Celtics 39.4 projected wins by box score geeks

[jyyzzoel]

I saw this site linked from wages of wins:
http://www.boxscoregeeks.com/previews

Interestingly Brooklyn has only 4 more wins. In hindsight I remember last year wages of wins predicted the failure of the lakers project nearly exactly with their win loss record when everyone else was screaming championship.

[DarkAzcura]

Props to them if they are right, but that has to be some of the worst projections I have ever seen. I'd love to see their formula. The Pelicans aren't going to be the third worst team in the league. That's just crazy. The Pacers won't finish under .500 (though I do agree with the sentiment that they are overrated). Brooklyn only finishing slightly above .500? Only happening if KG and Pierce play half the season, and seeing as you can't really predict injury, that's a really bad projection.

Seeing the Celtics and Pacers neck and neck..then looking at the rosters..just makes me laugh.

I wouldn't be surprised to see the Knicks fall off, though. Too many volatile personalities. Has potential to explode. BUT even if they do explode, they are still not the worst team in the division. I can't believe there is any formula out there that would have the Knicks 5th in our division. I can see 35-40 wins, but that would mean the C's and 76ers would probably bottom out.

I agree with their top 7 (different order, though, plus Brooklyn needs to be in there), but everything after that is a mess. Looks like they picked team names out of a hat.

I will give them a little leeway, though, because this offseason had TONS of player movement. There are a lot more question mark teams than any other year I can really think of (Pelicans, Wizards, Pistons, Knicks, Mavs, Bulls return, Brooklyn, Jazz, Raptors, Celtics, and the Cavs). It's probably hard to place these teams, but they have to do better than that haha

[Celtics4ever]

I think we will be lucky to win 29.

[D.o.s.]

Interesting to see how that'll turn out.

Is that equation run with the current roster?

[sofutomygaha]

Value comes from inter-connectedness, and the notorious problem with evaluating basketball players is how you can divide them and divvy it up. This will be a great learning experience for boxscoregeeks.

They've got a lot of our wins coming from Rondo, Wallace, Humphries, and Lee, but those are all complimentary players, or at least they have been thus far. It will be really, really interesting to see what kind of value they create when they are the best players on the court.

[LooseCannon]

Props to them if they are right, but that has to be some of the worst projections I have ever seen. I'd love to see their formula.

I can't seem to link directly to it, but the basic methodology is described if you click on the link to "Arturo's Preview Recipe" at the top of the page.

[erisred]

Value comes from inter-connectedness, and the notorious problem with evaluating basketball players is how you can divide them and divvy it up. This will be a great learning experience for boxscoregeeks.

They've got a lot of our wins coming from Rondo, Wallace, Humphries, and Lee, but those are all complimentary players, or at least they have been thus far. It will be really, really interesting to see what kind of value they create when they are the best players on the court.

Role players with new roles. If they can do with their new roles what they did with their old ones, then 39/40 wins is possible, but can they?

Personally, I think the C's are going to be a real wild card this year. They might win 40, they might win 20, but they are likely to come in somewhere in between. My guess is 28 wins, give or take 3...all things being equal. Injuries, trades and in-game substitution patterns could change that win total a lot.

[Q_FBE]

That is the biggest pile of malarkey I have ever seen. Miami with 57 wins is way too low in my opinion unless Lebron takes half the year off. How are we neck and neck with Indiana? Perhaps BoxScoreGeek is confusing us with Butler or something. Oklahoma and Indiana is severely underrated and Dallas is severely overrated. There is way too much parity here in the projected wins column. In reality, there will be 12 strong teams, 6 middling teams, and 12 weak teams that might win less than 33 games in a race to the bottom. Assumming the 12 weak teams go an average of 27 and 55 that is 660 out of 1230 losses used up. The average record of the top 18 teams will be some where around 49 and 33 and the playoff cut line in both conferences could be as high as 46 wins, 36 losses depending on how many teams tank.

[LarBrd33]

... I'd love to see a fan who looks at those projections and goes, "Yes!  see! ... I told you!... Boston will be the 8th seed. Woo hoo!"...  Cuz if that fan were to scroll down they'd see these same projections have the Lakers winning the Wiggins sweepstakes.   Talk about a nightmare projection. 

