Tuesday, July 10, 2007
Some bad advice from an “expert”Posted by Patrick DiCaprio at 5:55am
In a recent podcast, a fantasy baseball "expert" was discussing the expected rebound of certain alleged "second half" players. Let's assume for argument's sake that this is a fallacy; it certainly is based on his analysis. Why are some fallacies so persistent??
I don't mean the clearly crazy fallacies, like my continual efforts to get a date with Charlize Theron based on the fallacy that she likes balding older guys who play fantasy baseball (and what a world we would live in if this were true!). Baseball is particularly ripe for the belief in such fallacies; look no further than "The Curse," or that a fighting team will be roused and perform better the rest of the season (as debunked in the book Mind Game), or that players actually do hit worse with two strikes (as I discussed on my blog), or countless others.
Even geniuses are vulnerable; Isaac Newton was believed to be correct in his theories of gravity until quantum mechanics came along, and if he can be wrong so can I or anyone else! Aside from Newton though, most of the time that a fallacy remains persistent the reason is psychological; everyone wants to understand the world, yet not everyone has the knowledge or acuity to do so. In the absence of a proper understanding of how things really work, one merely devises an explanation that fits their knowledge and experience, rather than trying to add the knowledge to fully understand.
Here are three examples, one baseball related, one gambling related (I would wager that most fantasy owners are interested in gambling; Lord knows I am!) and one that just plain makes me crazy:
1. A slot machine player claims that "the machine is fixed." In a sense this is true; but not in the sense that the player means. Since the player does not understand how a random number generator works, he or she lacks the knowledge to understand what has happened. Similarly, when slot players hit for 100 coins, they say, "If only I played another coin I would have won more," ignorant of the fact that in the time it takes to put in another coin, the random number generator will have cycled through 10,000 different outcomes.
2. The current conventional wisdom on bullpen usage is far from an optimal pattern. Virtually every team uses its bullpen to maximize saves rather than to win. There are lots of reasons for this. They're beyond the scope of this column; there are many sources of analysis of bullpen usage. The point here is that the collective wisdom of managers is simply faulty; they "know" that only a pitcher of unique toughness and mental strength can pitch in the ninth inning despite copious evidence to the contrary. Similar comments apply for the existence of clutch hitters, or at least for the existence of clutch hitting as a repeatable skill.
3. A great many people believe in outright frauds like the psychics John Edward, Sylvia Brown and Uri Geller, to name some of the more famous charlatans. It boggles my mind that even desperate police will seek out these con artists. Since they do not understand how these despicable individuals ply their trade through techniques like "cold reading," they merely believe they can speak with the dead.
Now that I have vented on some of my pet peeves (and I must be crotchety today, as I could have written a whole column on these), lets look at the fantasy issue of the podcaster's list of players he thought would have second half rebounds. Why am I mentioning the second half rebound as a possible fallacy? Well, maybe it is true in some instances, but here, specifically are the players mentioned by this podcaster:
Matt Cain, Erik Bedard, Kelvim Escobar, Jon Garland, Joe Nathan, Francisco Cordero, Francisco Rodriguez.
These are all very talented pitchers. With the possible exception of Garland, all have tremendous skills that needn't be addressed here; we all know them. The premise of the podcast was that these were all "second half" players, because they all performed appreciably better in the second half last year. Do you see the problem here?
Here is a look at Joe Nathan. As I am writing this he has a 2.59 ERA, 1.40 WHIP, 14 saves. Presumably the podcaster was looking at that WHIP and relatively meager save total. So, he said Joe Nathan would rebound in the second half, since he has a history of "second half rebounds." However, the podcaster mentioned only the 2006 season. So let's even assume that there was a "history." Will this be the driving force behind any second half improvement?
Nathan has a 38% hit rate, which is astoundingly high for a pitcher of his caliber. Even for a pedestrian pitcher we would expect a regression to the mean of 30%. Nathan has, in his career, never allowed a BABIP of more than .300, yet now he is close to .390. If he improves, will it be because he is a "second half player"? Well, I for one am glad to know that an "expert" agrees that you should acquire Joe Nathan if you can. What great advice!
Looking at Erik Bedard, we see (as of this writing) a 3.60 ERA, 1.23 WHIP 10.9 K/9IP and 2.8 BB/9IP. This is a stud pitcher. He has a 33% hit rate, which is three percentage points above the norm, though Bedard has always been above the 30% benchmark, surprisingly. Given the K rate and BB rate (with a ratio of 3.9, which is excellent) and a somewhat unlucky hit rate, perhaps his ERA should be a bit lower. If he has a tremendous second half would anyone be surprised? He is a stud right now, and a stud should have a great second half. This isn't because he had a good second half last year; he is just a good pitcher.
Matt Cain is an even more puzzling choice by the podcaster. Last year was his first full season in the majors. Is it any wonder that he had a good second half? He was learning the league, and was about one year into his career (since he had roughly a half year in 2005). Moreover, he was unlucky with bullpen help in the first half, when only 64% of his runners were stranded (second half was a normal 73%). His improved second half, again, was not because of any innate ability to perform better in the second half. It was normalization of his statistics. This podcaster, who allegedly is a statistician, made a prediction on a second half sample size of one.
When looking for second half rebounders, or any "buy low" candidates, I generally don't go about it by looking at their historical splits. They may be valid or they may not be. But to make the prediction on those grounds is lazy, and there is no guarantee that history will repeat itself. Looking at a player's overall skill package, and whether he has been lucky or unlucky in relation to his skills, is how you find them. I am sure the podcaster meant well, but he apparently fell into the same trap as a hapless baseball manager who plays the slots while consulting his psychic.
Patrick is a member of SABR's Statistical Analysis and Science of Baseball Committees and writes about fantasy baseball at The Fantasy Baseball Generals blog. He has achieved the dream of all of his MIT classmates. No, not making millions in the tech markets, but writing about baseball for free. Feel free to send along all insults and comments here.