have similar statistical hickups on their final 2011-12 boxscores; so I guess that makes it okay. What it does not make is an accurate picture of how the team really is.
• This is also the case on defense, as defensively we're better than we looked because the team gave up 6536 total points over the time needed to play 67.35 games in a 66 game season. Over 66 games (which is accurate to how the season went) that's giving up 99.03 Points per Game. But over the actual time we played, that should only be 97.05 Points per 48 minutes. That moves the needle on the Jazz from "#8th most scored on team in the NBA" to "barely below average".
  
• I'm more than willing to admit that as a team we're barely below average on defense, and that our massive rebounding advantage made our offense look a lot better than it was. After all, you can't get better if you do not have an accurate frame of reference of what you ARE right now. The Jazz offense was effective, and they worked to their strength, but it was far from dynamic. And on defense the Jazz were damn good enough to EARN the right to play +1.35 extra games over the season. They didn't give up the Ghost, they played hard - they just had to play a lot of defense.

Effectively, as a team our group was both better than perception, and worse than perception -- depending on the quality being observed. We were still #9 in fg%, #27 in 3PT%, #4 in FTA, and #2 in Fouls. We were up and down, all over. Like Enes Kanter's worm dance move. Let's talk about some of the players now . . .

• "Derrick Favors and Al Jefferson have the same Defensive RTG, therefore they are equally good or equally poor on defense." Well, they both have a DRTG of 103, which isn't horrible. Last season Chris Kaman 's was 102. Dirk Nowitzki 's was 103. Carmelo Anthony 's was 102.Pau Gasol 's was 103. LaMarcus Aldridge 's was 106. And Kevin Love 's was 104. So . . . ballpark? Yes, I guess they are all pretty similar. But they all, for the most part, have different ways of coming about to that number. For example, Melo plays with Tyson Chandler, Dirk is always surrounded by good defenders. Pau plays for the Lakers and opposition teams get shafted against them. And Kevin Love doesn't defend well, but rebounds like a beast (which goes back into the previous issue of rebounds inflating possessions). Big Al gets a lot of stuffs on his man. Favors gets a lot of weakside swats. Per 36 mins Jefferson gets 1.7 blocks a game, and Favors only 1.0. But defense really is more than just these numbers. It's about the plays.
• If you look at the Synergy data they do not pass the smell test. Does that mean the data is useless if I don't like what it says? Of course not; but it's important to know that a tool is only as useful as it is accurate. Al Jefferson played defense in isolation 56 times last season, and Favors 51 times. Guys against Jefferson shot 50.0 fg% in this situations, taking 42 shots in 56 possessions. Favors? The man he defended only shot 35.0 fg% while they took 40 shots in 51 possessions. So, apples to apples it appears the Favors is the better ranked dude, right? Well, he is, his NBA Rank was #170, while Big Al was #194, and the PPP data also confirms that (0.80 vs. 0.84). So didn't the other thingy say Big Al was equal to him? And the other other thingy say Big Al was better? That's the problem, the more specific you look at something the less useful general information is.
  