Philly winning 35 is hilarious.

[Celtics18]

... I'd love to see a fan who looks at those projections and goes, "Yes!  see! ... I told you!... Boston will be the 8th seed. Woo hoo!"...  Cuz if that fan were to scroll down they'd see these same projections have the Lakers winning the Wiggins sweepstakes.   Talk about a nightmare projection. 

Philly winning 35 is hilarious.

Most people--but not all--don't use pre-season predictions as cause for "I told you so!!"s.

[PhoSita]

I think when you get results like this from your simulation, it's time to look at your simulation and question your methods.  It simply flies in the face of all common sense basketball analysis.

[RyNye]

There's no reason for people to get their panties in a bunch because they disagree.

Not only is their basic methodology very clearly described, but they have a separate post breaking down their logic for each individual team. They even have a big headline right on the home page that says, "Hey, these are just pre-season predictions based on our model and how we think minutes will be allocated. Expect these to fluctuate wildly as the randomness of the season begins to unfold."

Obviously, you are free to take it or leave it as you see fit, but it is extremely arrogant for random fans to just waive your hands and say, "No, these are wrong this team is going to be better" without even taking a modicum of effort to understand why they project teams where they are or explaining why you think that isn't the right way to think about things.

Whether or not you agree with their projected outcomes, at least these people have a specific model and methodology as opposed to just pulling opinions out of your bum. If you are going to respond, you might as well say things like, "Oh, well, the Pelicans will be better because I think Tyreke Evans will play better in a new environment" or "Oh, well, the Pacers will be better because they project a decline in David West this year due to age, but I think they overrated that decline." Don't just say, "deerrrrrpppp da Nets are gonna win 68 games rabble rabble." It doesn't make them look stupid, it makes you look stupid.

[RyNye]

I think when you get results like this from your simulation, it's time to look at your simulation and question your methods.  It simply flies in the face of all common sense basketball analysis.

That's an extraordinarily non-scientific way of thinking about things. If you are trying to develop robust statistical models like these people are, you can't just say, "Well, this doesn't line up with my preconceived expectations so it must be wrong."

You wait until the season plays out and see what your model got right and what your model got wrong. For example, for the Pacers, they project that Roy Hibbert won't sustain his high level of play and that both Scola and West will decline precipitously because of age. That's why they project fewer wins. If Scola and West don't decline as much as they think, and they win more games as a result, that doesn't make the simulation/methodology wrong, it makes the information they plugged into it wrong in retrospect. That's an extremely important difference.

But, of course, if you actually make any effort at reading into their methods instead of blandly dismissing them because you think you know better, you would already know this.

[PhoSita]

Value comes from inter-connectedness, and the notorious problem with evaluating basketball players is how you can divide them and divvy it up. This will be a great learning experience for boxscoregeeks.

They've got a lot of our wins coming from Rondo, Wallace, Humphries, and Lee, but those are all complimentary players, or at least they have been thus far. It will be really, really interesting to see what kind of value they create when they are the best players on the court.

Great points all.

I usually think of myself as being on the side of advanced statistics in basketball, but I think these predictions fail to take into account the big picture and focus too much on individual contributions from individual players. 

That approach might yield solid results in baseball, but not so much in basketball.

[PhoSita]

I think when you get results like this from your simulation, it's time to look at your simulation and question your methods.  It simply flies in the face of all common sense basketball analysis.

That's an extraordinarily non-scientific way of thinking about things. If you are trying to develop robust statistical models like these people are, you can't just say, "Well, this doesn't line up with my preconceived expectations so it must be wrong."

The thing is, sports are not science, and treating them like science is silly. 

We're not starting with a blank slate at the beginning of an NBA season.  We already know how basketball works; we already know how these players have performed in the past, and more importantly we know how well they have played together -- or we know that they haven't played much at all with the other players on their team, as the case may be.

Because of that knowledge, I think it makes sense to look at the results that their model spits out and say hm, well, either much of what we think we know about these teams is very wrong and past results were a fluke / generated by different factors than we assumed, or our model is failing to account for some pretty enormous influences on team performance over the course of the regular season.