Today’s post is going to make a simple point.
Stuart Gray was not a stiff!!!
Why am I commenting on a player whose career ended in 1990-91 and consisted of only 3,457 minutes played (and five starts)?
For an answer, we turn to JC Bradbury (yes, The Baseball Economist).
Bradbury recently wrote a post on Stuart Gray (a comment motivated by the fact Bradbury grew up in Charlotte, where Gray spent one season). Within this column is the following statement:
“(Dick) Harter felt that the Hornets could be competitive if they could get a good big man, and that big man was Stuart Gray, a man Harter felt was being underutilized by his former team, the Indiana Pacers.
The Hornets eventually did acquire Gray, I believe for a second-round pick, for the 1989-1990 season. Gray came to town and was nothing less than the pure embodiment of a “stiff.”"
Gray Responds
Two weeks after this column was posted, JC Bradbury received an unexpected comment from none other than Stuart Gray. And Gray was not a happy camper. And his response gives a remarkable level of insight into a neglected part of player transactions:
This post is in response to your post about Stuart Gray. Have the guts to post it or don’t. I don’t care. At least I have the satisfaction of responding to this nonsense. I wish my friend never showed me your wesite.
First of all, get the facts straight before you publish “crap” as fact. One of the things I do not miss about sports is self serving individuals talking out of their rear end and thinking it is factual!
You have no idea what transpired at the Hornets in 1989. You have no idea of the petty team politics and you have no idea why I was brought to the team. Loyalty was not the reason, it was to fill a specific deficiency in the roster. I am pretty certain that Dick Harter knew that he could not “sell” me starting in front of the ACC prodigy, J.R. Reid. The team was soft and they could not win unless they established that they would stick up for themselves.
My career stats never suggested Dick Harter was looking for a franchise player as you would suggest. You are looking to paint me as a “stiff” to try and make a point but that would only be accurate if Dick brought me in to be more than a role player. Nice writer’s trick but less than honest representation of the reason Dick traded for me.
He was looking for a center that would stand up to the other teams’ players that were beating the living crap out of his players on a nightly basis. Funny how the refs started calling fouls on opposing players once I went “berserk” as you referred to it. You think it was so easy? Then you do it!
Finally, Dick didn’t lose his job due to bringing me on board. He lost it due to political infighting and back stabbing that defined the team and organization in 1989. My job was to protect players like Rex Chapman and Dell Curry from getting beaten up each game and literally being hurt by the rough play of the other teams. That is why I came to Charlotte.
Please do not change the facts to make a point on your website. Your writings about Dick Harter and me show a profound ignorance to the finer points of basketball strategy. Dick is “Old School” and coached by a certain code that not all players understood or were willing to follow. Many chose not to.
By the way, I was traded to the Knicks to back up Patrick and protect his back. Different NBA back then but you wouldn’t know that would you? One final note, my career was defined by my rebounding and defensive skills against other centers. I also played for so long because I could get shooters open by setting picks that defensive players could not get through. Just a few more of the finer points of being a “Role Player” that you probably do not understand.
To the other posters, sorry about the rant but I felt JC’s comments warranted a response! Also, Dick Harter is an honorable man and deserves better than to have this inaccurate version about what “happened” at Charlotte in 1989 being told. This post is disrespectful to me and to the real reason for Dick losing this job.
Dick did the honorable thing and left quietly without making the bad situation the organization had created worse. I have no desire to discuss or does anyone need to know what happened out of respect for Dick Harter and to the sheer fact of “who cares.”
Enough time has past that this issue should have been relegated to the trash heap. Since it was given a new life, then get it right! You may run a small website but others read it so professionalism dictates that you don’t pretend to have knowledge of something you probably know nothing or little about.
One final word, loyalty to former players is not a bad thing, especially if you know they already fit into your system and can fill certain roles. Doesn’t this happen in business all the time? The problems arise when the player cannot fulfill the role due to team politics.
Many players are traded into a situation without a consensus in the upper management team. Politics takes over and the player is “caught” in the middle of a power struggle that they neither understand or want to be involved in. Players go where they are sent for the most part unless they are the fortunate few that have trade approvals in thier contracts. For the rest, you hope that everyone wanted you when the trade was finalized. If not, good luck, your career could be over!
And Bradbury responds:
Gray’s comment led to a response from Bradbury, a response that highlights the nature of criticism in public forums.
Dear Stuart,
Thank you for your response. While, I would like to respond “I meant no disrespect” I did call you a stiff. I was 16 at the time that you played for the Hornets, so my memory may be a bit hazy. Although, I do have a vivid memory of the Michael Cooper fight, and I believe I was actually at the game to witness the most bizarre sports fight that I have ever seen (though it could have been televised).
