Will George Hill's quality statistical performance follow him to Indiana, or was it merely Spur-ious (sorry)

On draft night the San Antonio Spurs made a trade: third-year guard George Hill was sent to the Indiana Pacers in exchange for the 15th pick in the draft (San Diego St. small forward Kawhi Leonard), the 42nd pick (David Bertrans) and the rights to 2005 pick Erazem Lorbek.  The move surprised me when I heard of it.  Hill had made such strides for the Spurs in the past two years, and was such a crucial, if not spectacular, cog in the machine that, let us not forget, produced 61 wins in the league last year.  But the general reaction around the internet was mostly positive.  Here’s John Hollinger:

So Hill goes now, before the Spurs have any drama over whether to extend him or risk losing him restricted free agency next summer.  San Antonio keeps its cap situation somewhat under control, and can plug James Anderson, Gary Neal, rookie Cory Joseph and whatever veteran backup point guard they sign into Hill’s former minutes without losing much in the backcourt.

Here’s Zach Lowe at the Point Forward:

It’s a risk, but it’s one a team in San Antonio’s position must be willing to take. Complacency can land you in mediocrity pretty fast, and Hill will be set to make at least a league-average salary once his rookie deal expires after next season.

And Jesse Blanchard of 48 Minutes of Hell:

In this instance Hill was in some ways a redundant strength. Granted, he was the Spurs best perimeter defender for the past two seasons, but he did so from a position that featured the Spurs best two scorers and playmakers… Of all the Spurs trade assets, Hill was the most desirable combination of youth, athleticism, skills, and contract the Spurs had to offer while still returning positive value.

What the Spurs lose is their best combination of shooting, defense, and competent (though not spectacular or dynamic) ball handling. But so far as skill sets go, Hill on offense was a dime a dozen shooting guard with some plus ball handling ability that fulfilled the role far better than most role players.

Timothy Varner (also of 48 Minutes of Hell) also contributed a very well-written (but hard to paraphrase) essay that argued that the trade was necessary to purge the impure presence of Richard Jefferson from the Spurs “pattern” (which is not a Scientology but a Wendell Berry reference — just go read it).

These are all good writers whose opinions I respect, but something felt wrong about these responses.  Given the current state of the Spurs, with their two rapidly aging stars in Duncan and Ginobili, the trade seemed at first blush to be exactly what Lowe described it as: a risk.  Upon further reflection, I’m not so sure.  Here are two reasons the trade could be bad, and one why it might be brilliant.

1) The Spurs have a very small window.

Tim Duncan is 35 and no longer the dominant player he used to be.  Manu Ginobili is still playing at a very high level (and is, I think, the Spurs best player) but will be 34 at the start of next season.  This is a team with a small opening for continued success as presently constructed.  If I had to bet, I would say that it’s extremely unlikely that the Spurs would have won the championship next season with George Hill, but I wouldn’t write it off as impossible.  Remember, the Spurs team that lost to Memphis featured a significantly weaker, injured version of their star player.  If Ginobili had been healthy for the entire series, could the Spurs have won?  And if they had won, could they have beaten Oklahoma City, and then Dallas, and then…?  Again, unlikely, but possible.  The Spurs are not in tear-down mode yet, even if their chances of winning a championship are slim.  Given this state of affairs, I would have expected the Spurs to make moves this off-season that maximized their chances of winning a championship in 2011-12, or at the latest, 2012-13.  And the trade of Hill for Leonard does not do that.  Why?

2) The development curve

It’s common sense (backed up with data) that most players follow a general development curve.  Generally, they start off as terrible, terrible rookies.  Then they get a little better, and a little better, and (again, generally) peak in their late 20s and start getting worse.  George Hill has followed this trend so far.  If we look at his WinShares/48 minutes, 1-year and 2-year Adjusted Plus/Minus numbers over his first three years:

Depending on whether you trust the one-year or two-year APM figures more, you could make a case either way that Hill improved or got worse over his first three years.  The WS/48 story is a little more straightforward — Hill had a bad first year, then got better, then got a little worse.  I agree with Blanchard when he says that Hill has probably plateaued, but it’s important to note that he’s still only 24 and could improve.  If he’s like most NBA players, he should still get a little better over the next three or four years, then start to decline.  And if he stays at approximately where he is now, then he is a very valuable player, and not quite as replaceable as some have made him out to be.  If you trust WS/48, then he was better last year than Joe Johnson, Carmelo Anthony, Monta Ellis, Rudy Gay and Jason Terry.  Of course, that’s unfair, because Hill is a lower-usage player than those players, but even if you stick to players who are more on the low-usage end, Hill is elite, in the .140-.150 range along with J.J. Redick, Arron Afflalo, Kyle Korver, and Tony Allen.  You may say that those are all role players, but they’re really good role players.  It may be true that you will never win a championship with Hill as your star, and that Hill probably can’t scale up his production with increased usage, and that he is probably just about as good as he’ll ever be, but even granting all that, I think it’s hard to deny that he is a much better than average supporting piece in the league.  Ask Dallas (or Miami for that matter) how important it is to have better than average back-up guards to spell your starter.

But Hill’s quality is only part of my point.  My main point is that San Antonio is essentially bailing on a development project after they have spent all the time suffering through the development.  They spent the game minutes, not to mention practice and training time, making Hill a better player, and now is when that investment should be set to really pay off.  You could argue that those are all sunk costs now, that all that matters is how much the Spurs got in exchange for Hill, but what the Spurs got in exchange for their completed development project was … a new development project.  Kawhi Leonard is clearly the unknown variable in this whole trade.  If he turns out to be the second coming of Scottie Pippen, we may look back at this trade and wonder why we ever worried about George Hill (although even Scottie Pippen didn’t really take off as a player until his fourth year in the league).  But if Leonard is like most non-superstar players, next year will be mostly a wash — even if he is a good prospect and a better prospect than George Hill.

What is Leonard’s ceiling, by the way?  Basketball Prospectus created similarity scores for all the big names in this year’s draft, and Leonard’s comparison player was Joe Johnson.*

* They noted that linking the outside shooting Johnson with the more interior playing Leonard was an odd comparison, but looking at Johnson’s college numbers, his FG%, FT%, rebound and steal numbers are not too far from Leonard’s.  Johnson was the far superior three-point shooter, Leonard the superior rebounder.

