Rec’d, most certainly. A very interesting take that is well reasoned and thoughtful, even in the parts I don’t agree with.

In this way, basketball-reference.com, Wages of Wins and other analytic sources are the CliffsNotes of the NBA.

Yes and no. CliffsNotes are abridged, certainly. Statistics do not capture everything but they are not abridged. Statistics, provided they are accurately recorded and while I’ve found a few errors in attribution over the course of time [Pietrus was credited with 2 rebounds that Biedrins got 2 years ago FWIW], by and large are comprehensive. It is not possible for anyone to watch every second of every game, but a statistical analysis can be the product of every second of every game. It’s not a complete recording of every action and interaction, but it’s a complete capture of much of the resulting details. It tells you “what” much more than watching every could, but it misses a base “why”, and that “why” is what makes games entertaining.

My counter to those who say “stats don’t tell the whole story; you have to watch the games” is that I don’t know of any sports stats freak who doesn’t watch many, many, many games. But when the game is over, we can’t get enough, so then we dive into the records to see what we missed.

Stats don’t tell the whole story, but if you work under the assumption that past results are at least a partial predictor of future [and if this weren’t true, then trying to get a “good player” would be meaningless], it’s pretty clear that evaluating the past by evaluating statistics is a good way of predicting future results. Knowing nothing but the numbers, you can come up with a rather accurate prediction of how teams will do given the personnel associated with those numbers and good analysis of those numbers [e.g. not getting swayed by ppg].

Watching, I’ve found (and I’ve watched a whole lot of basketball) certainly doesn’t tell the whole story either. Watching games tends to over emphasize scoring, underpenalize turnovers (unless they are very, very excessive) and missed shots and tends to credit how someone gets results even if it’s irrelevant to the final score. It also tends to lump defense into either “good” or “bad” when indeed the variation within is highly important. Everyone can form and opinion from “watching” and there tend to be a whole lot of people out there that seem to believe they’re among the best talent evaluators on the planet due to “watching”. Most are not.

This is the sort of thing that keeps me coming back to GSoM amidst a slew of trade machine fanposts.

Following your line of thinking further seems to yield the conclusion that no quantitative measure- not even who wins the game, or who wins the championship- can ever “tell the whole story.” I’ve always agreed with this, that the game itself validates its existence; it needs no winner/loser dichotomy to have meaning.

Yet what do players, coaches, and fans claim over and over to be the raison d’etre? “Win.” “Championship.” It’s as if all of us in this basketball-industrial complex are afraid to acknowledge that we, as grown-ups, are devoting millions of man-hours playing, coaching, and watching a game- it’s gotta have a higher meaning, right? Well… maybe not. It’s really just a game, a damn fine one, and that’s good enough.

I’ll start off by saying this was a well thought out and enjoyable post.

Now, I have a problem with some of content. I think you’re overcrediting how good many of the stats are.

It pisses me off that a website can tell me who played well and who didn’t without giving me anything else from that experience. The problem is, they pretty much can.

Sure, there are bits that statistical analysis doesn’t pick up on. It can’t make value judgements for us. We do require actual scouting to see what the stat sheet misses, but the truth is, it’s really not that much. As long as you look past PPG and get into the more advanced analysis, you’ll find almost everything that happened on the court represented by a number. It’s disgusting, but it’s just about dead on.

I know that basketball-reference.com will let you know almost all the relevant details of a player’s NBA career.

Now, I’ve only really started learning in depth about the statistical aspect of basketball in the last few months when I started posting here, so I don’t claim to be the most knowledgeable person on this particular topic. I do think, however, that my stats knowledge in general stacks up with pretty much anyone here, and I see a lot of problems with the stats that I’ve seen. Basically, the problem stems down to the individual vs. team.

It seems to me that team stats sum up almost everything worth knowing – between team rebounds, offensive/defensive efficiency, turnovers, steals, blocks and whatever else we can pretty accurately describe what happened and who won, as well as come up with a very good model for predicting the future. Basically team stats are at a very high level.

That said, when it comes down to giving credit to each individual for a team result (wins/losses), I think we have some serious issues that are often unaddressed. My impression is most of the more “advanced” player stats tend to be derived from a regression model based on the coefficients for how important each stat is at a team level. I have very strong reservations about this methodology. Unlike a sport like baseball that’s almost entirely individual, in basketball what each player does will affect their teammates and opponents. A good passer like Steve Nash or Chris Paul will cause their teammates to score more efficiently, and possibly in higher volumes. These teammate effects are a problem in this type of methodology. I also see some serious data issues with just taking box score results and trying to translate that into a players contributions. Many aspects of what goes on in a basketball game are simply not recorded – screens, ball movement, passing that doesn’t result in an assist, boxing out, off ball movement that gets a teammate open, etc – and so nobody is credited with these things. At the team level it doesn’t really matter because the part of it that matters – how it affects the end result – is taken into account because we record the end result for the team, but at the individual level it’s simply left out. This is a problem. I also see lots of issues with measuring defense (even baseball has serious issues measuring baseball) at the individual level, that again are irrelevant at the team level since we record the end result.

If someone wanted to pay me to figure out how to credit individuals correctly, I think a complete overhaul of how we measure a players contributions would be in order. The box score simply doesn’t do enough. So basically, my point is, there are still serious issues measuring an individuals contributions to the team result, so I strongly disagree with your points that we’re at the point where the stats are almost an end all to describing how good an individual was, or even how helpful their contributions were.