Like all sports journalists and bloggers, I write about the quality of players’ play. I don’t think it’s incorrect to say that you were not a very good NBA player. There is not shame in this in that you were clearly one of the best basketball players in the world. It’s a laudable achievement as I was once schooled in a pick-up game by the 12th man my college’s Division II basketball team. Compared to me, and most basketball players in the world, you are very good. The term “stiff” may be a bit harsh; however, I doubt I am the first person to use that term to describe you. I can only imagine what you must have said to Tom Sorenson or Rick Bonnell of The Charlotte Observer. Still, that doesn’t defend my use of the term, but I will defend it.
Writing is difficult. Unless it’s a diary, you must get the point across but keep it interesting. I could use objective terms without positive or negative connotations like “three standard deviations below/above average” to describe play, but that would get tedious for readers. Stiff is short-hand for a tall basketball player who commits fouls and doesn’t score much. That may have been your job, and you did it well. If that makes you a stiff, so be it. That I might not understand the value in what you did is my problem, and you probably shouldn’t feel disrespect, but instead just brush me off as an idiot. I do it all the time when people with little statistical training “correct” my “mistakes”. I’m also sometimes called a “stat-nerd” who should get out of the basement and see a game. They don’t know what they’re talking about and have no influence over my life, so I ignore them. These people don’t know anything about who I am as a person, and I don’t take those to be criticisms that influence how I view myself.
So what I ask of you is not to take any disrespect from my calling you a stiff or describing your attempt to decapitate Michael Cooper as berserk. That’s how I view your play on the court. You have every right to dismiss me as an uninformed idiot. If I’m going to write about sports, I’m going to have to be critical of those involved in the sport unless I want to run a rah-rah rag. As a person, I care about those things that matter most to players. I’m sad for players when things go poorly (illness, death, getting cut, etc.) and happy when things go well (marriage, birth, award-winning, etc.). If I worry about hurting feelings, I can’t do my job. Just as you probably didn’t enjoy physically injuring players with hard fouls to protect Patrick Ewing, I get no joy from the fact that I might hurt Jeff Francoeur’s feelings.
I would like to add, that I agree with your assessment of Harter. He was treated badly by Hornets owner George Shinn. He was fired on a road trip while his brother was having serious health issues after the team said it would not fire him. Gene Littles didn’t fare so well either. The politics behind the team sound ugly, and I’m sure you know more about that than I do. Still, the positive spin he put on your acquisition is worthy of criticism, and it served as an example that sometimes executives are a bit too smitten with past players.
Thanks again for writing, Stuart. I wish you well.
And Stuart responds again
Within two hours of Bradbury’s response, Stuart indicated there were no hard feelings:
Stuart just followed up with a response that I very much appreciate.
You are truly an interesting sports writer, fan and now I am going to say gentleman. I also enjoy someone that sticks by their principles in the sports world. Don’t lose that quality as it is rare!
Your assessment is correct about how trades or draft selections are spun in the media. I knew very quickly that there would be no way for me to get the minutes necessary to make an impression that was “expected.” As you remember, the town was alive with Hornet’s Fever those first few years and expectations were just a “little” out of whack with the reality of the team’s abilities.
While I rarely back down from a challenge, I knew this one was going to be almost impossible to win. I should have taken Hurricane Hugo trashing the city the day after I arrived as a REALLY bad omen!
Thanks for your comments. I will visit the site from time to time. Internesting story brewing with Greg Oden. Similar story potentially (except for the 100+ million that I never made).
Stuart Gray’s Career
There are a few interesting points about this exchange. Perhaps the least important is that I think Bradbury’s initial assessment of Gray- made when Bradbury was a teenager — is incorrect. Gray – especially the year he played in Charlotte – was not a horrible player. Here are some of his career marks:
- Career Record: 0.022 WP48, 1.6 Wins Produced
- 1989-90 (with Charlotte): 0.111 WP48, 1.1 Wins Produced
- Year 4 to Year 6 (1987-88 to 1989-90): 0.085 WP48, 3.6 Wins Produced
An average player will post a WP48 [Wins Produced per 48 minutes] of 0.100. So Gray was only above average the one year he played for Bradbury’s Hornets.
When we look at the career numbers we see that Gray was well below average his first three seasons. He then posted respectable numbers from season four to six (with Indiana and Charlotte). He then did very little in the 131 minutes he played in New York to finish his career.
When we look at the individual stats we see where Gray was good.
Table One: The Career Number of Stuart Gray
As Gray indicated, he was not a scorer. His strength was rebounding. As is often noted, the non-scoring aspects of the game are often undervalued. So it’s not surprising that many people thought Gray – the quintessential non-scorer – was not well regarded.