Pretty sweet player for a comp, right?  Well, Joe Johnson didn’t record a WS/48 score above average (.100) until his fourth year.  Leonard could buck the trend and produce right away, especially on the defensive end, but this would be unusual.  If he’s at all typical, there will be some growing pains.  He may get more of a chance than most rookies would get from Gregg Popovich due to the fact that he plays the same position as Richard Jefferson, but he will still need time to develop.  And the one thing these Spurs don’t have is time.  Leonard may blossom in three or four years into an elite defensive player and a solid, unspectacular offensive player (which seems to be the general opinion of his ceiling), but by that point Duncan and Ginobili may be retired, Parker will probably be gone, and we’re talking about a totally different team.

The transition from dominant team to non-dominant team to bad team is a tricky one in the NBA, because it’s impossible to know exactly when to cut bait on a team that no longer has a realistic chance of competing for a championship.  The Phoenix Suns, for instance, still haven’t realized  that the Steve Nash window has closed, and still cling stubbornly to the ticket sales and 35-45 wins Nash guarantees.  In a few years, that decision will, I imagine, look pretty foolish.  The Spurs are not the Suns, though.  They still have an outside shot at a title in the very near term, i.e. next year and maybe the year after.  The Hill trade, though, seems like a move of a team that is thinking about winning three or four years from now, not next year.  And that is a weirdly contradictory move for the Spurs to take.

Hollinger notes that this trade helps the Spurs with their cap situation at the end of next year, when Hill becomes a free agent, but my argument would be that the Spurs only have one year to worry about — this year.  If the Spurs still haven’t won a championship by next year, it may be time for a complete tear-down.  That year (the end of 2011-12) also marks the time that Tim Duncan’s contract expires, and if the Spurs choose not to renew him (unlikely) or he chooses to retire (somewhat more likely), then 2012-13 will be a complete rebuilding year anyway.  In the alternate universe where the Spurs had kept Hill and Duncan had retired after 2011-12, they would have happily let Hill go to free agency, because 2012-13 would have been a tanking year anyway.  Sacrificing a better shot at a last-gasp, improbable championship in 2011-12 for salary relief in a year where the Spurs may not even be trying to win seems like a poor exchange.  But like I said, there is one reason this trade could be brilliant.

3) The Spurs are wizards

George Hill certainly seemed like he was a very good player last year, but what if I am attributing too much to George Hill the individual and too little to the San Antonio Spurs, the system?  The Spurs have performed some astounding magic tricks in the past, making marginal players look good and good players look great.  Which players am I talking about?

• Brent Barry, when he joined the Spurs at age 33, was coming off of years with WS/48 scores of .141 and .151.  In his years with the Spurs, his WS/48 numbers were .168, .162, .201 and .203.  Once he left the Spurs to join the Rockets (at age 37), his production dropped dramatically, to .082.  He retired the next year.
• In his one year as a full-time Spur, Stephen Jackson recorded career highs in WS/48 and eFG%.
• Hedo Turkoglu, in his one year as a Spur, recorded a career high in WS/48 (.155, far above his next highest score, .140, and his career average, .121).
• Robert Horry followed up a .098 WS/48 in his last year with the Lakers with WS/48 numbers of .171 (a career high), .157, .145 and .125.
• Even Bruce Bowen (whose contributions don’t show up very well in WS/48) became an above average player by the metric, scoring a .111 at age 33 and a .108 at age 34.  This for a player whose high mark for another team was. 081 at age 26 for the Celtics.

There are more examples, I’m sure (Rasho Nesterovic, Malik Rose, Fabricio Oberto), of players who were never so good as they were with the Spurs.  Why is this?  Without getting into too much speculation, I’d wager that it has something to do with having a) a respected coach, b) hall of fame stars, and c) strong credibility from a) and b) that if players accept the discipline and well-defined roles of the Spurs system, the team will win.  Not all teams are so well-run, and many players who look, statistically, like all-stars for the Spurs look very average in the wild.  If the Spurs know that they have this effect on their players, they also know that George Hill’s true level may be somewhere far below his apparent level.  And if that’s the case, then the Spurs may have just pulled a fast one on the Pacers.  In this respect, the Spurs may be a bit like the used car salesman from Roald Dahl’s Matilda, rolling back the transmission and pouring sawdust in the gas tanks of their players, metaphorically speaking, and selling them for something of real value.  Of course there’s nothing underhanded about what the Spurs are doing — making your players look good is just one more perk of being a good team and a well-run franchise.

So which is it?  Did the Spurs get the better of the Pacers in this trade, or vice-versa?  While I will concede that George Hill will probably never play quite so efficiently as he did with the Spurs, the Spurs aren’t actual magicians.  Even in their most dramatic transformations, players don’t change their stripes completely.  Hill won’t forget how to shoot or play defense or dribble the ball when he joins the Pacers, and he will probably still produce as an above average guard and a quality perimeter defender.  Also, the 2010-11 Spurs are not the 2004-05 Spurs; Hill did not play on a team where Duncan, Ginobili and Parker had opposing teams so terrified that they were willing to leave a player like Robert Horry, Bruce Bowen or Brent Barry wide open.  Hill was more responsible for generating offense on his own than those players, and deserves more credit for his production.

But even if Hill suffers a decline next year, the timing and development issues still worry me.  Given the age of the Spurs stars, the most likely outcome is that Leonard reaches his peak long after Duncan and Ginobili have left the league.  The Spurs are taking a gamble, namely that Leonard is a true diamond in the rough and will be that rarest of things in the NBA: an extremely productive rookie.  Given that the Spurs are long shots to win the championship next year anyway, maybe that’s a justifiable roll of the dice.  It probably won’t work out, but the Spurs must think that the upside of the gamble on Leonard is higher than the upside on the sure thing of Hill.  I’m not sure if that’s sound reasoning (it’s the same kind of thing I said to convince myself that the Suns trade for Shaquille O’Neal was a good idea), but next season (if there is a next season) will provide a fascinating test case of the strength of the Spurs system, namely its ability to swap in new (and untested) supporting parts and still function.

One aspect of the epic comeback/colossal meltdown in the Miami/Dallas series that has not been commented on very much is how thoroughly Miami slowed down their offense in the last seven minutes.  According to the ESPN play-by-play, here are the number of seconds taken on the Heat’s last eight possessions before Dallas tied the game up with 57 seconds left: 20, 28*, 17, 26*, 24, 27*, 18, 24, 54.

