The Ribbie Reporter and Nachoman discuss stats vs. gut feelings

In the aftermath of Kansas City's Opening Day debacle, when home run prone righty Kyle Farnsworth was left in to face Jim Thome, who kills righties and hits lots of homers but is much less effective against lefties, I started thinking about the great divide between stats vs. gut feelings when you manage. I asked the Nachoman: what did he think? You'll find his answer below, and I'll post a rebuttal soon. We hope to continue this conversation over the next few weeks. --The Ribbie Reporter

The Nachoman holds forth: You'll find me right in the middle. I believe in stats. But I also believe they're most useful in the GM office, not the dugout. Tony La Russa famously has data for every batter vs. every pitcher that he uses for in-game managerial decisions. I would suggest that small sample-size data like batter-v-pitcher matchups are not useful. Let's say David Freese is 1-15 against the opposing reliever, while Brian Barden is 4-11. It sure seems like TLR should pinch hit with Barden, right? But that discounts other issues. Is one of them hot? Is one of them better in a pinch hitting role? In the clutch? What about the lefty-righty split? Who's more useful later in the game defensively? What if the game goes extra innings, which hitter do we want to burn now?

Sure, one can quantify each of the above issues I raised. Problem is, much of that small sample-size data might be contradictory. (That is, maybe Barden overall hits worse against righthanders, but he's 4-11 against this guy in particular.) And how can a manager process all of that data to make a stat-based decision within a few seconds during the game? Weighing all of these options on the spur of the moment and making an educated decision is what you refer to as "managing by feel."

I get upset with managers who assign more significance to statistics than can be properly inferred. For example, let's say your right-handed middle reliever has sailed through 1.2 innings giving up only one bloop hit and a stolen base. Now there are two outs in the eighth inning with a man on second in a tight game. Here comes a left-handed hitter who hits better against right handers; the current reliever gives up a higher average to lefties. Bring in the LOOGY, right? That's what virtually every manager today would do. I wouldn't. Not unless the platoon advantage is enormous, not unless the upcoming hitter is an elite run producer, and not unless I have absolute faith in my LOOGY.

I'd rather have the current pitcher who's in a groove than the cold, nervous guy from the bullpen. How many times have you seen this exact situation, where the LOOGY walks the lefthanded hitter? Or goes 3-1 and gives up a hard hit ball? You see, usually a guy is a middle reliever for a reason, that reason being he's not good enough to start or close. If the LOOGY were really that much of a better pitcher, he wouldn't be a LOOGY. Use the guy who's already proven he can get the opponent's hitters out.

In the case above, we can certainly quantify lefty-righty splits. Can we quantify how many times a reliever has fallen behind the first hitter he faces? Probably, but that's not generally an easy to find stat. Can we quantify the effect on the defense, who is forced to pick dandelions for five minutes while the manager goes to the mound twice, the LOOGY walks in, warms up, and then throws one pitch per minute?

I would suggest that in the heat of battle, for any given pitcher-batter confrontation, there are too many variables involved to make a truly stat-based managerial decision. (Sure enough, Baseball Prospectus has suggested that the best managers are those who do nothing strategically -- they play their studs, they DON'T bunt or hit-and-run.) Managing by feel is, I think, the only possible way to manage. The difference between Grady Little and Terry Francona is primarily the talent surrounding them, and partly the better feel for the game possessed by Mr. Francona.

Now, this all changes if we're talking GENERAL managing. Statistics over the course of a season or a carreer can meaningfully inform personnel decisions. But that's a different issue.