Commodore, with this applet, given your handle, I am very impressed. How difficult would it be to use this applet for other NBA teams? By the way, DoD, halflingenius is Morey's preferred moniker.
Would there be a way to use modifiers? For example minutes played as a percentage somehow? Something that would show more of an overall impact than just +/- ? This is really cool. DD
Well you can see the spreadsheet (the raw data tab at the bottom left), it's just organized box score data (the columns in pink) and some other columns that calculate things that aren't in a box score (but derive from box score data). That can be done for another team, the difficulty is that you have to input all that data. I'm not knowledgeable enough to write a scripting program that would automatically pull that information and organize it in a spreadsheet.
Not sure if I understand what you mean. You can change either axis from '+/-' to MinPG and also Total Min played.
I am thinking of impact. For instance, Chase's overall +44 to me is more impactual than Trevor's because Chase has done it in far fewer minutes....... Is there a way we could represent the overall imparct of the +/- by including Mintutes played per game as a percentage modifier? Does that make sense? DD
Well I've heard people refer to an 'adjusted +/-', which might account for what you are talking about. If you can find a formula for it, it shouldn't be difficult to incorporate.
I think DD is referring to +/- per minute (or +/- per 48 minutes). So, you could just add that as a category on the left axis.
If i'm not mistaken adjusted +/- is more about adjusting for the strength of the players on the floor with you and the players playing against you as opposed to adjusting for minutes played. The nature of +/- already adjusts for that. If anything, playing more time makes the data stronger. I'm not a huge fan of adjusted +/- but is better than the regular +/-. I think there are some walkthroughs online of how to go about calculating adjusted +/- on your own if you're interested.
Is that necessary? Why do you need to compare +/- by the minute when +/- is already a comparison of how the team performs while you are on the court?
I wouldn't say +/- already adjusts for minutes played. You could have a guy who consistently is a +3 every game who plays 40 mpg, and another guy who consistently is a +2.5 every game who plays 20 mpg. In terms of +/- per minute, the second guy should rate higher. But based on raw +/-, the first guy will always rate higher.
Well, I thought one of the main purposes of per minute stats was to account for production while a player was on the court instead of penalizing a player for playing less minutes. How important is it to know that the +3 guy play 40 minutes and the +2.5 guy played 20 if their +/-'s are only drawn from their time on the court anyway? Ideally wouldn't the guy playing 20 minutes maintain the same +/- in extended minutes?
Yeah, that is what I am saying, trying to find the true impact number...whereas you take the percentage of minutes per game and use that as a modifier to the +/- stat to get a truer impact number. For instance if a guy played 48 minutes and was +10, his impact per game would be +10... But if a guy was +5 in 12 minutes, his impact per game would be +20. DD
The reason adjusted +/- exists is because of the flaw(s) in the "plain" +/- stat where a player can have a high +/- simply because he plays with other good players for most of the game or if he plays against sorry competition (the other team's scrubs) for most of the game. FYI : adjusted +/- won't necessarily be a simple formula - it's a statistical regression analysis.
I don't follow your reasoning. Its analogous to any other stat. If one player scores 30 points in 40 minutes, and another player scores 25 points in 20 minutes, you can see that the second player scores more points per minute than the first player and was likely the more potent scorer (based on this information alone). Similarly, if one player is +3 in 40 minutes, and another is +2.5 in 20, then the second player has more "+/- impact" per minute, and is presumably the more impactful player.
But perhaps we could come up with a pseudo adjusted +/- that is a simpler formula without doing the regression. For example, we could consider MPG as a first-order proxy for how good each player is. Then, adjust each player's +/- based on the average MPG of the players he's playing with and against. I haven't put much thought into it, but maybe something like that could yield interesting results.
Ok I added a category for '+/- per 36 min' (basketball-reference.com uses per 36 rather than per 48, not sure why but I thought I would as well). Problem is Cook and Taylor skew the scale such that it's unreadable, and the applet doesn't let you remove their data from the chart. So I removed Cook/Taylor/T-Mac/Yao from the raw data. If they start playing statistically significant minutes, I can add them then.
This is the failing of +/-. Who was he playing with and against in those minutes? This is why scrubs would show up with elite players in the original +/- system. His "impact" may have been when the team was already up or down 25 points. If you take pure +/- and you do it per minute, you're frying an already screwed up system of evaluation. Adjusted +/- already takes into account the strength of his teammates for every minute he's on the court and this is what pmac may be referring to. It alleviates to an extent many of the problems with "per minute" stats.
OK, I didn't know this was the case. If a player scores 25 in 20 minutes it's almost obvious that he'll score more if he played twice as many minutes. I thought, in the case of +/-, that you couldn't apply the same logic because it is already based only on the time a player is on the floor. So you can reasonably assume that a player's +/- increases with minutes like other statistics? Isn't +/- more like fg%, it takes into account the good and the bad?
