Batter versus pitcher data – other than the legality of DFS, there’s no greater divisive element in the game. While I have an opinion on the topic which I’ll share in a bit, first I need to vent a bit as both sides of the issue are putting me on tilt.

There’s obviously a dichotomy among DFS analysts and players with respect to the utility of batter versus pitcher (BvP) data and DFS lineup construction. My main issue is neither side is doing itself any favors when defending their philosophy.

Before I go on, not everyone is guilty of what I am about to describe. But the faction is large enough on both sides that the overall impact is detrimental to what we should be trying to accomplish. The goal should not to be right about BvP but to strive to provide the best analysis possible. Instead of trying to demean the other side, both sides should be pooling their efforts and not engaging in what amounts to a pissing contest. Again, there’s a significant, perhaps even majority number on both sides that indeed get it. But as usual, it’s the vocal minority that gets the attention, and thus my ire.

Let’s start with those against using BvP in DFS analysis. The defense for their position is almost always citing a chapter in The Book: Playing the Percentages in Baseball, written by three well-respected sabermetricians, Tom M. Tango, Mitchel G. Lichtman and Andrew E. Dolphin. The book was published in 2007 and the study in question is even older. The research looked at BvP numbers from 1999-2001 and compared those to the 2002 outcomes between the same batter and pitchers. There has been other research conducted which is accessible on the Internet, but The Book is the preeminent source referenced.

Here’s the problem. Not everyone supporting their argument has read the chapter. For if they had, they wouldn’t be making a monumental mistake that is doing their case far more harm than good. The conclusion of this study, as well as other related, is the sample size of the data is too small to be predictive. In far too many instances, when asked if one uses BvP data, the answer is “no, because there’s no such thing”, or “It’s wrong” or “it doesn’t exist.”

THAT’S NOT WHAT THE RESEARCH CONCLUDES!

The data in non-predictive. Maybe there are certain hitters that see certain pitchers better and indeed perform consistently above their baseline level. All the research says is if there are two such players, there’s no concrete means to be 100 percent sure which is real and which is happenstance. By misstating the conclusion, those arguing against using BvP lose credibility since it’s apparent they either did not actually read the study or they’re misinterpreting the results. Not to mention, the flippant, sometimes elitist means they answer the question perpetuates the geek in the basement persona. If you’re going to cite research, make sure you have read and understand the work.

My beef with those that use BvP to assist with their lineups isn’t the fact they use it, but are dismissive to the notion the question can be addressed via numbers and instead cite intuitive, anecdotal and circumstantial reasoning. To be 100 percent truthful, what really pisses me off is some of the pro-BvP populace sink to straw man arguments, often revolving around the geek in the basement wisecrack. Prove your case, don’t be the playground bully.

Examples of the pro-BvP assertions are those that have played or worked in the game say it’s real. Some relate their own experiences from High School, Legion or even College ball. OK, fine. But here’s a question. If it DOES exist, why don’t the numbers bear it out? And if the answer is not everyone’s BvP is truly because they fare better against that pitcher, how do you know which is real and which is fluke?

The other argument that gets my goat is arbitrarily stating a sample size at which data is useful. This sort of thing can be statistically corroborated. Citing a number to suit your purpose is as egregious an offense as referencing research you haven’t read or understand. In both cases, you’re just making something up.

As an aside, for as much fun as those for using BvP make against those using numbers to refute the notion, if a study existed that demonstrated BvP were predictive, you can be damn sure those that use BvP would be climbing up mountaintops sourcing the work. And if you don’t believe that, perhaps it’s time to check out Craig’s List for cheap bridges? See, now I’m resorting to a straw man argument too.

I’ve said my peace. I’d be lying if I said I felt better but that’s my problem.

Personally, I don’t use BvP in my analysis. I’ve read the study and while I have some problems with the methodology, not to mention how long ago the research was conducted, I’ve seen ample follow-up work to confidently exclude it from my process. To reiterate, I don’t want to be right about BvP. I want to be using the most salient methods when I write and play.

That said, I truly respect those that have played and worked in the game. Plus, you couldn’t tell by looking at me now, but I played and agree there were some pitchers for which I stepped into the box with more confidence. More often than not this was because I was better able to pick up the ball from their delivery. I swear, it was just coincidence the same pitchers didn’t throw very hard or had a crappy curve ball.

I also respect the fact that the President of SABR, Vince Gennaro, has chosen BvP as his primary research endeavor. I come from an academic background so here I’m interjecting some intuition of my own. If Mr. Gennaro is looking for a proxy for BvP data, he too must feel what’s currently accessible in non-predictive.

From a practical perspective, almost all of the BvP examples that I’ve seen deemed actionable are also supported by an analytic method I in fact utilize. Most often this is handedness but there are others. Occasionally, there’s an exception such as the success Paul Goldschmidt enjoys off Tim Lincecum. On these occasions, I’ll usually fade the hitter as a means to be contrarian, not so much because I don’t believe Goldie truly doesn’t own The Freak.

We’re now at the point of this discussion where I was planning on comparing using BvP to flipping a coin. It’s non-predictive, it’s not wrong. Using it neither helps nor hurts. You’re going to be right around half the time. The other half the time it doesn’t matter.

Then I got a little scared. Maybe I’m completely missing something. If there’s a 50 percent chance of being right and a 50 percent chance of it not mattering, why not just use BvP on the chance you’re in the good half?

After taking a deep breathe, I had an epiphany. Bear with me, I’m about to get a little basement-geeky.

Let’s say normal performance is 5. Now let’s say there’s a subset of players that fare well against a certain pitcher and their performance is 7. The research suggests future performance of both subsets against their respective pitchers will be collectively 5. Something’s not right.

There are two sets of players: those with tangible success against a specific and those whose success is just dumb luck. If the former is 7 and the latter is 5, any weighted average would have to be greater than 5. But, the research says combined result is 5.

How can that be? There needs to be a third subset: those hitters who hit WORSE against that pitcher and this was real!

To make the math work out, this subset needs to balance out those whose success was real. Let’s not quibble over exact numbers, but yeah, there needs to be a group of hitters where using BvP is detrimental and there’s about an equal chance you land on the right side of this by employing BvP as there is you land on the wrong side. Of course, there’s still the group which is akin to flipping a coin.

To answer the obvious question, who’s to say there aren’t some pitchers that are aware of their history with specific hitters and make a point to do something different? Maybe they study film to find the hole in the hitter’s swing or a pitch they can’t handle. There’s just as much of a chance of this narrative being the truth as there is a hitter seeing the ball better off a certain pitcher.

So where does this leave us? Other than my personal frustration over how this issue is discussed, I’m a bit flummoxed. I’m still not going to incorporate BvP data into my analysis. I prefer to utilize methods which I can defend. I’m able to mathematically demonstrate players perform better at home. I can show that as a group, left handed hitters fare better against right-handed pitchers than they do versus southpaws. For those feel this is a naive approach, show me. Prove it.

I just have a feeling that neither side really knows the answer. The study to convince me either way has not been conducted. Let’s just say of the many lessons my Mom taught me, if you want something done right, do it yourself is near the top of the list. Who would’ve guessed that my stubbornness to do the dishes would one day aid in my vocation?