SAY YOU'RE AN NBA GENERAL MANAGER and, much to your surprise, the NBA actually decided they wanted to play in 1998-99. All of a sudden, you'd have to get off that phat leather chair and your not-so-skinny wallet and make a deal. Sprewell for Mashburn sounds good. Marbury for Ward and Houston and a few draft picks. Find someone who is not going to demand $40 million per year, someone who isn't going to choke the coach, someone who doesn't have friends who surreptitiously stash drugs in his luggage, and someone who doesn't wear a lace teddy under his uniform.
It's tough these days, especially when you consider that you also would like to get a player who actually helps you win.
That's where this article comes in. It may not be as important as a babysitter, but properly evaluating a player's effect on your win-loss record provides a sense of security that even a salary cap can't provide. (But that's your own fault.)
...How To Evaluate A Player's Effect on Your Record?...
|
How many wins and losses does a player add to your team? |
So how can it be done? What does it even mean for a player to affect your win-loss record?
Doing this kind of research means answering tough questions, thinking about the details. For instance, the easy answer to what it means for a player to affect a team's record is just, "If I put him on my team, will it get better or worse?"
Bear with me as I sort through the ramifications of this. Let's say you're Phoenix and you acquire Kevin Garnett and, because you have so much sunshine, because you don't have to pay Rex Chapman more than dirt to stay, and because your players can go golfing every day, you don't actually have to pay Garnett $20 million a year AND you can keep all your existing players. (Hey, if the players and their owners can live in a multimillion dollar fantasy land, I can use my own fantasy example.)
All you're doing is adding Kevin Garnett to your stacked lineup. Is the team going to get better? Probably. He can step in for the constantly injured Danny Manning and give you a few wins. He'll probably also take some time from Hot Rod Williams, Antonio McDyess, and Cliff Robinson. The amount of time Garnett gets from each of these slots depends on his relative ability compared to the players he's replacing, and on the coaching decisions of Danny Ainge, if not 23 other things as well. Let's just say that his shoe contract has no effect on it -- I already told you, it's fantasy.
Now we would have to estimate the number of minutes Garnett gets and reduce the minutes of four other guys. That estimate would be based on some estimate of quality, offensive and defensive ratings, for example. We then assume that everyone's performance level for the coming year is going to be roughly the same as the past few years. We do some addition, some division, then a bit of multiplication, and we realize that Garnett only adds 2-3 wins to the Suns. (Only a partial fantasy.)
Now let's say reality kicks in and you find yourself as general manager of, gasp, the Clippers. Instead of sun, cheap off-guards that produce, and perennial golfing in Paradise Valley, you have brown haze, Lamond Murray, and mini-malls in Orange County. All of a sudden, you want to pay $20 million and you can dump a bunch of players. The reason is because Garnett now replaces Rodney Rogers, Ike Austin, Eric Piatkowski, and Lorenzen Wright. Four goons and a superstar. Guess who gets a lot of time. The Clippers get better by much more than just 2-3 wins. Garnett is worth a lot more to the Clippers than he was to the Suns.
...The Point Is...
The point of this little satire is that a player's value to a team depends on the context. Who is being replaced and who is getting more time? Maybe we can eventually predict what a coach is going to decide -- it's not unfathomable -- but it's hard.
What I conceived of long ago was that I could calculate the wins per minute that every player is worth, then multiply these values by the estimated number of minutes they'll play.
I was 20 years old at the time and clearly naive. I actually thought I could do it.
It turns out I can get close. I can actually assess wins and losses fairly well. I can determine wins and losses per minute, too. But what I find is that translating a players record from team to team is difficult and that there may be a better way to assess a player's performance on a new team.
Over the next couple weeks, as the NBA owners and players take some time to catch up on the Monica Lewinsky affair, I'll get down to business and present some of the logic behind individual win-loss records. I have a couple ways to calculate them, including a new way that looks at every game to see if players outplayed their opponents.
In the meantime, there is some background reading that you could do.
What none of these do is adequately discuss the theory behind the method, when it must be modified, or how to improve it. Nor have I established a decent library of numbers to compare. These are some of the goals for the next few weeks.