offense >> defense

How about this reasoning?

If you took the best defender in the league, let's say Bruce Bowen (at his peak), and clone him four times to create Team A then take the best offensive player in the league, let's say LeBron James, and clone him four times to create Team B, which team would you bet your paycheck on?

That is probably an easier way to look at the relative importance of offensive compared to defense.

Offensively-skilled players are more important than defensively-skilled players on an individual basis. The top draft picks are almost always offensively-gifted players and the biggest contracts in the league almost always go to offensively-gifted players.

I think the best type of player is typically an offensive-minded player that has the right discipline and attitude to play both ends of the floor. There is always room for a Bruce Bowen or a Shane Battier but the majority of players on every team, including championship teams, are offensively-capable players.

What usually separates these championship teams from the rest of the pack is partially talent but also the ability to put everything together and get it done consistently on both ends of the court. Most championship teams were above average, or better, both offensively and defensively.

 

Bowen has too much offense and James has way too much defense.

Five Steve Novaks vs. Five Hayes. That would be better.

 

that wouldn't be right because he says defence is important but offence is more important

Players like Bowen have no offence what so ever so it doesn't fit his criteria.

His thinking players like Shawn Marion and Dwight. Guys who can marginally defend well but not considered as a true stopper like Battier. But can still supply masses of offence. Artest is the only guy in the league i can think of that really fits what his saying. But the main thing is his theory just got totally killed by the Suns signing Shaq. They signed him to play in the halfcourt for defence. His not even there for their fast breaks. Just half court sets and someone who can defend and rebound against a premier post player (i.e. Yao). The Suns were his poster team but they just went and signed for defence.

 

This was in today's paper

http://www.azcentral.com/sports/suns/articles/0218barkleyside.html

Barkley still loves basketball more than just about anything. He still loves Suns fans more than just about anyone. And he still wants the Suns to win more than any team.

But just as Suns fans' love for Barkley is tested by his on-air critiques, he said their takes on him are off.

"There's one thing that has disappointed me about the fans in Phoenix," Barkley said. "They never came to me and said I was right. They've never said, 'This little team's too soft.' And I was right. I've been right every year. They are like, 'You don't like our team.' I really don't care who wins.

"They can't win the way they were. You can't outscore people. It's got nothing to do with hating. That's just strictly watching the game. That bothers me when they say I was a Suns hater. The game comes down to defense and rebounding. They were terrible in those stats. That's the only reason."

 

From the Daryl Morey chat earlier today:

[rquoter]durvasa: What's the best way to rate a team's offense and defense, statistically (e.g. ppg, FG%)? Do you think one side of the ball is any more important than the other historically? And in particular, is it true that defense is more important than offense in the playoffs? Thanks, and good luck with the rest of the season!

Daryl_Morey: best way to judge offense and defense is points/poss scored and given up per possession. Defense historically has been moderately more important but you generally have to be top 10 at both and very good at something. We are currently 2nd in defense and 17th in offense.[/rquoter]

Frankly, I'm surprised. I thought he would say that both are equally important, but it looks like his analysis agrees with the conventional wisdom that defense is more important in the playoffs. He does stress that it's important to be top 10 in both.

 

I would say that suggusts that we need to be efficient in both. Top 10 in both would be awesome for any team obviously getting it done on both ends of the floor. Rockets offensive efficiency will increase i think.

 

blackbird, I need some time to consider your post more carefully. I'll respond later.

Agreed. Balance is the key thing. You can't expect to win a championship with a really good defense, if your offense is just mediocre. Hopefully, the recent tweaks (inserting Scola in the starting lineup) makes us a much more efficient offensive team. How our rank in offense and defense changes over the course of the season will be well worth tracking.

 

You need both a good coach and good players.
You need both a leader and good teammates.
You need both offense and defense.

The Spurs aren't ONLY a defensive team, they're a great defensive team with good offense as well. That's why they have 4 of the last 9 titles.

