This is a topic I touched upon previously in a prior post dealing with second half players. One goal in life, fantasy or otherwise, should be to steer clear of bad advice. There is no shortage of it, and no shortage of experts, mostly self-proclaimed. Just recently I heard two more instances where an expert (who has his own snarky name) gave out some bad advice, so I wanted to share my thoughts. But I won’t be sharing in a touchy-feely Human Potential Movement way. This type of bad advice, couched under the guise of “sabermetrics” or ‘statistical analysis” is one of my pet peeves.
The first item is some pretty poor advice on closers and save percentages. On a podcast around the time of the All-Star break, this “expert” was discussing the save percentage stat. The recommendation was to only view as secure those closers that had 90% save percentages; namely they save 90% of the opportunities. Otherwise, they are to be considered “vulnerable.” Some specific vulnerable targets were Jose Valverde, Todd Jones and a few others. Hearing Valverde listed as “vulnerable” naturally piqued my curiousity.
Clearly this is lazy advice. 90% is way too high a standard and in any event save percentage should have very little to do with your evaluation of whether a closer is vulnerable or not. For reference here are the career save percentages for a few closers:
There are some obvious factors that are more important than save percentage that are not news to anyone, but just to point them out:
- Skills (K-rate, BB-rate, Hit-rate etc): This is probably 75% of the battle right here.
- Manager Usage/Opportunity: This is another 15-20% of the battle.
- Competition in the pen
- History of Success in closer role (only because some managers prefer retreads to unknowns)
Do I think the save percentage is a factor? Well, I suppose it is in a vague sense, but I can say that I have never looked at it as a guide to whether a closer is successful or vulnerable in and of itself, and certainly I would not use some arbitrary standard that is not attained by some of the greatest closers in history.
Here are the percentages for the single season leaders of this century, which tell a similar story:
John Smoltz 2002 55 saves/59 opportunities (93%)
Mariano Rivera 2001 50/57 (87%)
Mariano Rivera 2004 53/57 (93%)
Eric Gagne 2002 52/56 (93%)
Eric Gagne 2003 55/55 (100%)
Francisco Cordero 2004 49/54 (91%)
Chad Cordero 2005 47/54 (87%)
Jason Isringhausen 2004 47/54 (87%)
Jose Mesa 2002 45/54 (83%)
Robb Nen 2001 45/52 (87%)
Kaz Sasaki 2001 45/52 (87%)
These are the top-10 save seasons since 2000. Only five were above the 90% standard. Going a bit further down to the top-25, only nine met the standard.
Will this advice hurt you? Probably not, unless you go and dump Jose Valverde or Francisco Cordero. That is not to say that the save percentage is irrelevant; however it is just a number and is not as important as looking at the overall picture.
The second piece of advic was from the same guy, not surprisingly. The podcaster “invented” a new stat that he called an efficiency stat. Is it a complicated metric of the kind you might find here at The Hardball Times? Not quite. It was a simple ratio stat; for runs it was AB/runs, for RBI it was AB/RBI, etc.
Two problems here. The first is that he called it a ratio of runs per at-bat and RBI per at-bat. But that is wrong and sloppy. What he is measuring is the converse: It is the number of at bats per counting stat. This is the type of error that no analyst or expert should make.
With regard to the “stat” itself, as he defined it, the stat can have value in analytical terms, or perhaps in leagues where you have an at-bat maximum (which he specifically referred to as the “goal” of his stat). But the basis for the stat was that it was supposed to target guys that are “efficient” at accumulating runs or RBI.
Amusingly, this stat was used to give us sage advice like “try to acquire A-Rod or Gary Sheffield.” Thanks for the tip. To be fair though; he did point out that Carlos Pena was having a surprisingly good year and should be targeted over some other big name players. That advice is helpful. But you don’t need a contrived, incorrectly defined stat to figure it out.
Another problem is that at-bats is not the most useful measurement here. Using at-bats in the numerator ignores walks. When trying to accumulate runs or stolen bases why ignore walks? A stolen base or run scored on a walk is just as good as one scored on a hit. What is even worse, if you are in a league that caps at-bats, don’t you want your guys to get lots of walks? Lots of walks means lots of runs, hits etc. without accumulating at-bats.
The use of at-bats also changes the result. Lets take a quick example. Nick Swisher has 314 AB, 42 runs and 70 walks. By the “efficiency” stat Swisher has a ratio of 314/42, or 7.47. Including walks, his runs stay the same, but his numerator changes to at-bats + walks, giving us a ratio of 9.14. A lower number is better, according to the description, so Swisher is penalized for walks.
In the case of a player with very few walks, the ratio will not change nearly as much, for obvious reasons. Take Kenji Johjima, for example: he has 291 at-bats, 32 runs and 11 walks. So his ratio using at-bats only is 291/32 or 9.09. Using walks in the numerator, we get 291+11/32 or 9.43.
Simply put, the definition used by the podcaster radically alters the results in the instance where a guy has a lot of walks. Yet if the goal is to measure someone’s efficiency at accumulating a counting stat why ignore, in the case of Swisher, those extra 70 plate appearances? This is especially true where those walks do not count against an at-bat maximum.
Nothing here is earth shattering, but on the other hand, it shows that even simple analysis can reveal that self-anointed experts can get it wrong. Or it may be that a self-described expert is even more likely to get it wrong. Believe me, this guy is no Tom Tango. Taking his advice is akin to taking career advice from this guy.