Blackmill wrote:drza wrote:I understand your reasoning, as I've seen variations of it before. But I think, in this case, you're first sentence is correct...I'm not going to agree with you on this. Frankly, the stance that you espouse makes no logical sense to me. I'm willing to, as I mentioned in the post you quoted, agree that pace normalizing isn't an exact science. But, as I also mentioned, I believe that we have to at least TRY to normalize, because otherwise we get results that (to me) make absolutely no sense. Consider.
1977 Lakers had 104.7 possessions per game. The 2003 Spurs had 90 possessions per game.
1977 Lakers sored 106.9 points per game, of which Kareem scored 26.2. So, Kareem scored 24.5% of his team's points.
2003 Spurs scored 95.8 points per game, of which Duncan scored 23.3. So, Duncan scored 24.3% of his team's points.
Even in raw numbers, their scoring averages are similar. But for the premise to be true that star players aren't opportunistic, to use your terms, and thus that the pace adjustment isn't closer to even ground than raw stats, that would suggest that if the 2003 Spurs suddenly played at 105 pace, Duncan would for some reason average the same (or very similar) numbers of shots as he does at 90 pace. Well, then...who is getting the shots in those extra 15 possessions? Duncan scored the same percentage of his team's points as Kareem did...why would playing faster suddenly cause that to change? Why would Duncan not get the expected 3 or so shots out of the extra 15 possessions? Why would we assume that Parker, or Bowen, or Jackson, or Robinson would all of a sudden monopolize all of the extra shots to thus change the proportion of the offense that Duncan utilizes to any significant degree?
I understand the notion that we can't know for sure what would happen. But to me, we can only use the best logic that we have. And in my opinion, I can't use raw numbers if one player is getting 15 - 20 (sometimes more) more possessions than another. It seems that you're settled in your feelings on this, and I respect that. But to me, pace normalizing is much more useful than raw numbers (or, at the very least, required in addition to purely looking at raw numbers for a different perspective), if one is going to put any kind of premium on boxscore volume in the evaluation process.
I happen to agree. My original post was not to prove pace doesn't matter, but rather, to establish the interaction between pace and production is contextual. In fact, your opinion and mine are not so orthogonal as they may seem, and I think could be made consistent with one another.
Just like there's team statistics and individual statistics, I think there's team pace and individual pace. Team pace depends greatly on team composition and offensive philosophy. I would argue that Kareem played at a relatively slow pace and his teammates at an absurdly fast pace.
For instance, right now I'm watching G4 of the '77 series between the Lakers and the Blazers, and I've noticed that the Lakers play much slower when they attempt to post Kareem. This is most obvious when Kareem goes to the bench. Without him, the Lakers looked to be shooting before the post entry pass would typically be made, were Kareem on the court.
I wouldn't be surprised if the pace, when Kareem touched the ball, was fairly similar to the pace Duncan played at. Granted, that's just a guess, but any one who's watched Kareem's post play knows it's far from fast paced. Given this, and that Kareem can get a shot whenever he receives the ball, I think his per-game stats would be largely unchanged had he played in the 2000s.
I've been thinking about this since I posted last night, and based on your response I do think we can find some common ground. I do recognize your point about faster pace meaning the play style is different, so the exact output change isn't fully known. But one thing that started to solidify for me as I wrote my post last night is...I don't know that this uncertainty really matters to me.
Because yes, pace would potentially change the game in unknown ways. But, so would offensive and defensive schemes (for example). 70s style offense and defense are both dramatically different than 2000s style offense and defense, on a scheme and execution level. Not to mention things like rule changes, degree of talent saturation/scarcity in the league, or (for players further separated in time) training methods, nutrition, racism (thinking Oscar/Russell generation), etc. If the point of pace adjusting (for me) were to generate an accurate look at what a player might average for one period vs another, then I think these other factors would create likely more uncertainty than pace and would thus render the projections untenable, anyway.
But, as solidified for me last night, I'm not LOOKING for the pace adjustment to be an accurate prediction of what one or the other would do in the other's place.
No, instead, I'm looking for what amount of production/weight a player carried in his own time, which is going to be proportional. As I pointed out in that last post, Kareem in 1977 produced 24.5% of his team's points while Duncan produced 24.3%. In the exact environments that each played in, this is the proportion of the team's scoring that each carried. THAT'S, I think, what I want to know. NOT what one WOULD HAVE produced if subbed for the other, but how much weight DID they carry, in their given environment.
And, to finish that thought...pace adjusting allows me to see the amount of box score weight that each player carried in their given situation, but still using the norms. It's harder to mentally contextualize the difference of 1% of rebounding percentage (for example), but with pace adjusted numbers I can see that one was producing 17.5 rebounds to the others 16.6 rebounds, when normalized to their time.
So, ultimately, maybe (?) I'm saying that I can agree that pace adjusting wouldn't yield exact estimates of what either could do in the other's shoes. But what I'm looking for, and what I feel that I get from pace-adjusted numbers, is a better idea of how each performed in their own shoes, relative to the time and team that they played in/on (as measured by the boxscore stats). Then, with those quantitative estimates established, I can consider things like how pace, schemes and talent saturation/scarcity may change how impressive I believe their relative production to be. I don't think that it's necessary that I fit those latter elements into the numbers themselves, in order for the numbers to still be useful to me.
