I think you can still have a useful statistical model even if it doesn't always accurately project everyone's win-loss record. I mean, if your stats model always agreed with the majority view of a team, it really isn't that useful. In this case, its a pessimistic projection. Well, what variables does it take into account? What areas does it therefore suggest that the Rockets will be deficient in? As long as you understand the model and what its capturing, and you recognize what are the significant factors it is not capturing, you can still potentially learn something about the team from the results. It takes a little digging, though, not simply looking at the final win-loss prediction.
It doesn't capture things like roles players are asked to play etc, and types of games and how they fit together. Stats are great, a wonderful tool of incomplete information that do not tell the entire story. I love stats, but they don't always pass the seasoned eyeball test. DD
It also makes sense for Morey to endorse something like this. Disseminating what amounts to false advertisement for advanced statistics. If the publication actually came upon anything actually valuable, or pertaining to some of Morey's own proprietary systems, you can bet that Morey would use all possible resources to acquire/shut down the publication, or at least distance/disassociate with them. It is a field still in its infancy, why would Morey want to promote useful information, jeopardizing his own edge? EDIT: Maybe a little exaggerated, but at least that's what I would do..
So stats are great, you love them, and they are a wonderful tool ... and yet they are way flawed? Everyone recognizes that stats don't tell the whole story. But neither do your seasoned eyeballs (or mine, or anyone else's). I wouldn't say your opinions are flawed on that basis. When someone says something is flawed that implies something that it is logically fallacious. IMO, its typically not stats (unless they are based on false data), but rather bad interpretation of stats that are flawed.
A properly formulated, thought out, and researched combination of those would be much better, don't you think?
Exactly. I would, I think that you need both sesaoned eyeballs and stats to get a more complete picture and even that is not totally accurate. The problem is that most advanced stats are based on bogus made up formulas to begin with so the end result is always going to be flawed. If you start a stat with incomplete data the final outut will also be incomplete. I do, but I am not sure that what is tracked right now would allow you to make a very complete picture. I would love to get my eyeballs on the Rockets internal stats, I am sure they are closer to what I would consider decent analysis. DD
Morey's forward to the Basketball Prospectus last year: [rquoter] If you are an NBA team reading this foreword, please move on to the next book in your pile. There is nothing to see here. Move along ... we’re walking, we’re walking ... I repeat, NOTHING TO SEE HERE! Are all the NBA teams gone? OK, down to business. Bill Simmons, are you still reading? If so, you are about to be a happy man. You know how you said you wanted access to all the interesting analysis teams are doing? This book has it. While I am sure you will quibble with some of the findings (your beloved Allen Iverson does not fare well), anyone who reads what Brad and Kevin have put together cannot help but come out a smarter fan on the other end. I think this is the part in the foreword where I am supposed to come up with a grand connecting thought on how data is revolutionizing sports and the NBA is the latest part of the trend. Sorry, but I am having too much fun watching and analyzing basketball to figure out where it is all headed. All I know is every day we try to make the right decisions, we are pretty sure that we are going to make many wrong decisions and we hope data and analysis will help us avoid a few. You would not expect it, but I think data and analysis makes it more exciting to be a fan. Jeff Van Gundy, our former coach, is my favorite broadcaster on TV and consistently gets kudos for the fantastic job he does. One key to his success is that every time you tune in and listen to him you learn something you did not know before. This book is like that. Every time you read part of this book your enjoyment of the game will go up because it will teach you to look for the things that really determine whether a team is going to win or lose on a given night. Basically, if you consider yourself a basketball fan and you are not reading this book, then you ought to turn in your serious fan credentials. Enjoy! Daryl Morey General Manager Houston Rockets [/rquoter]
Things I gleaned from the write-up: 1. SCHOENE doesn't like Brad Miller, and think that giving him Chuck Hayes minutes will cause the Rockets defense to go down. 2. SCHOENE also thinks Martin will miss a lot of games to injury, causing the Rockets offense to basically stay where they were a year ago overall. Not sure if I agree. First, I think Chuck Hayes will still play plenty of minutes. Second, I think Brad Miller can be a decent defender-- certainly did a decent job on Lopez (but then again, Miller, though not 100%, was largely responsible for Jerome James getting an MLE contract out of a playoff series). Third, I think the offense will improve more than people think. If nothing else, they seems to be moving the ball more as the preseason goes on. Maybe the X-factor is guys finally running the motion offense seriously. Anyhow, we'll see, the SCHOENE projection in the 09/10 version of the book had its hits and misses. I'd like to see what you would get if you had gambled on the over/under win totals in Vegas based on the 09/10 proejction... my guess would be about even money.
Good points. SCHOENE didn't properly value the defensive value Yao would bring to a team last year, and as a result it projected that the Rockets would rate very highly on defense (they ended up being very mediocre). It probably overrated Chuck's defense as a 5. Move Chuck to his more natural position, put some good-sized, veteran 7-footers in the middle (even if they're slow), and add a dose of Jared Jeffries, and the Rockets front-line should be much improved defensively.
No, what people interpret and determine those stats to mean is what is flawed. The numbers, resulting from whatever formulas are used to derive them, are still simply useful data. They cannot be flawed. The rankings, extrapolatied meanings, and thereof, that are attached to the data is where the "flaws" come in. "1 + 1 = 2" is statistical data. "1 + 1 = 2 and therefore the Nets will be the best team in the East" is what you would call flawed statistical data.
Say what? From last season, "According to SCHOENE, they are right on the money. SCHOENE sees Houston slipping to 37-45...." I'll give the Schoene credit for predicting the drop offs in ORTG and DRTG pretty well, but they were still 5 games off. I would not call 5 games off as "played out largely as SCHOENE projected." http://www.basketballprospectus.com/downloads/HOU essay sample.pdf
After skimming through that whole block of text, I think I see the problem: Their SCHOENE methodology hates the Rockets and predicted us to have a worse record last year. As another poster mentioned, its 5 games worse, which is actually pretty relevant. Considering their expectations were a lot worse than what actually happened last year, they should have put that margin of error into their calculations instead of applying the exact same methodology that they did before. What you have in the end is another ridiculously pessimistic prediction that results in a worse record than what we actually got last year. It seemed to me that after they got their result they then had a brainstorming sessions to make up reasons why we would be worse off after adding Yao Ming, Lee, Miller and Patterson to a team that didn't have them a season ago. Or at least that's what it seemed to me, the whole "Hayes is better at D than Miller so the Rox will be worse overall" thing to me doesn't really fly given that Yao will already take at least 24 mins of PT a game. I just wanted to thank the OP for posting this, at least it gave us something to talk about...
Correction: I took a look again, and their SCHOENE projection last year was actually not far off from where we ended up as a team. They projected we'd rank 19th in offense and 18th in defense. We ended up 16th in offense and 17th in defense. They projected 37 wins. Our pythagorean win total last year based on offense/defense was 40. Their concerns about the defense turned out to be correct, while our offense did not drop off as much as they (or, probably, anyone else) expected. Moreover, unlike most of us, when making their projection they pretty much assumed we'd get nothing out of McGrady, which turned out to be correct. I think a lot of fans might have been "closer" in their projection, but for the wrong reasons -- i.e. they assumed McGrady would be a contributor.