* Some of these numbers are too high for a possession without an offensive rebound, so there must be something wrong with the play-by-play transcription here.  The big 54 number is the epic, two-offensive-rebound possession the Heat had starting at the two-minute mark.

Assuming that the possessions in excess of 24 seconds are actually below 24 seconds (let’s say 22 seconds), then the Heat averaged about 21 seconds a possession from 7:13 to :57.  On those eight possessions they scored two (2) points, for an offensive efficiency of .25 points per possession.  Dallas, meanwhile, scored 15 points on their nine possessions (they started with the ball after a timeout, and so had one extra possession), for an outstanding offensive efficiency number of 1.67.

Clearly, Dallas got unusually hot for that stretch, and Miami got unusually cold.  But what about the strategic move of “slowing the game down”?  Is it a sound move?  And exactly when is the right time to start your slowdown?

To answer this question, let’s assume one thing: Miami’s offensive efficiency down the stretch is representative of “slow-down” offensive efficiency.  This may be exaggerating the case, because Miami’s offense was exceptionally wretched, but for the sake of argument, let’s say that it’s true, i.e. that a team playing “slow-down” ball will average .25 points/possession.  Next, let’s look at two cases of the opponent’s offensive efficiency: a) Dallas last night (1.67 points/possession), and an average team (1.04 points/possession).  Lastly, let’s assume that the “slow-down” team averages 21.13 seconds/possession, and the other team averages the average number of seconds/possession in the NBA, which happens to equal 15.2 (I’m assuming that the team that is trailing is not hurrying excessively, due to its urgent need to score points on every possession).  Given a lead of a particular amount, how long would it take in these two scenarios for the trailing team to catch up?*

* The basic idea here is to compute the differential per possession (e.g. 1.67-.25 in the Dallas/Miami case from last night), by the average length of a full two-possession sequence (15.2+21.13=36.33 seconds), to get a differential per second.  In the case of Dallas and Miami last night, for instance, Dallas was catching up to Miami at a rate of .04 points/second from the 7:13 mark to the :57 mark.

As you can see from the graph, it takes longer for the trailing team to catch up when they are not blazing on offense a la Dallas last night, but even in the case of Miami versus an average offensive team, it doesn’t take very long for the trailing team to make up even large differentials: 7.7 minutes to make up a 10 point lead, for example, assuming the “slow-down” team plays with the offensive efficiency of Miami last night.  And that lag time shrinks to just 4.3 minutes if the trailing team plays offense like Dallas did last night.  And in the case of smaller leads, such as five points, the time it takes for the trailing team to catch up is quite short: 2.1 minutes in the ultra-efficient offense case, and 3.8 in the average efficiency case.

The point is that slowing your offense down at the cost of offensive efficiency only makes sense when the trade-off between your lead and time left in the game is right.  Teams that are only up a handful of points may be tempted to think they can run the clock out by eating up 24 seconds on every possession and then throwing up a prayer, but the clock has to be pretty far down and the lead pretty far up before this is a wise strategy.

The best way to win close games with a lead is, not surprisingly, the best way to win games in general: make sure your offensive efficiency exceeds that of the other team.  It’s generally acknowledged that starting offense early improves efficiency, which makes sense: it’s easier to turn down a bad shot when it becomes available with 10 seconds left in the shot clock than it does when there is 1 second left.  And if you maintain your offensive efficiency to at least near your average level, even if the team trailing is on fire, the slope of their ascent to catching you will be much more slower; for example, if you’re averaging .9 points/possession and the opponent is averaging 1.1 points per possession, it will still take 25 possessions for your opponent to make up a 5 point lead.  So, in a paradoxical way, the way to truly slow the game down is to continue to play at the same, not-too-fast, not-too-slow pace you did all game.  Merely by ensuring that there is not a massive discrepancy between you and your opponent in points/second elapsed, you will force your opponent to use more possessions to make up the gap.  Possessions = time elapsed, and time elapsed is your friend; the situation is analogous to the difference between holding an hour glass slightly on its side to make it take longer to run out, if that makes sense.

Here’s the quote:

Ask any NBA starter how many minutes would make them happy and they’d say 36 to 38 (one rest per half). There are 240 minutes available in a basketball game. That means you need to allot 180-190 minutes for your five starters to be happy.

I was curious, so I looked up the number of players in the NBA who have played 36 minutes or more per game this year. Here is the list in its entirety:

That’s 27 players. So… either 80% of the league’s starters are terribly unhappy, or Simmons is just wrong about how common it is for people in the NBA to play 36 minutes or more every night.  Among the unhappy players in the league: Russell Westbrook, Josh Smith, Al Horford, Paul Pierce, Dirk Nowitzki, Joakim Noah, Steve Nash and Kobe Bryant.

I could go on, but my point is this: very few players consistently play 36 minutes or more a night, and very few teams rely on more than one starter to take on that kind of load (the Miami Heat are the only team with three players in that top 27).  And teams can be successful that way!  The Bulls have two guys who play above 36 (Deng and Rose), the Celtics have two (Rondo and Allen), the Magic have one (Howard), the Thunder have one (Durant), the Lakers have one (Gasol); the Spurs and Mavs both have zero. I get Simmons’ point, which is that it’s better to have superstars playing heavy minutes than average players playing average minutes, but he exaggerates how easy it is to find five guys who can all take on that kind of load and remain effective.  Every team has to find a way to fill up the 19,680 minutes of a grueling 82-game season , and most teams choose to do that by not killing their starters.  I can even conceive of a universe in which an aging star with a bad back (like Steve Nash) or a sore leg (like Kevin Garnett) might even be relieved — nay, happy — to hear the coach say “we only need you for 30 minutes tonight.”  Considering that most teams set winning the playoffs as their primary goal, it might make the coaches and the general manager happy, too.

"Shavian", as in "Brian's head is looking quite Shavian tonight."

Frequently when writing about the NBA, one is confronted with the need to describe a player in terms of another player, either contemporary or historical. In these moments, many writers reach for the obvious — “Jordan-style”, “Magic-like”, or, for the more European writers, “Bird-esque”. In addition to being somewhat clunky, these adjectives lead to terribly non-standard writing. What we need is a handbook of official NBA player comparison adjectives. And here is a start:

Afflalonian – like unto the gritty defense and efficient scoring of Arron Afflalo.

Birdanic - having the skill and cutthroat intensity of Larry Bird; not to be confused with Birdmanic, referring to the exploits of Chris Andersen.

