[Storyteller]

The purpose of this thread is to compare the +/- value of Damon Stoudamire vs. that of Jeff McInnis. I hope to update it after each game (at least for a week or two), starting with the 1/15/04 game against Phoenix. Each date will show the difference between points scored by the team vs. points allowed by the team: 1)when the player is on the court; 2)when they are off the court; and 3) net difference between the two values.


1/15/04 - Phoenix
Stoudamire on the court: -20 [31:38]
Stoudamire off the court: +11 [16:22]
Stoudamire net value to the team: -31

McInnis on the court: +6 [23:19]
McInnis off the court: -17 [24:41]
McInnis net value to the team: +23


1/17/04 - Dallas
Stoudamire on the court: -15 [26:56]
Stoudamire off the court: +11 [21:04]
Stoudamire net value to the team: -26

McInnis on the court: +8 [18:31]
McInnis off the court: -6 [29:29]
McInnis net value to the team: +14

 

[Storyteller]

OK, after correcting a few numbers from the first couple of games and adding the last two games, here's the update:


1/15/04 - Phoenix
Stoudamire on the court: -22 [31:38]
Stoudamire off the court: +13 [16:22]
Stoudamire net value to the team: -35

McInnis on the court: +10 [24:09]
McInnis off the court: -19 [23:51]
McInnis net value to the team: +29


1/17/04 - Dallas
Stoudamire on the court: -15 [26:56]
Stoudamire off the court: +11 [21:04]
Stoudamire net value to the team: -26

McInnis on the court: +8 [18:31]
McInnis off the court: -12 [29:29]
McInnis net value to the team: +20


1/18/04 - at Phoenix
Stoudamire on the court: -4 [37:32]
Stoudamire off the court: +0 [10:28]
Stoudamire net value to the team: -4

McInnis did not play


1/20/04 - at Sacramento
Stoudamire on the court: +2 [28:05]
Stoudamire off the court: +3 [24:55]
Stoudamire net value to the team: -1

McInnis on the court: +0 [25:40]
McInnis off the court: +5 [27:20]
McInnis net value to the team: -5


Tonight's game was the first that Damon scored better on the +/- scale than McInnis.

I'll keep this thread going until we get a full week of games, post an "average score" for the week, then start a new thread for the next week.

After a couple of weeks, we'll be able to see some legitimate patterns (I hope).

 

[Bwatcher]

If I am understanding your system, I think it is a bit unusual. Generally +/- refers to the effect a particlular player has on the team score while that player is playing. To assign a value to a player when the player is not on the court is a bit questionable.

To take an example: Jim plays 10 minutes and the team is +12 during that time. He doesn't play the other 38 minutes and the team is -20. So, you are assigning Jim a net of -8, when most other +/- systems would assign Jim a +12. To put it another way, you are assigning responsibility to Jim for something he had no control over. He wasn't involved in team play at all for that 38 minutes, so why should any events then, be assigned as his responsibility? To follow on, in your system, every DNP player would receive a -8, even though they never played in the game.

 

[dkap]

I think the discrepancy is the point of view. You're looking at it from the player's perspective. SoCal is looking at it from the team perspective. In the latter, you look at two cases:

1) Impact when player is in.
2) Impact when player is out.

Combined, that gives you a way of looking at the net impact of that player to the team. I agree that the composite of the numbers makes for a questionable statistic, however.

Dan

 

[Bwatcher]

I looked at 82 games and found that they are using this same system.

I think I understand the difference you are trying to draw about effect from the point of view of the team, but from the point of view of my training and use of statistics, it doesn't really hold water. The better way is just to compare the numbers of the players for their effects while on the court. The "not on the court" numbers don't hold the same level of meaningfulness. To combine the two reduces the quality/value of the 'on court' numbers.

 

[Minstrel]

Originally posted by <b>dkap</b>!

Combined, that gives you a way of looking at the net impact of that player to the team. I agree that the composite of the numbers makes for a questionable statistic, however.

Yeah...it seems like a double penalty. When a player "earns" a +/- while on the court, his +/- off the court is simply going to be: (final margin between the two teams) - (player's +/- on the court).

Where the "final margin" could be positive or negative, depending on whether the team you're looking at won or lost.

Therefore, the +/- on the court is conceptually identical to +/- off the court and informationally identical. One is simply the inverse of the other. It's not like each is calculated with a different method.

Thefore, adding them together is simply giving double weight to a single piece of information, making the player look twice as good or twice as bad as they actually were.

I think, anyway. Conceptually, a number minus its inverse is essentially twice the starting number.

