Perfect, thank you. Now, I see an assumption and a simple logic. Are you really sure this assumption is perfect? Have you ever heard "You can never put your foot in the same river twice!"? And you call it is a simple logic? B plays A, then B plays C, do you really think B will play the same level of defense again? Even if, B plays A, then B plays A again, do you really think B will play the same level of defense again? Have you ever heard "coach's game adjustment"? Have you ever heard "players' game adjustment"? Anyway, how can B plays the same "level of defense" against both teams? When you build your stats on this kind of crappy assumption and flawed logic, what can I say? .
You may not think that's realistic, but for now assume it is. I will explain it further after you answer the question. If Team A is able to score more than Team C against this hypothetical opponent, can we infer that Team A is playing better offense than Team C against them?
If assume the assumption is true, I can say yes, we can infer that Team A is playing better offense. I think I know what you are trying to say. But in reality, beware that the assumption is false, the hypothesis is false. In logic, you need to know "A false hypothesis implies any conclusion." .
I'm just trying to ensure we are on the same page. I know that it's not realistic to assume that team B will play exactly the "same level of defense" each time. But we established a rule that I will use: (1) if two teams plays against similar level of defense, then the team that is able to score more points (per possession) is the better offensive team. Now I can continue. Suppose you have Team A and Team C that play in the same league with a finite number of teams and a randomly chosen schedule. They each play n games. It follows that as n increases, Team A plays against approximately similar defense (on average) as Team B. Therefore, as n increases, from (1) we can get an increasingly clearer picture of which team plays better offense by comparing their points scored (per possession). Do you agree or disagree with that argument?
I'm sorry, we are not on the same page. The problem is here, you think it is a rule, but it isn't. It is an assumption. If you want to use it, you must make sure it is perfect (or true). But how can you prove that the assumption is perfect? You must make sure two teams plays against similar level of defense first. The problem is that how can you know they are similar level of defense? Now, go back to the original question, how to separate offense and defense? Actually, you are in a circle, aren't you? You are still in a circle. Actually, all your stats are based on the above assumption. You need to make sure the assumption is perfect. .
blackbird- You are overly concerned with the potential for error due to erroneous assumptions. As the sample size increases, that potential for error decreases exponentially and you can actually disregard it for practical purposes.
It is not just a small error, it is totally wrong. I meant the assumptions(foundation) were wrong. My question is how to separate offense and defense? You can't separate offense and defense from stats such as "95-89", can you? There is nothing (no relative offense and defense information) in those stats. No matter how to increase the sample size, you can't create anything from nothing, can you? I think I have already said everything very clearly in my post #158. .
I'm sorry, but you're just wrong. There IS "relative offense and defense information" in a single game score. Relative to Team A's defense, Team B's offense was able to score 89 points. See? What's lacking is any absolute information on the level of offense/defense played by the two teams. Now, when you add more games you begin to see the big picture. No, a team does not have the same capabilities from game to game, but over time the data will converge towards their average capabilities. This is what you're not understanding: the more data available, the less likely we are to be wrong about the AVERAGE level of a team's offensive/defensive capabilities. You have a very good opportunity to learn something about statistical analysis from somebody who has a good understanding of stats and what the mean (durvasa). Your arrogance, however, has prevented you from taking advantage of this opportunity. I suggest you at least entertain the notion that you may be wrong as a starting point.
kid, think first before saying anything, don't celebrate too quickly. Ok, Relative to Team A's defense, Team B's offense was able to score 89 points. Then, tell me what Team A's defense was? Please do tell me that relative to Team B's offense, Team A's defense was -89 points. And again, no matter how to increase the sample size, you still can't overcome the above dilemma, you still can't separate offense and defense. Maybe it is too difficult for you to understand. .
This will be my last attempt to show you why you're wrong. After this post, you and your ignorance are on your own. Let's say there is a basketball league with four teams: A, B, C, and D. Each team plays the other three once in a season. In Team A's games, it averages 110 points per game and allows an average of 90 points per game. Team B averages 95 points per game and allows an average of 105 points per game. Team C: 105/110 Team D: 90/95 In other words, against the average team in this league, Team A scores 110 points and allows 90 in a game. Overall, the average team scores and allows 100 points in a game. Therefore, Team A is a better-than-average offensive and defensive team (10 ppg better both ways). Do you agree?
This is so basic. If they are playing in the same league, then over time they will face similar competition (consequentially, a similar level of defense faced, on average). I don't know why you still can't see it. It follows directly from the law of large numbers.
kid, I've already said namy times: "It is not that easy". This topic is just too difficult for your little brain to understand. I've already said everything very clearly in my post #158. A plays B in 1 games: score 105-88 (There is no way to separate O & D) A plays B in 3 games: average 102-91 (How can you separate O & D?) A plays B, C, D in 3 games: average 110-90 (How can you separate O & D?) .
I've already answered your questions very very clearly in my post#158. The key is: you can't create anything from nothing, can you? No matter how many teams or how many games, you still can't create anything from nothing. Seems, you did understand that there is no way to separate offense and defense even if A plays B 10000 times. Why 30 teams or even 300 teams are so important? .
Actually, as durvasa has pointed out, it is a very easy concept to understand. I hope one day you will finally get it. Until then, peace.
Statistically, playing a big number of games can give a good prediction of future results - how many points a team will score and will give up. To measure offense and defense, I think you also need to look at stats like FG%, opp FG%, points per possession stats,etc...just looking at total ppg can hide the pace of play of the teams For example, a team can milk the shot clock on every offensive possession - that will likely result in lower number of possessions and points but it doesn't mean they are a good defensive team.
blackbird, you lack some very basic reasoning skills if you're still asking that question after all of this. Or perhaps you just don't want to admit you're wrong, since that would make you look pretty foolish after all the insults you've been throwing at people. The issue of how to evaluate a team's offense and defense has been resolved long ago. It's been accepted both by people who have a deep understanding of the game (e.g. Dean Smith), and people who have a deep understanding on analytical methods (e.g. Daryl Morey). Experimentally, it's been verified to work. This is pretty solid science. It's sheer arrogance and stubbornness on your part to dismiss all of this as "creating anything from nothing". I mean, using your so-called "logic" one could argue that rebounds tells us nothing about a team's ability to rebound the ball (can't separate the team's ability to rebound from the opponent's ability to rebound), and turnovers tells us nothing about a team's tendency to turn the ball over (can't separate the team's tendency turn it over from the opponent's ability to force turnovers). That simply is not a tenable position if you want to thinking analytically about the game. Good luck trying to prove offense is more important than defense with that view. I've spent far too much time trying to make you understand this, but it's becoming increasingly clear that this is headed no where. I don't know if it's a lack of aptitude or willingness on your part, and frankly I no longer care. I'm done.
Raw points isn't helpful, that's true. Ultimately, though, maximizing points scored per possession and minimizing points allowed per possession is the goal of any offense or defense. FG%, rebounding, limiting turnovers, etc. is just a means to that end.
I dont know what to say Durvasa. Dude just doesnt believe that you can get solid defensive or offensive stats and probabilities. This IS NOT football. A large lead does not cause the other team to panic and abandon their game plan to "air it out" to try and move the ball and score quickly. In football, your theory holds a bit of water Blackbird. Some teams who score lots of points put the other team at a disadvantage and this can make their defense appear better than it is. (Even in statistical analysis)