Brickowskotic - used to describe an otherwise average player who makes a surprisingly large contribution in the playoffs (see also “Hansensian” and “Homely”), after Frank Brickowski.

Dark – refers to the cast of an owner’s expression after he realizes he has drafted a player far too high; refers to Darko Milicic, related to Olowakandescent.

Fishisch - describes any player whose shooting percentage correlates highly with the importance of the game situation, after Derek Fisher.

Hansensian - similar to Brickowskotic; refers to former Jazz and Bulls guard Bobby Hansen.

Homely - similar to Brickowskotic and Hansensian, although with more of an emphasis on three-point shooting; refers to current Heat guard Eddie House.

Jordickish - refers to any former star who becomes sour and angry in his post-NBA days.

Laimbeeroise - refers to any player who gets in physical confrontations for tactical and not emotional reasons.

Manupulative – refers to any player who tricks referees into calling offensive fouls through flamboyant flopping; see also Vladulant.

Olowakandescent – refers to the brightness of the smile of a player on draft night after he has been drafted 60 spots higher than he ought to have;

Redickulous – the quality of exceeding widespread expectations that a player will flop; antonym of Thabeetan.

Shavian – referring to any player who sports a particularly clean-shaven head; although Brian Shaw fits the bill here, the name derives from the word “Shave” and not “Shaw”, because, duh, “Shaw” doesn’t have a “v” in it.

Thabeetan – refers to any former star college player who flops just as much as everyone thought he would; antonym of Redickulous.

Vladulant – characterized by egregious flopping, syn. “manupulative”; refers to Vlade Divac; take care not to confuse with “vlatulant”.

Are the Heat Wyatt Earp, or Annie Oakley?

There was an interesting conversation in the comments on Basketball Reference the other day.  The author, Neil Paine, used the binomial distribution to argue that it was unlikely for a true .500 team to go 5-13 in close games.  A couple of commenters pointed out that Paine may have stumbled into a “Wyatt Earp Effect” fallacy.  The effect refers to the concept that we as observers are more likely to find unlikely events remarkable after they happen, while ignoring the fact that an unlikely event was bound to happen to someone.  In the eponymous example, Wyatt Earp was seen as remarkable because he survived many, many gunfights, but someone was bound to survive all those gunfights, and it could just have easily have been Bob Smith as Wyatt Earp.  Applying that lesson to the Heat, the idea is that given that a number of close games are bound to happen every season, it shouldn’t really surprise us that one team happened to win only 5 out of 18.

I’m not sure the Earp effect works here, because the NBA hasn’t been “eliminating” teams that don’t get in close games and leaving only Miami playing at the 23rd game.  But I might be misunderstanding what is meant here.  In any event, I decided to look at the matter from another angle.  The real question, to me, isn’t just whether the Heat are winning close games, but whether they are getting in close games in the first place.  If we think of a team’s performance as varying around some mean, then a good team — one with its mean further along the positive end of the x-axis — should closely overlap with the rest of the league relatively less than worse teams.  And the Heat have been a good team, by the standard definition given by statistical types: their scoring differential per possession has been high overall (+.0646).  This is partly the result of the Heat winning blowouts against lesser teams, but you can’t discount the fact that the Heat also don’t get blown out themselves very often.  Losing to the Knicks by one hurts, of course, but it’s not as bad as losing by 30.

So here was my thinking: I would look at each of Miami’s 64 games (as of Wednesday), and run a Monte Carlo simulation using the Heat’s average scoring differential and their opponents’ average scoring differentials.  The Heat, as I said, have averaged a margin of .0646/possession, with a standard deviation from game to game of .135.  Their opponents’ average scoring differentials are easily calculable from available data.  It would take quite a bit more work to figure out the standard deviation for each Heat opponent, so I made one big assumption: that all the other teams had the same standard deviation as the Heat, .135.  This introduces a degree of imprecision into my results, but my assumption is that all teams have their efficiency bounce around to roughly the same degree.  That is, one team may have a standard deviation of .09 instead of .135, but nobody is going to have a standard deviation of .8 or something ridiculous like that.  My guess is that these results are roughly the same as you would get with a more detailed model.

I ran 1000 trials of each game, using Excel’s NORMINV function* to generate values around the mean for both the Heat and their opponent.  If the value generated for the Heat’s scoring differential/possession was greater than their opponent’s, I said they won.  If it was lower, I said they lost.

* So for each game I used Excel’s random number function RAND() to generate a random number, then used that random number in the NORMINV function, which takes three variables: a probability (given by the random number), a mean, and a standard deviation.  The function outputs a value that would produce the given probability for the given normal distribution.

I had a couple questions I wanted to look at:

1. How many close games would this Monte Carlo version of the Heat have?  The line dividing a “close” game from a “not close” game is fairly arbitrary, but I’ve heard 5 points given, so I used 5 points/94.7 possessions (the league average pace) as my benchmark.  That translates to a margin of .0528/possession.  The Heat have had sixteen such games in reality, so 16/64 = 1/4 of their games have been “close” by this definition.  They have won five of these games, for a winning percentage of 23.8%.
2. What is the winning percentage in close games for the simulated Heat?  Is it better/worse than reality?
3. How often do the simulated Heat have a winning percentage lower than their actual winning percentage, i.e. 23.8%?

Here are the results of the Monte Carlo Simulation:

• Number of close games (64 games*1000 trials = 64,000 trials): 13290, or 20.8% of all trials were “close”
• Number of close wins: 7251
• Number of close losses: 6039
• Winning percentage in close games: 7251/13290 = 54.6%
• Games with highest number of close outcomes: GM1 (BOS) 244, GM42 (CHI) 235, GM32 (LAL) 225, GM58 (CHI), 225
• Games with fewest number of close outcomes: GM27 (CLE) 162, GM29 (WAS) 162, GM48 (CLE) 164, GM18 (WAS) 170
• Number of seasons with winning percentage in close games < 23.8%: 12, or  12/1000 trials = 1.2% of the time.

This is a pretty simplistic simulation — it doesn’t take into account home court advantage for instance, and there is the problem of the standard deviation for the Heat’s opponents — but it does do a decent job of matching the number of “close” games the real Heat have had (real Heat = 25%, Monte Carlo Heat = 20.8%).  It also does a decent job of matching the Monte Carlo point differential with the actual point differential:

Note that there are only five games where the simulation predicts the Heat will have a point differential less than zero.  I couldn’t label the graph, but those games were GM9 (BOS), GM15(ORL), GM33(LAL), GM49(ORL), and GM55(BOS).  Also notice that there is a small cluster of games where the model predicts Miami will win by a small margin but in reality Miami lost.  These are the close games everyone is talking about.