Politics and Criticism
Beyond the issue of the quality of Gray’s contribution, Gray’s initial response highlights a key aspect of how teams are built in the NBA. People outside of the NBA often act like general manager and coaches in the Association are simply well-paid fantasy league owners. My understanding of the process – which Gray’s comments confirm – is that the personalities of the participants play a significant role in how a team is constructed. Inter-firm politics are certainly a part of the NBA (as they are everywhere) and the politics often dictate the decisions teams make.
Let me close this discussion by moving past the lessons about the NBA to the lesson learned about criticism. Bradbury’s explanation of his initial critique was excellent. And it was wonderful to see Gray’s response to this explanation. Although the exchange between Bradbury and Gray did not begin well, it ended with both parties behaving like adults. One has to note that often this doesn’t happen on the Internet. Often people cannot agreeably disagree. Instead, often things escalate into personal attacks. Again, it is good to see that not happen in this instance.
- DJ
The WoW Journal Comments Policy
Our research on the NBA was summarized HERE.
The Technical Notes at wagesofwins.com provides substantially more information on the published research behind Wins Produced and Win Score
Wins Produced, Win Score, and PAWSmin are also discussed in the following posts:
Simple Models of Player Performance
What Wins Produced Says and What It Does Not Say
Introducing PAWSmin — and a Defense of Box Score Statistics
Finally, A Guide to Evaluating Models contains useful hints on how to interpret and evaluate statistical models.
11 responses so far ↓
Tommy_Grand // November 20, 2008 at 11:25 pm
you are so full of baloney!!!
Lior // November 21, 2008 at 1:03 am
That NBA players can roughly be divided into “starters” and “role players”. It may be worth it to calculate the average WP48 for these two classes separately. After all, WP48 of 0.100 being “average” is an average weighted by minutes played (not a simple average over all players) and thus heavily influenced by the better players (usually the starters) who play most of the minutes.
In other words, I expect a WP48 of 0.085 to actually be above average for a backup center, and would not be surprised to learn that a WP48 of 0.115 is below average for a starting center.
Lior // November 21, 2008 at 1:06 am
An additional point: I don’t think WP48 is such an accurate measure of playing strength that a difference of 0.015 is scientifically significant. In other words, Gray’s performance in years 4-6 should be thought of as being “about average” in any case.
dusitnc // November 21, 2008 at 2:08 am
Lior, good role players can have good WP48. Joel Pryzbilla, Josh Childress, Ronnie Brewer, Paul Milsap…
Another Pete // November 21, 2008 at 7:37 am
Along the lines of Lior’s comments: you often note that the average WP48 is 0.100; what’s the median WP48? Unlike the average, I’d guess the median changes from season to season, but a rough number from even one season would still be interesting.
K'liph7 // November 21, 2008 at 9:10 am
any comment on the politics of Billups trade?
Lior // November 21, 2008 at 12:22 pm
dustinc: There are certainly reserves who (in their minutes) are very productive — I am definitely not saying that Gray was a great player. However, strong reserves are rare — I am saying that Gray was likely not much less productive than the typical employee in his position.
Pete: WP48 averages to 0.100 by definition (that is, this is part of the calibration of the stat).
Regarding my second point: WP48 of 0.085 is 15% below average, almost surely a meaningful number in terms of WP48 itself. But WP48 is merely a proxy for the actual strength of the player. With the error bars in the total WP of the team compared to its actual wins (where the WP/WP48 model is stongest) being what they are, I don’t think a 0.015 discrepancy in WP48 numbers can be sufficient to meaningfully distinguish players.
Jason E // November 21, 2008 at 2:11 pm
0.100 is the average per-minute production of all players. It has this value by definition. It is the weighted mean for all players, weighted by minutes played. WP48 does not appear to be normally distributed. There are more players with low wp48 than high wp48, something that should not be surprising, since coaches tend to play better players for more minutes. While many (most? all?) coaches are not aware of a player’s wp48 and do not assign minutes in straight accordance with its recommendation, better players by this metrics are *usually* ones that are better in a coach’s opinion as well. This means a majority of the players have below average wp48, but these guys get far fewer minutes.
In a random matchup of any two players, if you have a 0.100 wp48, you’re likely to be better than your opposing number. However, this is not true in any given matchup on the court, as there’s selection against the random matchup in who gets on the court.