• Overall in Synergy Big Al looks better, but that's because of another flaw -- a number of times Al gets beat, but his man is picked up by a rotating defender. Synergy logs the outcome of that situation to the help defender, and Big Al is absolved. Who do you think is a likely candidate to be a help defender on the Jazz? Yes. Favors. He (like Hayward) pick up extra defensive assignments by being good team mates and end up having to defend guys at an implied disadvantage. For example, Favors doesn't look so hot when defending guys off of screens or spotting up. Why? Because if you watch those plays that happens when he has to leave hi man and run out at someone he wasn't supposed to be defending in the first place. Synergy is far from perfect, but we kind of "know" Favors is a better defender than Big Al -- because we know Favors will go out there and defend more than just his man.
• Their DRTG is the same, Big Al is rated higher in synergy, and he gets more blocks and rebounds. BUT -- Favor's man shoots only 38.4 fg% overall, while Big Al's shoots 43.1 fg%. 82games has Jefferson's opponent getting a production value of 16.7, while Favors' man only gets a 15.0. The point is, the stats only tell part of the story, and you and I and Tyrone Corbin know who is better than who.
• Another example of how stats are deceptive could be Raja Bell's fantastic shooting (39.1 3pt%, 84.0 ft%, 46.6 fg%, 1.25 PPS despite only taking 25 free throws total) . . . . which looks GREAT. Except he next shot, passed up shots, and he hurt the team by doing so. In fact for the season he ended up with 1.0 offensive win shares, which despite his opportunity for playing time (23.4 mpg) was really low. Alec Burks played nearly 10 mpg less and had more Offensive Win shares, and combined Win Shares per 48 minutes. That's just what the numbers say. On the court we saw one of those two guys helping the team more than the other, and it wasn't the dude with the great shooting percentages.
• But numbers obfuscate more than just how we see teams and players -- they also help, hurt, or hide coaches. Tyrone Corbin has coached 98 total games (regular season and playoffs combined), and he has gone 44-54 in that span. That's only a winning percentage of 44.9%. Over an 82 game season that's winning only 37 games. We know that's a bad mark. I'm not an Ty apologist, pretty much a Ty rationalist though. Who know who's a really good coach? Larry Brown. Let's look at his win% the last few seasons: 32.1%, 53.7%, 42.7%, 28.0%, 65.9% . . . . so Ty Corbin has had a better average than Larry Brown in 3 of the last 5 seasons. So does that mean Corbin is a better coach? Or that he's a better coach than P.J. Carlesimo who has a career win percentage of 40.8% (lower than Ty?) . . . or is viewing coaching in terms of wins short sighted?
  
• I think the potential for a coach to win games is increasingly dictated by the GM's moves, and the types of players the GM gets for the coach to work with. After all, we all know that Phil Jackson is a coward who would never coach a rebuilding team that did not have two of the top ten players on it, in any given season. Does Tyrone Corbin even have two of the top 20 players on it? Nope. He does have to fight for his job security though, which means he's going to have to get some wins. Why? Because even the great Jerry Sloan was canned from his first NBA coaching job with the Chicago Bulls years ago when he went a combined 96-125 in three seasons. Oh, that's including playoffs too which gives Jerry a 43.4 winning % there. I can only imagine how the blogs would have wanted him gone too back then. Was Sloan a bad coach then? Or coaching bad players? Or coaching under a bad GM? Or a combination of all three? Probably the last one -- but after he started coaching HOFers for good GMs he became this epic figure.
  
• Ty's 44.9% is currently greater than Sloan's original 43.4% with the Bulls. Give Ty the Lakers front office, or the Thunder's talent and then perceptions of him almost automatically change as his player talent is better and wins more tough games. But the coach is the fall guy, I get that. Especially so in terms of statistics *and* the smell test. We do not like the rotation, the substitutions, and the game plan. So then, Ty must be a bad coach. After all, he's been the head coach for 98 games already. He is who he is . . .
  

Unless of course the greatest failing of statistics isn't the way they incompletely record the game, or incorrectly ascribe blame or value, or is it the inability to accurately relate cause to effect. The main problem of statistics is that these 'slice of time' views of the game we love fail to understand how to accurately project and predict the future. Stats are numbers based upon observations. These observations happen in a game. This game is played by people. People can grow, learn, gain experience, confidence, and improve over time. And that's not something stats can quantify right now. Kirk Snyder won the NBA draft combine. Greg Oden won the NBA draft. The Portland Trail Blazers won the NBA draft lotto. Players can win a game. Coaches can lose games. GMs can win free agency. And franchises win rings.

And all of that involves too much of the human element for a computer to accurately enumerate. So as a stats guy I have to say that sometimes you just have to watch the game. And I'm really glad we have games to watch again. (For me to obsess over in Excel . . . the circle of life)

The rest of the season previews will start this week, but keep this in mind as the grain of salt to temper the stats with. There will be a lot of stats, but stats aren't perfect. Neither are the players who create them. But you gotta hope that the human element means there will be some movement between who our players are 'right now' and who they could one day become.