Take this little simulation with a grain of salt, but I think it lends some credence to the theory that something weird is going on with Miami in close games.  It’s not, I don’t think, that Miami is just the random Wyatt Earp who was unlucky enough to not “survive” a series of close games.  It’s more like Miami is a truly great marksman (Annie Oakley?), who would be expected to hit 50% bulls-eyes, and instead is hitting 25%.  That could happen by random chance, but once the number of trials gets large enough you have to question how likely that is.

A lot of people, myself included, have had some fun with Brook Lopez’s rebounding numbers this year.  The sheer incongruity of a 7’0″ center grabbing only five or six rebounds in a game is comical.  But after reading an intriguing piece by Phil Birnbaum (“Box-score statistics are the RBIs of basketball”), I’ve had to rethink my view of Lopez’s rebounding prowess.  Birnbaum holds the position that many rebounds in the NBA are essentially “stolen”: that is, if there are several players on the same team in the vicinity of a rebound (usually a defensive rebound), it is possible for particularly motivated players to increase their individual numbers without affecting his team’s actual rebounding ability in any way.

So in the case of Brook Lopez, the bigger question is not whether Lopez, individually, has become a worse rebounder (the statistics, whether rebounds per game, OREB%, DREB% show he has), but whether the New Jersey Nets have become a worse rebounding team.  And as it turns out, they haven’t:

[numbers from www.hoopdata.com]

Those are the three years Lopez has played for the Nets.  And as you can see, the Nets rebounding numbers have actually improved this year, especially when compared to last year.  The Nets are, in fact, an above-average team in terms of rebounding ability: they rank 14th in total rebounding rate, 16th in offensive rebounding rate, and 7th in defensive rebounding rate.  So in spite of Lopez’s anemic numbers (TREB% = 10.1%, OREB%= 7.8%, DREB% = 12.5%), the Nets are doing a decent job clearing the boards.

What is going on here?  The answer comes down to the play of one man: Kris Humphries.  He is on the #1, #3, and #5 five-man units in terms of minutes for the Nets this year, and all three of those units happen to feature Brook Lopez.  And Humphries’ rebounding numbers are just nuts: 21.7% TREB% (2nd in the league), 12.4% OREB% (5th in the league) and 31.3% DREB% (2nd).  If it weren’t for the presence of another rebound vacuum, Kevin Love, Humphries would be arguably the premier rebounder in the league this year.

The story here, then, is not one of Brook Lopez suddenly becoming a much worse rebounder (although he may have regressed a little — his shooting, block and assist numbers have also gotten worse this year), but of a so-so rebounder (Lopez) playing significant minutes with a fantastic rebounder (Humphries) and the fantastic rebounder hoovering up the lion’s share of the available boards.  If you look at who played power forward along with Lopez in the past, most of the minutes went to Yi Jianlian, with a little time for Josh Boone and Ryan Anderson.  It’s not surprising that Lopez was able to grab a few more rebounds a game playing opposite the willowy Jianlian, and it shouldn’t surprise us that those marginal rebounds have disappeared with the arrival of a traditional rebounding power forward.

Something similar has happened, I think, in Minnesota, where Kevin Love is putting up numbers on the glass unseen since the days of Moses Malone.  If you think in a broad sense about what is different for the Timberwolves this year as opposed to last, two things come to mind: a) Love is playing more (36.6 minutes a game, up from 28.6 in the previous year), and b) Darko Milicic is playing a lot (24.7 minutes per game, the highest total of his career).  If you look at the five-man units for Minnesota, the top five by minutes all feature Milicic and Love.  Why should this matter?  Well, Darko is, as it turns out, a pretty mediocre rebounder: for his career, he averages an 8.6 OREB%, 18.4 DREB%, and 13.5 TREB%.  That’s better than Brook Lopez, and even above average when compared with the league averages in those categories (roughly 6%, 15%, and 10%), but I would guess if you filtered your list to include only those players 6’10″ and above, Darko would rank pretty near the bottom.

And if you look at how the Timberwolves have changed from last year to this year in terms of defensive rebounding percentage, things really haven’t changed, in spite of Love’s incredible season: the team was at 73.60% last year, and they are at 73.55% this year.  Where things have changed is offensive rebounding, where they have jumped up from 26.71 to 30.81.  This is almost surely due to Love, who is leading the league in offensive rebounding percentage and contributing a completely unique skill to the team.  But offensive rebounds make up 30% of Love’s per-game rebounding numbers.  The other 70%, which amounts to 11 rebounds per game in Love’s case, come at the defensive end.  And at that end of the court, Love’s only real “competition” for rebounds is Milicic (or the even worse rebounder Michael Beasley).  There are a lot of empty calories available on the defensive boards for Love to eat up, and to his credit, he’s grabbed almost all of them.

Does all this mean Lopez is actually a good rebounder or that Love isn’t?  No.  Rather, the point is that Brook Lopez was probably never a good rebounder, and that, although Kevin Love is an outstanding rebounder, especially on the offensive end, his numbers would likely be slightly less gaudy if he were paired with a competent rebounding front court.

The real question for Brook Lopez and New Jersey, then, isn’t how to improve his rebounding or the team’s rebounding — he is probably just not a very good rebounder, and the team’s rebounding numbers are fine, anyway, thanks to Humphries — but how to improve the other parts of his/their game.  And those problems, at least pre-Daron Williams, are many: the team ranks 20th in defensive efficiency and 27th in offensive efficiency, 26th in true shooting%, 27th in eFG%, and 9th in turnover rate (a high turnover rate is bad).  Put simply, the Nets are really bad on offense, and merely bad on defense.

For Lopez specifically, the problem that is easiest to see is offense (although his reputation on defense is not sterling and his block and steal percentages are down this year), specifically a lower shooting percentage from the field.  He is shooting 48.1% on field goals (all twos), down from 49.9% last year and 53.1% the year before that.  His shot attempts are up by 2.6 FGA/game, which suggests there may be a bit of a usage/efficiency trade-off here.  His assist percentage is down from 2.3% to 1.5%, but his passing numbers have never been great.  His free-throw shooting has remained steady at about 6 FTA/game and roughly 80% percentage.  Perhaps with the addition of Daron Williams, Lopez will have less of a burden in terms of generating offense, and his shooting efficiency will creep up.