The 3peat Lakers weren't ONLY a great offensive team, they shut off the paint and had strong perimiter defense (for the most part).

The Pistons had both a fluid, talented offense and the ability to put the clamps down on a team. That's why they have a ring.

Those are the last 8 seasons for you.

All 3 of those teams had talented players, a superstar/leader, and a great coach.

I would say that defense is slightly more important, but that's just my opinion. I don't know if there are any hard facts to prove it. When your offense has an offnight, you will still score enough points to barely win, granted that your defense shuts down the opposing team. That you can count on every night and you don't need a player who's a talented scorer to play good defense.

But even if you score, if you don't stop the opposing team, you will lose. Or, you will probably lose. The Suns have enough offense (obviously) and they had enough defense with Marion (i'm not sure about Shaq) to win titles, but they had TERRIBLE lead management. They go up by 13 with their brilliant shooting and "early shot clock" offense, but even when they are cold they would do the same thing and piss away that lead.

Hopefully when they dump it into Shaq, that won't happen.

That's my reasoning.

 

That is just a dumb comparison. James is probably a better defensive player than Bowen. He just doesn't concentrate all his energy on it.

If you want to make it a little more realistic how about Olajuwon vs Jordan. They both are great offensively and defensively but Jordan was the most formidable offensive player of his day as was Hakeem the best defensive player in that same era. If you could clone 5 Hakeem's and 5 MJ's and have them play against each other I think the Olajuwons would school the MJ's. Think about all the block shots and rebounds the Olajuwons would have over the MJ's and the drives to the baskets would be almost null going against the Hakeems. WHich means the MJ's would simply become a jump shooting team and the Olajuwons would still have the ability to shoot over and back in and dunk over the MJ's. It's also why almost every championship team is anchored by a defensive center and Russell outplayed Chamberlain. All things being equal defense beats offense.

 

Man, that would be an awesome commercial.

 

I'd like to first recap, just to ensure I am understanding you correctly:

The "fatal error" for a 2-team league, as you put it, is that you can not determine just by looking at the stats if a team is playing good offense or good defense.

My claim is that in a 30-team league where every team plays a roughly similar schedule, you can separate the good offensive teams from the bad offensive teams, and you can separate the good defensive teams from the bad defensive teams. The reason you can do this is that every team, over the course of a season, should face roughly similar defenses and offenses (because of the similarity in schedule).

You are claiming that this is a logical fallacy. You say that a 2-team league can be reformulated in terms of a 30-team league, and a 30-team league can be reformulated in terms of a 2-team league.

For 2-team -> 30-team league, if I understand you correctly, you are essentially saying that you can split up the schedule and rename the teams. As an example (correct me if I'm wrong), we could convert a 5-game schedule between two teams into a 5-game schedule between 6 "teams":

So, we have the following initially:

1. A: 105, B: 100
2. A: 103, B: 98
3. A: 106, B: 105
4. A: 99, B: 104
5. A: 110, B: 101
Average. A: 104.6, B: 101.6

The only thing we can conclude from this in comparing the two teams is that Team A played better than Team B. We can't say that Team A is better offensively (Team B may have just played worse defense), nor can we say that they are better defensively (Team B may have just played worse offense). One thing we can say, and I hope you agree, is that Team A must have played better on at least one end of the court than Team B. For if Team B played both better offense and better defense, then they logically would have scored more points.

Fine, now via your conversion, we might get something like this:

1. A1: 105, B1: 100
2. A2: 103, B2: 98
3. A3: 106, B3: 105
4. A1: 99, B2: 103
5. A1: 110, B3: 101

Note that we have effectively done the following:

A = {A1, A2, A3}
B = {B1, B2, B3}

This particular conversion was for 6 teams, but in theory it could be for 30 teams, or n teams more generally. I will note that, in general, when converting from the 2-team domain to the n-team domain, it must be a "one-to-many" conversion. Otherwise, the conversion wouldn't be reversible, and therefore the 2-team and n-team schedule wouldn't truly be logically "equivalent". For example, it wouldn't make sense for "A1" to represent team A in one game and represent team B in another game. Rather, "A1" will only represent team A, "B1" will only represent team B, etc.

