The end is Neyer

Rick, you always say "scientifically analyzed," but it isn't. True science depends on a control group. You can analyze a million innings of data about baseball, but there will never be a control group in that data. To the question of walks as one example, you will never know what the opportunity cost was of taking a walk. Did the batter pass up a pitch he could have hit out of the park? Would the runner have taken an extra base on a hit to right?

Also, the analysis is not peer review as the scientific community defines it. It's other hobbyists and website operators taking a look, but even though DIPS looks very reliable, in no way has it become "scientifically" accepted.

A lot of conclusions can be drawn in the numbers, but it isn't science. It's more like stock-market analysis, and we all know how horribly wrong those equations can look in retrospect.

 

Yeah, Rick, my thinking was in acknowledging some intangibles, and character and work ethic and "upside," that don't get captured by snapshot numbers. Not disputing the numbers for own sake.

Still trying to find the stats, for instance, that adequately assess the so-called "glue guys." Not just energy grunts who flail about but the players who indeed are more helpful to good teams than they are to bad ones. The ones who know, like a good butler (going by the movies here, not personal experience), know what's needed even before the head of the house knows he needs it. You spill the wine, turn around and Jeeves already is there with a towel and a fresh bottle. On a bad team, which needs so much, those guys are wasted and often not even good enough to get on the floor.

 

Of course there are control groups.

Let's say I want to study if the a high batting average with RISP is a skill separate from batting average. I rank all the players for my sample (one season, 10 seasons, whatever) by the difference between their RISP BA and their overall BA. I take the top 20% from that sample, that's my subject group. I take the bottom 20%, that's my control group. If there's a clear difference between the two groups the next year (or 10, or whatever sample I want to study), then it's a skill. If they both regress to the mean, it's not a skill.

As far as your scenario about the "opportunity cost" of taking a walk, that's only relevant if we are interested in comparing the batter's value to the value of an imaginary version of him who didn't take the walk. But we aren't doing that. We are comparing his value to other real players.

Your "Hobbyists and web site operators" are pretty darn expert on the subject. Their peer review is a lot more rigorous than you think.

 

My friend is a huge Reds fan and he constantly grips about Adam Dunn. How glad he is that he's gone. How he strikes out too much. And on and on and on.

I try to explain to him that you've got to compare Adam Dunn's production to other players to determine his value. Not Adam Dunn's production to my friend's ideal version of Adam Dunn.

It never sinks in.

And he's a freaking CPA.

 

That is still retroactive data analysis, very valuable but not science. It would be science if the batters were hitting the same pitches from the same pitchers on the same day in the same ballpark.

 

So you are arguing that nothing that analyzes data retroactively is science?

Goodbye economics, political science, and pretty much all social sciences.

 

Plus/minus has shown promise on this front, and it's so simple: points for and against when a player is on the court. Obviously this can be skewed by a player's spot in the rotation, but there's some value to be had there. Last year the Warriors had Monta and Ellis and Stephen Curry, and as good as both of them looked individually, they were awful playing together. CJ Watson paired with either of them was a much better player in the plus/minus. Watson is gone this year and now the Warriors are realizing they can't continue with Curry and Ellis in the same backcourt.

 

I'd love to hear from a math/science professor (doctorquant?), but it has always been explained to me that the only way to draw an absolute conclusion is with a control group and observation of the data and experimentation methods.

Economics? Political science? Social sciences? I'm trying to think of a study that is accepted as truth regardless of political affiliation, religious belief, etc. Paul Krugman tells us three times a week that he has proof that the Republicans are full of shit.

 

Take, for example, the political work of Nate Silver. What he does is science.

Or climatology. They can't reboot the earth and try the same weather patterns over again with a control earth.

Science is about observation. Experimental observation, which is what you are thinking of, is not the only way to get data.

 

Apparently, your sense of humor is .7 of the normal SportsJournalists.commer sense of humor.

 

The character of those guys may be in question but not their athletic abilities. The egg Heads probably never played sports and this is their idea of competing.

 

"They never played the game"? From a sportswriter? Priceless.