I don’t want to make it seem like the Nets’ woes B.D. (before Daron) have been all about Lopez — I’m pretty sure you could make an equally strong case against Devin Harris or even Travis Outlaw — but just to give some context for why Lopez’s rebounding might matter less than you think.  The shift to Humphries at the four has been an unquestionable upgrade over Yi Jianlian and Josh Boone: the team’s rebounding has improved, their defense is better (down to 106.4 defensive efficiency from 108.0 last year), and Humphries even seems to be a decent offensive player (he’s shooting an eFG% of 52.4% and has a very nice assist rate of 7.1%).  His addition has been by far the Nets’ second-best move all year (the trade for a certain point guard would be #1).  He is leading the team in WS/48 and is tied for the lead in WS.  If anything, the Nets should play him more.  But to get back to Lopez, my point is that it may be unrealistic to expect him to start pulling down 12 rebounds a game — even when he was paired with Yi he didn’t average that much.  Lopez is not quite the all-around two-way big man who can hold down the fort on offense, defense and rebounding; in other words, he’s not Tim Duncan, Dwight Howard or Pau Gasol.  What he needs to offer to be effective, then, is superior offense.  So far this year, that hasn’t been there, and it’s one reason the Nets are 19-43.

Is Melo worth a dollar?

Bill Simmons has a column out reviewing the recent trades, and it’s definitely worth reading, but I had to quibble with him on one point.  Concerning the Carmelo Anthony trade, Simmons notes that whenever one team has traded three or four bit players in exchange for a star, that team has wound up being the winner (by far) of the trade.  He then goes on to cite the following 15 “four quarter for a dollar” trades:

1965: Philly trades Connie Dierking, Paul Neumann, Lee Shaffer and cash to San Francisco for Wilt Chamberlain.
1968: Lakers trade Jerry Chambers, Archie Clark and Darrell Imhoff to Philly for Wilt Chamberlain.
1970: Milwaukee trades Flynn Robinson and Charlie Paulk to Cincinnati for Oscar Robertson.
1975: Lakers trade Elmore Smith, Brian Winters, Dave Meyers and Junior Bridgeman to Milwaukee for Kareem Abdul-Jabbar.
1983: Philly trades Caldwell Jones and Cleveland’s 1983 first-rounder (No. 3, Rodney McCray) to Houston for Moses Malone.
1993: Phoenix trades Jeff Hornacek, Tim Perry and Andrew Lang to Philly for Charles Barkley.
1994: Washington trades Tom Gugliotta, 1996 first-rounder (No. 11, Todd Fuller), 1998 first (No. 13, Keon Clark) and 2000 first (No. 7, Chris Mihm) to Golden State for Chris Webber.
1996: Houston trades Robert Horry, Sam Cassell, Chucky Brown and Mark Bryant to Phoenix for Charles Barkley.
1996: Phoenix trades Michael Finley, Sam Cassell, A.C. Green and a 1998 No. 2 (No. 53, Greg Buckner) to Dallas for Jason Kidd, Tony Dumas and Loren Meyer.
1999: Phoenix trades Danny Manning, Pat Garrity, 2001 first-rounder (No. 18, Jason Collins) and 2002 first (No. 9, Amare Stoudemire) to Orlando for Penny Hardaway.
2004: Houston trades Steve Francis, Cuttino Mobley and Kelvin Cato to Orlando for Tracy McGrady, Tyronn Lue, Reece Gaines and Juwan Howard.
2005: New Jersey trades Alonzo Mourning, Eric Williams, Aaron Williams, Philly’s 2005 No. 1 pick (No. 16, Joey Graham) and a 2006 No. 1 (No. 20, Renaldo Balkman) to Toronto for Vince Carter.
2005: Miami trades Lamar Odom, Brian Grant, Caron Butler, a 2006 No. 1 (No. 26, Jordan Farmar) and a 2006 No. 2 (No. 50, Renaldas Seibutis) to the Lakers for Shaquille O’Neal.
2007: Boston trades Al Jefferson, Gerald Green, Theo Ratliff, Ryan Gomes, Sebastian Telfair, a 2009 No. 1 (No. 28, Wayne Ellington) and the rights to Minnesota’s 2009 No. 1 (No. 6, Jonny Flynn) to Minnesota for Kevin Garnett.
2008: Lakers trade Kwame Brown, Jarvaris Crittendon, Aaron McKie, the rights to Marc Gasol, a 2008 No. 1 (No. 28, Donte Greene) and a 2010 No. 1 (No. 28, Greivis Vasquez) to Memphis for Pau Gasol.

Simmons also throws a jab at the “advanced metrics” crowd (they are “getting caught up in small-picture stuff”), implying that these folks are making the same mistake of valuing a dollar (Carmelo Anthony) as much as four quarters (in this case, Danilo Gallinari, Timofey Mozgov, Wilson Chandler and a draft pick).  Here’s the thing, though: of those 15 players, 10 of them were much, much better by advanced metrics than Carmelo Anthony.  Here’s the list:

The WS/48 figures are taken from Basketball Reference and refer to season-long WS/48 for the year indicated.  Anthony has been having a bit of a down year — his average is .124, and he’s hit peaks in the .140-.150 range in the past — so bump him up a bit if you want, but any way you slice it he is worth less than a “dollar.” Apart from a few players who were either past their peak or before their peak (Penny Hardaway, Jason Kidd), Carmelo is nowhere near this group in terms of production.  And that’s what this debate has been all about: is Carmelo Anthony, in fact, a superstar player?

Some people say yes, some say no (and just to be clear, not everyone in the advanced statistics community is on one side of this debate — there are plenty of Melo defenders out there), but the general consensus is that Anthony is a good player, possibly a star player, but a definite notch below true superstar players.  He is not Wilt Chamberlain, he is not Moses Malone, and he is not Charles Barkley.  Of that list of 15, Anthony resembles Vince Carter more than Kareem Abdul-Jabbar.  Compare Carter and Anthony over their first seven years:

If anything, Vince Carter had the higher peaks, but the general pattern is similar, at least in terms of Win Shares, True Shooting percentage and effective field goal percentage.  Carmelo is the better rebounder, Vince gets more assists and fewer turnovers, but overall these guys are roughly the same.

