Joe Morgan hates Moneyball -- igophoto Life Hosting

Joe Morgan hates Moneyball

< prev

next >

Joe Morgan hates Moneyball -- Wed, Jul 6 2005 04:09PM

This article in SF Weekly takes Joe Morgan to task for dismissing the school of Sabermetric thought. I hadn't really noticed Morgan's disdain for Moneyball, Billy Beane, etc., but then again I haven't been able to watch a ballgame on ESPN for about a month now.

I think my favorite quote from the article is this one that comments on Morgan's broadcasting (in)ability:

"there were things he'd say that were completely counter to the way he played the game. It was the way he'd combine certain ideas. He'll make a reasonable statement, then combine it with a totally outlandish statement that makes no sense whatsoever." ... "His logic takes this leap. It's kind of ingenious in its own way." (Surprisingly, Carminati enjoys the ESPN Sunday Night Baseball broadcasts. "You have to be in the right frame of mind," he says. "It's like watching Reefer Madness.")

I've always seen (and appreciated) baseball as a stats-centered game. I don't understand why Morgan can't/won't give some credence to the argument that Moneyball is a valid approach to team building (as opposed to, say, the Steinbrenner approach of buying all the big names).

Is Morgan simply failing to adapt to a new age of baseball appreciation? In the age of the Internet, Sportscenter and ESPNNews, Moneyball, Bill James, and SABR seem to be more valuable than weathered scouts and crusty veterans.

Am I making too much of baseball statistically? Or is Morgan in the minority?

Posted By

Snee

Posted In

My two cents

Baseball

Linked to Users

oreo

aham

Similar Images

76.35%

75.34%

74.89%

74.33%

73.40%

Comments

Well I know nothing about baseball, but I do know quite a lot about math/statistics/game theory. I went ahead and read the article for the hell of it.

From my perspective you have a classical John Henry/Casperov battle with the machine. In this case you have Joe Morgan who has the experience and a quite sophisticated neural net (brain) vs. the most complicated and in depth statistical analysis available.

Now lets look at 2 games other than baseball: chess and go and how they are played. A grand master chess player does look ahead a certain number of moves, however often he makes a move that feels right rather than knowing it will cause checkmate in 40 moves. A machine plays by simply by crunching the numbers and looking as far ahead as it can. It turns out that the second approach is more powerful than the first as was proved in the nineties. However, chess is a game with relatively few options every turn. This allows the exponential growth of possible game positions to be curtailed.

Go is played on a 19 by 19 board and every move you can place a single stone on any space. Your opponent may then respond by playing on any remaining space. Every turn there are an enormous number of options making it exceedingly difficult to look ahead many moves. Even more than in chess the master Go players use instinct and often cannot explain themselves why they chose a move. The result is: computers are completely inferior to humans at playing Go. From what I have read even decent amateur players are better than the best computer programs.

So is baseball more like Chess or more like Go? It isn't too complex on paper in terms of possible outcomes, but still the computer doesn't have access to the same information people do. A coach could look at a player on a particular day and note that he looks tired. Now you could imagine a computer sitting on the field that asks the coach to rate the player's tiredness on a scale 1-10, but certainly that sort of information is less useful than actually seeing the guy. The same thing is true when deciding to hire players. One guy has great stats, but is a notorious alcoholic, drunk driver, racist and beats his wife. How do factor bad PR into the program?

I will close by saying that from the article's biased perspective Joe Morgan does seem pretty ignorant and I doubt I could have an intelligent conversation about this with him. However, it is possible that his brain still knows how to play baseball.

- crazy1armman, Thu, Jul 7 2005 08:10AM edited Thu, Jul 7 2005 08:15AM

I'm familiar with chess, but not with go, so based on what crazy1armman has said I'll argue that baseball is more like chess. Joe Morgan, I think, would argue that it's more like Go because he's played professionally and used instinct/gut, whatever you want to call it, very successfully.

As a fan, I see limitations (as in chess) in terms of the team's roster size and elements (i.e., players=pieces). Just as you're not likely to take your opponent's queen with a pawn, you're not likely to force the opposing team's star slugger to make an out by bringing in your worst available pitcher. Sure, it could happen, but your defense would have to be aligned perfectly to respond to a ball put in play. However, this probably isn't the most useful analogy because the article is more focused on the offensive aspects of Moneyball.

In the (biased) article, it sounded like Morgan's main gripe was that not enough credit was given to elements of "small-ball," like the hit-and-run, stolen bases, etc. I guess the chess equivalent would be risking one piece in an attempt to give another piece a better opportunity at putting the king in check--you might lose the piece you risk, but if you don't you're in a better position than you were before. I'm no grandmaster of chess, but I would think it'd be foolish to sacrifice a piece without being absolutely sure of the ramifications of that sacrifice. Sure, the payoffs are potentially big if the gamble works (see Dave Roberts, game 4, 2004 ALCS), but there's an equal (maybe greater?) potential for mucking things up.

I think there is a strategy (albeit complex) to baseball that edges out players' instinct. The trick is putting together a team of players with the right skills and the mindset to resist instinct every once in a while.

