I personally think more time should be spent conquering the concepts of “going by the book” than on “Moneyball” types of batting and pitching data and derivatively compiled data like OBP + slugging. These are basically useless because data like that are affected by situational play that is hidden in these kinds of gross compilations.

For instance, a guy who hits behind Rickey Henderson isn’t going to have the batting or slugging percentages or RBI’s that he might have had hitting sixth with that same team; he is taking good pitches to let Rickey steal, then hitting to the right side to get him to third with one out. His purposeful at-bat is good baseball, almost as good as a double in that situation, and, more importantly, is what management asked of him. But Moneyball punishes him for that.

Forty years later, his numbers are unimpressive until you realize this guy with his meager numbers was in the playoffs half his life, and he was really twice as useful as the guy who did hit .300 on that team, but was incapable of letting Rickey do what he did. Guys with all kinds of numbers frequently weren’t winners because of situations, ie, Ernie Banks; other guys had numbers expressly because of situations, ie, Ruth and Mantle who always had protection in the lineup.

What intrigues me instead is the mathematical “statistics”—not data, but probabilities and odds that can be compiled from the mass of situations to develop in-game and season long strategies of play. (To be clear about the difference, the number of cards in a deck is technically data, and the probability of getting a pair in a poker hand is a statistic.)

For instance, what is the overall probability of scoring when the situation starts with a man on first and no outs? If a man is on second with no outs, do you play for a big inning, or simply try to secure the almost sure run that comes from that situation? Do the odds say it really makes sense to bunt a man to second with one out because there is a higher probability of getting him in from there with two gone than from first with one out, even if the pitcher is up? Should we reconsider the concept of never bunting a man to third with one out because of the odds?

Should a visitor be more willing sacrifice in extra innings that otherwise because a one run difference is about as good as a larger lead in that spot? Is a 1-1 game after 9 more likely to go 6 extra innings than a 7-7 game, so the pitching has to be juggled based on that?

If a pitcher has given up two long hits in a row, overall, is he more likely or less likely to give up hit than in other situations, and how should hitters approach those at-bats?

Does playing the odds (going by the book) mean preferring to play for big innings or playing small ball at different stages of a game?

It is this type of Q&A;that really helps people watching a specific game appreciate what is actually going on in front of them and to predict how it turns out than some esoteric derived number that doesn’t separate situations where there is pressure from others.

So, I’m putting in a poll about Moneyball, I’d like to hear comments too.

Ros - 22 April 2008 12:19 PM

I personally think more time should be spent conquering the concepts of “going by the book” than on “Moneyball” types of batting and pitching data and derivatively compiled data like OBP + slugging. These are basically useless because data like that are affected by situational play that is hidden in these kinds of gross compilations.

For anyone who hasn’t read Moneyball (which includes me), could you summarize what particular ideas you’re referencing from that book?

Ros - 22 April 2008 12:19 PM

For instance, a guy who hits behind Rickey Henderson isn’t going to have the batting or slugging percentages or RBI’s that he might have had hitting sixth with that same team; he is taking good pitches to let Rickey steal, then hitting to the right side to get him to third with one out. His purposeful at-bat is good baseball, almost as good as a double in that situation, and, more importantly, is what management asked of him. But Moneyball punishes him for that.

I partially agree with your basic principle. But batters are placed in the lineup according to their expected production. The best batters aren’t put at the 6th and 7th slots. And a #2 hitter can’t generally be just someone who moves runners over. Even Rickey Henderson gets out 60% of the time. Also, with two outs, the #2 hitter isn’t going to be giving up an out. And the #2 hitter is going to lead off his share of innings as well. So it’s somewhat of a misnomer to say that the #2 hitter’s job is to move the leadoff hitter over.

Ros - 22 April 2008 12:19 PM

Guys with all kinds of numbers frequently weren’t winners because of situations, ie, Ernie Banks; other guys had numbers expressly because of situations, ie, Ruth and Mantle who always had protection in the lineup.

Team context can be adjusted for. I’m a proponent of team-adjusted R and RBI.

Ros - 22 April 2008 12:19 PM

What intrigues me instead is the mathematical “statistics”—not data, but probabilities and odds that can be compiled from the mass of situations to develop in-game and season long strategies of play.

From the examples you’re giving, this would lend less to evaluating a player’s worth and would lend more to strategizing in situations.

Have you used Diamond Mind simulators? I’ve heard good things about it, but haven’t used it. I’m wondering how much it lets you experiment with these things.

Ros - 22 April 2008 12:19 PM

It is this type of Q&A;that really helps people watching a specific game appreciate what is actually going on in front of them and to predict how it turns out than some esoteric derived number that doesn’t separate situations where there is pressure from others.

But then there are different goals, as mentioned before. One could be to measure the relative worth of each player, while another could be to make the best strategic move based on probability. I think they can peacefully co-exist.

For anyone who hasn’t read Moneyball (which includes me), could you summarize what particular ideas you’re referencing from that book?

The big concept is that a player’s performance statistics can be massaged and then divided by dollars and cents so as to rate him economically, and thereby assemble a team that is the most cost effective. Sometimes this is done by amalgamating the odds of the results of particular situations, and projecting from how often a given player creates or finds himself in that situation to how many additional runs are scored or prevented, and which is then further projected into how many games are won or lost due to the situations found.

The grand idea is that figures don’t lie, whereas vague impressions, such as body type, intelligence, and so forth are easily misunderstood and overweighted in analyzing players.

But what it misses, it seems to me, is the greater concept that while figures don’t lie, liars figure. Sometimes the math that is done, it seems to me, is outcome determinative; ie, the figurer selects the math that helps justify his case, and ignores the rest.

It is fine with me to go through the process of questioning the “conventional wisdom,” ie. that a successful team needs a specialty closer. I’m not sure that is capable of objective determination, but it is a proper question. What I don’t think works is trying to build a team around the manipulated and derived and massaged minor league results of player X or Y; the “moneyball” numbers.