It’s important when you’re framing this trade to figure out what your comparisons are.  When you think of trading Gallinari, Chandler, Mozgov and a pick for 2007 Kevin Garnett or 1993 Charles Barkley, you fall all over yourself saying “yes.” If you plug 2005 Vince Carter in the same trade, though, you pause for a minute and think it over.  Which is not to say that a theoretical trade of that variety would not be a good trade.  It very well might be!  Vince Carter, as much as he’s derided these days, was a pretty good player back in 2005, especially on offense.  He wasn’t going to transform you into a championship contender, but he was certainly above average.  Was that Vince Carter worth more than Gallinari, Chandler, Mozgov and a pick?  Maybe — it’s hard to know the future.  If you could have known in 2005 that Vince Carter would have two more All-Star years and then several fairly pedestrian years, you might have thought twice about the trade.  And of course we still don’t know how Gallinari, Chandler or Mozgov will turn out.  The point is, when you think of Carmelo in terms of players he is truly like, and not players that far surpass him in production, the quality of the trade for the Knicks becomes much more murky.

And that’s all the advanced metrics people mean, I think, when they question the Carmelo trade.  They want people to have the proper referents in place: Vince Carter, Bernard King, Hedo Turkoglu — yes.  Wilt Chamberlain, Kareem Abdul-Jabbar, Moses Malone — no.  Carmelo is a high-profile, stylish player, and that makes it hard to think about what his actual production has been, but when you do take that extra look, it turns out he is a good, sometimes very good, player who just happens to be a step below the league’s elite.  He is a good six bits, but not a full dollar.

Whether it is wise to trade three quarters for three quarters, then, is the question.  And it’s a tough question to evaluate, because Anthony and the three players mentioned weren’t the only component of the trade.  There just happened to be a “dollar” player who went from Denver to New York not named Anthony: Chauncey Billups, he of the .179 WS/48 career average (right in line with the “dollar” average, if you’ll recall).  So it could turn out that the Knicks did make a great “dollar for four quarters” trade, just not for who they thought.  Even at 34, Billups is still an elite player and a definite upgrade from Raymond Felton (whose WS/48 is just .098 for the year).  If it turns out the Knicks surge with the new tandem of Billups and Anthony, it’s important to keep in mind the possibility that Anthony is not the only person responsible.

Perkins doing what he does best, getting in some chump's face

If you haven’t heard, Kendrick Perkins has been traded from the Boston Celtics to the Oklahoma City Thunder.  Most observers are a bit stunned, but some have defended the trade by saying that since Orlando has declined somewhat this year, the Celtics don’t need a tough interior defender anymore.  This reasoning strikes me as dubious on its face — in general, height, tough defense, and rebounding are good things in in basketball, regardless of the opponent — but I think it’s worth looking specifically at the possible match-ups the Celtics will face in the playoffs and seeing how that reasoning holds up.  Here is the bracket if the playoffs started today:

So let’s think this through:

Round 1:

Probable opponent:New York Knicks

Defensive Center needed?

My answer would be “no.”  Amare Stoudemire is not a traditional low-post center, and in any event, Kevin Garnett will be marking him.  The only other low post presence for the Knicks is Rony Turiaf.  Perkins could be useful defending the rim and taking hard fouls on Carmelo and Stoudemire, but there is no clear match-up that makes him indispensable here.

But, then again, is any one player on the Celtics indispensable against the Knicks?  We’ll see how the Carmelo-Billups trade works out for New York, but my guess is that the Celtics would dispatch the Knicks in five games with or without any one of Ray Allen, Paul Pierce, or Rajon Rondo.  Their team is just that good.

Round 2:

Probable opponent, Chicago Bulls

The Sixers are ranked surprisingly highly in John Hollinger’s Power Rankings (#9 in the league), but the Bulls are ranked even higher (#4, which happens to be higher than the Celtics), and it’s hard to imagine the Bulls, with Joakim Noah and Carlos Boozer, getting knocked off by the smaller Sixers.

Defensive center needed?

Yes.  This is where the “we don’t need Perk” theory starts to break down.  Joakim Noah is a beast on the boards and on defense, and Perkins could be very valuable a) to block out Noah and prevent offensive rebounds and b) set solid screens so that Noah cannot harass perimeter players.  There’s also the matter of defending Carlos Boozer inside.  Garnett will probably handle that duty, but it never hurts to have a tall, strong help defender to help/take fouls.

Conference finals:

Probable opponent: Miami Heat/Orlando Magic, with an outside chance of the Atlanta Hawks.

Defensive center needed?

Quite possibly.  It’s true the Heat do not offer a traditional low-post presence, so Perkins might be more superfluous in that series (although it never hurts to have someone out there who is actually physically capable of giving LeBron James a hard foul), but here’s the thing: there is no guarantee that the Heat will make it to the Conference Finals.  Orlando, although looking lost right now, could still make it.  Even if it’s a 1/3 chance, does Boston really want to face Dwight Howard without their reliable backstop to check him?  It’s possible that Orlando is so much worse now that the Celtics are not particularly worried about beating them, Perk or no, but it’s still frightening to think of what Dwight Howard could do marked solely by Shaquille O’Neal, Jermaine O’Neal and Glen Davis.  Atlanta is another possible opponent, but the Hawks seem more like the Knicks than the Magic: a team where Perkins would help, but would not be necessary for the Celtics to win a series.

Finals

Probable opponent:

• San Antonio, 37.7%
• L.A. Lakers, 23.1%
• Dallas, 15.6%
• Oklahoma City, 9.1%

These percentages come from Hollinger’s playoff odds table.

Defensive center needed?

This is the one round I think supporters of the Perkins trade have ignored.  People are talking about beating the Heat, the Knicks or the Magic, but not the Spurs, the Lakers or the Mavericks.

In the case of the Spurs, it’s debatable whether Perkins would be absolutely necessary.  The Spurs’ best low-post player, Tim Duncan, would probably be matched by Kevin Garnett (although a big body like Perkins would be helpful to give Garnett a break), with Perkins or his replacement covering DeJuan Blair or Antonio McDyess.  The Spurs also have a penchant for going with a small lineups featuring Matt Bonner at the four and Duncan at the five, and Perkins would be no help there.  So if the Spurs keep up their strong play throughout the playoffs and the Celtics meet them in the Finals, I could see how Perkins would not be absolutely crucial to their game plan.