- Snee, Thu, Jul 7 2005 10:03AM edited Thu, Jul 7 2005 10:04AM

"I'm no grandmaster of chess, but I would think it'd be foolish to sacrifice a piece without being absolutely sure of the ramifications of that sacrifice."

okay, so bringing it back to baseball, does simply knowing and acknowledging all the ramifications of a steal justify sending the guy at first? you know its possible he could be thrown out (especially if someone like pudge is behind the plate), so is it worth the risk? i think in most cases, joe morgan would say "yes, it is" and moneyball would say "no, its not."

to me, that's all the argument is- how one defines risk/reward in baseball. at some point, as in all things in life, you have to take a big risk to gain a big reward, regardless of the odds. when in moneyball does it tell you to take that risk? at what point do you "throw out the numbers" and go balls to the wall? does that point even exist in the moneyball frame of mind? i think that's what joe morgan is angry about; he feels that moneyball removes the "feel of the game" from the game. that it doesn't allow players to play on instinct and take as many chances because they are doing what is statistically the best option.

- aham, Thu, Jul 7 2005 10:55AM

how one defines risk/reward in baseball

I don't know exactly how the money ball strategy works, but when it comes to taking risks there is a basic concepts in economics/game theory that are often danced around in baseball and poker, because most people don't know exactly how to describe them. One such concept is called average expected utility. To explain this lets take 2 games outside baseball and analyse them to define what this number means.

First lets say someone offers to flip a coin and pay you $1 dollar on heads and take $2 on tails. Now the idea is to evaluate how much winning $1 is worth and how much losing $2 is worth to you. Typically most people would rate these as 1 and -2 (this is actually slightly innacurate as we will see in a moment). Then you multiply each utility by the chance of it happening.

Average expected utility = 1/2 * 1 + 1/2 * -2 = -1/2
The fact that the expected utility is negative is a sign that you don't want to play this game.

Now lets look at a similar game. You are worth X dollars total if you add up all of the money in your bank account, car, etc. Someone offers to pay you X if they flip heads and to take X if tails comes up.

Average expected utility = 1/2 * X + 1/2 * -X = 0
The utility is 0 ie. its a fair bet. However, most wouldn't ever take this bet (unless X is so small that your life already sucks). Sure it would be great to double your worth, but the cost of losing everything is devistating. The problem is that money's utility is not linear. You instead have to look at the outcome of hitting heads and tails. More realistically you might say heads is good and has a utility of 100, but tails is absolutely awful and has a utility of -100000. Based on those numbers you wouldn't take the bet.

The point is that you cannot evaluate utility in a vacuum. Money is worth more or less depending on the context of how much you have and what kind of person you are. It does have an absolute utility in every circumstance though and that means there is a definite right answer when it comes to taking any risk. Whenever you walk across the street you have a very small positive utility (you cross a street yah!) with a very high likelyhood and an extremely negative utility (you get hit by a car) with an extremely low probability. If you walk across the street it means that based on your evaluation of utility it was worth it.

Now back to baseball. When you steal a base the situation is very similar. You can not just say gaining a base is worth +1 and has a 20% of happening, but getting an out is -10 and has an %80 chance of happening so you shouldn't take the risk. You have to look at the big picture. Suppose you are at the bottom of the 9th and down a run with 1 out and a guy on first. If the batter consistantly hits doubles and is followed by crappy hitters then it might very well worth stealing the base. In this case the utility of stealing the base means a high chance of tying the game, where as if the runner gets out you are no worse than you were, because you were probably going to lose. Now lets complicate it further and say you already won enough games to go to the playoffs. Then all of a sudden winning the game has a very low positive utility and the fact that the man at first could injure himself sliding in on the steal could actually have a higher negative utility.

The point is that when deciding on a risk all factors have to be taken into account and you have to decide as a whole how happy you are with each outcome. You then have to consider the likelyhood of each outcome. I think the computer has to be programmed with absolutes ahead of time. Getting to the World Series probably has utility similiar to keeping your king.

Now if players are allowed to take risks and do what they want thats more fun, but it will probably win less World Series. Unfortunately when managers/money are concerned players having fun has 0 utility.

- crazy1armman, Thu, Jul 7 2005 12:22PM edited Thu, Jul 7 2005 12:23PM

when in moneyball does it tell you to take that risk? at what point do you "throw out the numbers" and go balls to the wall? does that point even exist in the moneyball frame of mind?

Good point. I don't think that point exists in the Moneyball strategy. That said, I think it's also important to keep in mind that this strategy is most useful for small-market teams who can't stack their line-ups with guys batting .300+ with 30HRs and 100+RBI.

If a team adopts the Moneyball strategy of working pitch counts, focusing on getting men on base, and avoiding situations that risk an out (e.g., stealing, bunting, trying to stretch a basehit), they're more likely to win more games over the course of the season than a team that is more reckless on the basepaths or has batters that like to swing at first pitches.

- Snee, Thu, Jul 7 2005 02:17PM

In 1995, during a Reds-Dodgers game on ESPN, Joe Morgan made a comment that might stand as the best combination of ridiculous obviousness, phony profundity and pomposity ever heard on air. Here's what he said:

"The Dodgers seem to play better when they're hitting the ball out of the park."

I challenge anyone (you too, Joe) to say it didn't happen.

- alexanorak, Mon, Jul 11 2005 01:16PM

The mathematicians point of view:

Statistics are very misunderstood, and sabremetrics have a lot of bad statistics in them. On the whole however, they're providing some good heuristics for estimating play value. The reason why Joe Morgan really doesn't like them isn't because he feels they are wrong, it is the same reason why you can play computer generated music to a composer in a double blind test and have him or her call it art, but in a single blind test they will say it is crap.

People feel threatened by machines, and by mathematical models which can make good predicitions.

End result: Sabremetrics appears to provide good heuristic results, but will always be just that, a heuristic. Old timers will always bitch and moan when they feel the game is changing, and we should just learn to humor them.

- illini, Wed, Aug 10 2005 02:15PM

Add a comment

igophoto home [bugs/comments]