 

[Storyteller]

Originally posted by <b>Bwatcher</b>!
If I am understanding your system, I think it is a bit unusual. Generally +/- refers to the effect a particlular player has on the team score while that player is playing. To assign a value to a player when the player is not on the court is a bit questionable.

Over the course of one game, I'd agree with you. But if a pattern develops over the course of a season, I'd say it's signficant. That is to say, if the team scores significantly better or worse when a certain player is on the bench. That's a significant pattern.

To take an example: Jim plays 10 minutes and the team is +12 during that time. He doesn't play the other 38 minutes and the team is -20. So, you are assigning Jim a net of -8, when most other +/- systems would assign Jim a +12.

Actually, with that scenario, Jim would have a net value of +32.

To put it another way, you are assigning responsibility to Jim for something he had no control over. He wasn't involved in team play at all for that 38 minutes, so why should any events then, be assigned as his responsibility?

You're right. Jim has no direct control over what happens when he's off the court. That's why these numbers are just that - numbers. Their power is not so much in their value but in their use of comparison over an extended period of time. If, as I said before, a significant pattern develops in relationship to other players, I think it says something.

If the team consistently outscores its opponents when Jim is on the court (and he plays significant minutes over the course of an extended period of time) while being outscored while Jim is on the bench, that says something. If the reverse is true, that also says something. Especially if the values are not in line with the rest of the team.

If Jim's net +/- is -10 over the course of the season, while every other player's is between -2 and +2, wouldn't that communicate something significant about how Jim affects the team? It's not a definitive statement, but I think it would have to be taken into consideration.

To follow on, in your system, every DNP player would receive a -8, even though they never played in the game.

Every DNP gets a 0.

 

[Minstrel]

Originally posted by <b>So Cal Blazer Fan</b>!

Every DNP gets a 0.

Why? Isn't a DNP just off the court time? If the team wins by 7, then shouldn't a player with a DNP get a +7 off the court +/- and if the team loses by 7, shouldn't the player get a -7 off the court +/-?

What's the significant difference between recording "off court +/-" in a game a player plays in and in a game a player doesn't play in? If seeing how the team performs without the player is valid for 30 minutes in a game, what makes it invalid for 48 minutes?

Just wondering.

 

[Storyteller]

Had to edit this - it was rambling due to fatigue.

I'll post tomorrow when I'm more coherent.

 

[Bwatcher]

When there is data, there is usually a pattern. That's not the issue, really. The issue is "is the pattern meaningful".

I'm not saying there is "no" meaningfulness in the "off court" statistic, but at the moment, my intuition is telling me that I don't think it has as much meaning as the "on court" statistic.

A way (not necessarily the best) to examine the issue is to think about the "error bands". The on court numbers are probably pretty representative for the starters who play long minutes. They become less representative (wider error bands) for players who play just a few minutes. Thus, comparing starters is likely to give good results, but comparing players who play few minutes with any others is less likely to accurately represent differences. With the off court numbers, the opposite result is obtained. The bench players have representative numbers for "not playing" while the high minute starters have numbers with bigger error bands.

When the two numbers are combined for a starter we get a high quality (small error band) combined with a lower quality (bigger error band) number. Using the "off court" number, increases the total average error.

For the low minute bench players, using the "off court" number would decrease the total average error, since the off court number most closely represents their per minute contribution to the team.

So, from the standpoint of error bands, this system helps us get a more precise sense of how low minute players affect the team, but it gives us a bit less precise sense of how starters affect the team, than the on court alone statistics would.

 

[Blazer Ringbearer]

I don't think it is a perfect comparison for all players, but for this particular situation I think it is fairly good so long as Damon and McInnis play similar minutes...

It serves as a nice rough look at the situation. The nice thing is that the way SoCal has been presenting it, he gives you all of the data broken down as well as summed so that you can look at it however you like.

Thanks for taking the time to do this, SoCal. I'm excited to see how this story develops...

Cheers

 

[Minstrel]

Originally posted by <b>Blazer Ringbearer</b>!

Thanks for taking the time to do this, SoCal. I'm excited to see how this story develops...

Yeah, I'm thankful to SoCal for doing the work. It's definitely interesting...I just like to discuss stuff like statistical modeling. Baseball habit.

 

[blazerfan4life]

Well now that JEFF is gone ..this post that SCBF worked hard on is no longer needed..unless it is a DAMON vs Q post....:laugh:

 

[gambitnut]

Originally posted by <b>blazerfan4life</b>!
Well now that JEFF is gone ..this post that SCBF worked hard on is no longer needed..unless it is a DAMON vs Q post....:laugh:

If Woods starts getting enough PG minutes to make them comparable, that could be interesting.