For the Lakers, though, the absence of Perkins would prove crucial.  Does anyone remember Game 7 of last year’s Finals?  Assuming Andrew Bynum and Pau Gasol are healthy, the Celtics would be at a significant disadvantage on the offensive and defensive boards against the Lakers.  If they matched Gasol with Garnett, that would leave the slower Shaq or shorter Davis to box out Bynum, and that could be deadly for the Celtics.

The Mavericks? It’s hard for me to picture the Mavs beating the Lakers and the Spurs, but they’re a good team, and it could happen.  The Celtics could actually regret not having Perkins to box out Tyson Chandler and Brendan Haywood (Garnett would probably be assigned to Dirk Nowitzki).  I would guess that the Celtics could still beat the Mavs without Perkins, but Chandler has been on a rebounding tear all year, and, similar to Noah, he is a pest defending the pick and roll.  Without Perkins to set bone-crushing screens, Chandler could have a much bigger impact on the shooting of Pierce and Allen coming off of screens.

The really intriguing case is the dark horse team, that up-and-coming squad from Oklahoma that just signed a rangy defensive center.  It would be a crazy Finals if the Thunder faced against the Celtics, but it could happen.  In that series, the Celtics would, I think, match up well against the Thunder: Rondo could take Westbrook, Pierce could take Durant, Allen could take James Harden/Thabo Sefolosha, and Garnett could take Serge Ibaka/Nick Collison, and do quite a bit of roaming on defense.  But the Celtics would again lack a strong interior presence to keep a tough opposing rebounder off the boards; Boston would find itself in the odd cosmic quandary of wishing they had Kendrick Perkins back so that he could match up with … Kendrick Perkins.

Conclusion

I can sort of see the pro-Perkins point of view.  There are several teams out there that don’t rely on traditional low-post centers anymore, and Perkins’ big weakness — poor foot-speed — could be exploited by teams that stretch the floor and use non-traditional/non-offensive centers (Miami, New York, San Antonio, Dallas).

But even in those best case scenarios, I think the Celtics will miss Perkins.  He just did so many things for the team beyond covering Dwight Howard:

1. He set punishing screens that constantly freed Allen, Pierce and Garnett for open shots.  Perkins is not heavier than Shaquille O’Neal, but he always seemed somehow to take up more space and set more disruptive screens than the old man.  Perkins is just all elbows and muscle and seeks contact in a way Shaq rarely does.  A critic might say Perkins sets illegal screens, but if he does, he gets away with it, and sets the best illegal screens in the business.  That’s a useful skill against any team.
2. He punished people, period.  On offense, on defense, in the lane.  Perkins was the guy who took hard fouls and made teams wary of going in the paint, the guy Kobe Bryant, Dirk Nowitzki and Derrick Rose had to run through if they wanted to make a cut to get open on the baseline.  It’s hard to describe, but Perkins gave Boston a quality of … “density” is the closest word I can think of: their defense seemed like a thicket of elbows, hands and knees, and Perkins was the biggest, thorniest bramble in there.  He played the same asshole-with-elbows (to coin a phrase) role that Bill Laimbeer and James Edwards played for the Bad Boy Pistons, that Bill Cartwright played for the Bulls, or that Charles Oakley played for the Knicks.  That psychological edge is hard to measure, but I think it’s real, and could be very relevant against teams like the Heat or the Knicks which feature players that rely on forays to the rim.  If Dwyane Wade knows that he can fly to rim and never have to worry about a giant meat-hook taking him to the floor, the Celtics could be in for a long series.
3. He rebounded well.  I don’t care who the opponent is, you need to rebound the ball.  Perkins OREB%/DREB% numbers this year are 9.6%/27.7%; Shaquille O’Neals are 8.4%/20.3%.  Perkins presence will be missed on the boards.

The Celtics will miss you, Perk.  It’s the end of an era in Boston, an era of truly frightening, physical defense.

Here is the Heat offensive efficiency by game as the year has gone on:

And here is the Heat offensive efficiency minus the season average defensive efficiency of their opponent on that night:

There’s a very, very small rise in offensive efficiency and an even smaller rise in offensive efficiency minus opponent defensive efficiency, but I don’t see a clear trend.  The trendlines bear that interpretation out, and a simple regression of the efficiency measures by time shows no statistically significant trend component in the Heat offensive efficiency.  This is a pretty simplistic view of things: it doesn’t account for home court advantage, for instance, and the Heat six more road than home games (31 road, 25 home), so that could change things; also, I’m calculating possessions with the quick and dirty equation POSS = FGA + TOV – OR +.44*FGA (averaged for both teams), so that might be a little off.  But my quick take is that the Heat have not had some “eureka!” moment where their offense has suddenly gotten much better; instead they have just been more or less pretty good over the course of the year.  Maybe when they get Udonis Haslem back and they can finally put together line-ups of Wade, James, Bosh, Haslem and Mike Miller out there, we’ll see an up-tick in offensive efficiency, but I wouldn’t call that “learning” so much as “becoming more talented.”  And my suspicion is that a line-up of those five players would not be vastly better than current Heat line-ups anyway.  The main problem I see the Heat having on offense is defenses packing it in the lane to stop drives, and Haslem does little to alleviate that problem.

Anyway, it’s an interesting question.  Here is the data for your perusal:

The show's cast of colorful locals

The Channel Television-produced show will take the basic setup of the popular MTV show — a reality format following a group of friends from a unique cultural subgroup — but transplant it from the boardwalks of New Jersey to the cow pastures and fishing boats of the island of Jersey in the English Channel.

The location of the scandalous new show

The show will give viewers an insiders look at the style, lingo and habits of the island’s Norman-descended Jèrriais inhabitants (who call themselves “Georges” and “Georgettes” due to the popularity of those two names on the island) and their devotion to the GTL lifestyle: Groceries, tea, and les mervelles (a traditional fried dough snack).  The first episode of the show will feature Michael “the What’s What” St. Helier and Snew-kie the Jersey cow hooking up after drinking too much apple brandy on La fête Nouormande, with The What’s What shamefacedly trying to hide it from his friend Paulie De Gruchy the next morning.

[All information about the island of Jersey taken from the Wikipedia entry on the subject.  I apologize for any inaccuracies about the island's culture.  Apparently Jersey has made the transition from agricultural community -- i.e. Jersey cows -- to a sort of British Channel Cayman Islands by specializing in offshore financial holdings, so maybe these characters should be suit-wearing sharps rather than farmers.]