Lior, you are absolutely correct that there must be some standard error of WP48. What sort of significance you want (e.g. p = 0.05) is itself somewhat arbitrary. The low p cutoffs seen in most scientific studies are perhaps not appropriate for sports where the end result is a binary outcome: win or lose, and we’re really looking here at a system that assesses the relative probabilities of these two outcomes. My *guess* (read: I’m not computing this here) is that a 95% confidence interval on wins and losses probably encompasses a wide, wide range, but just because the errors overlap doesn’t mean the values are the same. It just means if we want to be certain that we’re very rarely wrong, we cannot make a call that there is a difference in such circumstances.
I am not exactly sure how one would go about computing this error in any event. Is it merely a function on the error estimates on the rate activities (e.g. rebounding percentage, fg%) translated into raw numbers to calculate Wp48? That seems as good a stab as any.
I’ve been thinking more and more about wp48 in terms of likelihood ratios. The definitional average of 0.5 assumes that both teams have equal prior probabilites of winning before player performance is evaluated. If this is the case, then players producing above this value raise your probability of victory, players below it lower your probability. If the prior probability is lower, then sub-average players may actually help their team *relatively speaking* even if they do not make it such that their team is still more likely to win the game. They can just make it *more* likely.
Since we are living in a binary world in sports (w or l) it might be that a narrow confidence interval (p = .5) can still be useful for making predictions. If you’re betting consistently with 51% odds of winning, you’ll come out ahead in the long run. This is the formula casinos have followed for years: many games with odds slightly in the house’s favor = a constant stream of people losing their money albeit irregularly and slowly enough with enough individual variation to keep the suckers coming back.
Vince Gagliano // November 21, 2008 at 2:58 pm
By the way, there has just been news that Zach Randolph has been traded to the Clippers and Jamal Crawford to the Warriors.
Here is the lowdown for each deal from a Win Score perspective:
Crawford has posted a est. career average of .058 WP48 as a shooting guard, in exchange, they’re getting Harrington, who posted -.005 WP48 as a PF.
Randolph has an estimated .111 WP48 as a power forward, while Tim Thomas, depending on your point of view, is either a slightly below average small forward (.077 WP48) or a really unproductive power forward (-.025)
On a side note, Cuttino Mobley (.085 at SG) is being traded for Mardy Collins (-.033 at SG) as part of the Randolph deal.
In the long run, the Knicks are conceding the ‘09 title race to try and make a run at the LeBron Sweepstakes in the future.
Lior // November 21, 2008 at 9:21 pm
Jason: I’ve been meaning to post the following here for a while.
As WoW makes clear, statistics are mostly kept in an attempt to evaluate player productivity. It is not obvious to people who haven’t thought about this, but there are three distinct kinds of errors associated with stats like WP, WP48, or PER (Adj. PM is a bit different)
0. Measurement error
1. Statistical error
2. What I call “scientific” error but social scientists surely has a better term for.
Let’s say a player has an “intrinsic” rebounding rate of 10 rebounds per 48 minutes. What this means is that, on average, the player will get 10 rebounds every 48 minutes of playing time. We’d like to know the player’s rebounding rate. So we total the number of rebounds the player caught, and divide by his playing time.
But note that the number of rebounds the player actually gets is best thought of as a random variable fluctuating around 10 rebs/48 min, not as a deterministic value. The player could have had unusually good games or bad games, have chanced to against some teams on days they shot badly or particularly well (so there are more or less rebounding opportunities) etc. In other words, the stat we defined in only a statistical measure of the true rebounding rate of the player. So far there are only two errors in our measurement: “measurement error” comes from the fact that the recorder might not have recorded the correct number of rebounds, or the correct number of minutes. I think we can ignore that for the NBA. “Statistical error” is the main effect I’ve just talked about. It means that, if the player played for 576 minutes and caught 120 rebounds, his rebounding rate is probably close to 10rebs/48mins but we can’t know it exactly.
In the same way we can distinguish between the player’s “intrinsic WP48″ and the actual value we measure at the end of the season based on actual performance. The resulting is the “statistical error”, and I suspect this is what Jason was trying to quantify with his “confidence interval”.
One way estimate the noise in WP48 (or WS48 or PER etc) is to split the games in the season randomly into two groups and measure the stat separately in each. The difference you get is probably due to statistical error.
Now the other kind of error, the one I really care about, is the distinction between WS48 and the abstract concept “actual playing strength”, or if you want “player productivity”. In other words, WP48 is a measure of productivity, but it is not exactly the productivity of the player (for example, the total WP for the team is not exactly its total number of wins).
The latter error is the one I was actually thinking about. Prof. Berri might have actual data on this, but I would be hesitant to say that a difference of 15% in WP48 numbers actually says one player is better than another. The reason is not statistical error (especially if we are talking about many thousands of minutes played) but the limitations of the model.
Italian Stallion // November 22, 2008 at 1:03 pm
Someone is going to HAVE TO analyze the recent Knicks moves.