The consequence is that the same problem arises as before. The opponents faced by one set of teams (renamed from A) will be distinct from the opponents faced by the other set of teams (renamed from B). This is much different than a normal NBA schedule, where every team faces each other. Consequentially, in the NBA there is actually considerable overlap in opponents faced, which means that we can assume that offense and defense faced is approximately the same for every team over the course of a season. But leaving that aside, let's examine the properties of our new 6-team schedule in a little more detail.

If I understand your point correctly, you claim that since this new schedule is effectively equivalent to the 2-team league (just renamed the teams on a game-by-game basis), we can not say that one team played better offense than the other, nor can we say that one team played better defense than the other. Actually, I believe you are only partially correct. Before we could not definitively say that Team A was a better offensive/defensive team than Team B. Likewise, we now can not say that any of Teams A1, A2, or A3 are better offensively/defensively compared to Teams B1, B2, or B3. Just as in the 2-team league both Team A and Team B played non-overlapping opponents (trivially ... A only faces B, while B only faces A), in the equivalent 6-team league the opponents faced by teams A1, A2, and A3 do not overlap with the opponents faced by teams B1, B2, and B3.

However, unlike in the 2-team schedule, in our equivalent 6-team schedule there can be cases where there is some overlap in opponents faced amongst certain teams. For instance, Teams B1, B2, and B3 all face A1 in the schedule I constructed above. B1 scores 100, B2 scores 105, and B3 scores 101 points against A1. And so, we have a little extra information to work with. Because all three teams face the same opponent, one may have an expectation that the defense from A1 faced by the three teams is roughly similar. And, consequentially, B2 played better offense in their game against A1 than B1 or B3 did. Now, this is all very rough and not very reliable, since we're only dealing with a 3-game sample. That's not nearly a large enough sample of games to say with confidence that the difference in points scored is due almost entirely to offense and not variation in defense. However, if as in the NBA, B1, B2, and B3 faced a very similar schedule over a span of 82 games, then we can be much more sure that the average defense faced by the three teams over their respective 82 games would be very similar, and therefore we can be significantly more confident in saying that the difference in points scored is due mainly to a difference in offense.

To summarize, when going from a 2-team to 30-team schedule, a "one-to-many" relationship must be used in the renaming process (e.g. "A1" can not represent A in one game, and B in another game). Consequentially, in the resulting 30-team schedule you have two sets of teams, each representing of the initial teams in the 2-team league. And no team from one set will face opponents that a team from the other set has faced (which, let's be clear, is not how a normal 30-team NBA schedule works). Therefore, I agree that we can not compare offense/defense of a team from one set to a team in the other set. However, it may still be possible to compare offense/defense of teams within a given set as I tried to explain above. I will say, and this is something I probably didn't make clear enough before, that one can never be certain that the difference in points scored is due to exclusively to offense (and likewise for points allowed with defense). So, yes, this issue of "separation" (in the sense you talk about it) always exists to an extent. Unfortunately, that's just the nature of analysis in any field... we can only observe and draw probabilistic conclusions, not certain ones. The important thing is that as teams play more and more games over the course of a season and the schedules face become more and more similar, we can get more "separation" for the offensive and defensive stats.

When converting from a 30-team league to a 2-team league, I don't agree that "the disjointed sets problem" no longer exists. For the 2-team => 30-team conversion to make sense in my opinion, you have to create two disjoint sets of teams (e.g. A1..A15 and B1..B15). To reiterate what I said above, it would not make sense to use "A1" to represent Team A in one game, and use "A1" again to represent Team B in another game.