Sunday, January 12, 2014

La Russa Agile Innovation #10 of 17: Once You've Found an Optimal Tactic,
 Treat it as Sub-Optimal  

Adhering to a process you've proven works effectively and efficiently in the context of a team is optimal. But you need to vary the use of Optimal methods gently and continually over time.

No line of work is more effective and illustrative Agile and Lean norm than Baseball. One great practitioner of the mix of art and science of knowing when to simply apply the Optimal move and when to riff off it is now Hall of Fame manager Tony La Russa, who desribed it in his recent. He documents it in a way any manager can "get" it in his recent book, One Last Strike (p. 231-233).

In the book, he discusses "intentional walk", a managerial technique he uses far less than the norm. Statistically, the walking a batter to get to the next batter gives the other team in the composite average an advantage. That is, that if you roll up all the cases in which a manager chooses to apply it, the choice nets out as negative over all cases. Baseball Prospectus last year published this info for the 2012 season rolling up every event for the 2012 major league season. It reports what difference on average events correlated to runs the batting team scored in the inning. For example, "HR", a home run, was worth, in the overall mean average case, +1.4 runs for the batting team in that inning.





















As you can see, the IBB, the intentional walk, nets out negative for the team of the pitcher that issued it, on average about .17, or about one-sixth, of a run per instance. In general, any sensible manager would lean against this net-negative tactic. La Russa is a metrics-informed manager, and also leans against it.

But here's the Truth -- he doesn't automatically NOT use this sub-optimal choice. In the beginning and middle of his career, he was usually at or near the bottom of use in the IBB. In his final year of managing, 2011, he still used it 44 times, third lowest for the league.

NOTE: If you haven't thought through the consequences of the differences between the National and the American leagues, managers in the National use the intentional walk far more often than their American peers. This is because the pitcher is part of the batting line-up in the National but has a designated hitter in the American. Pitchers, as whole, bat worse than players not named Robert Andino who play any other position, so the intentional walk of the batter hitting in front of the pitcher to make the pitcher either bat or be dislodged with a pinch-hitter, is the most frequent application of the tactic.

And while La Russa's frequency of intentionl walks is very low for the National League, American League use is so much lower, his ranking would be more moderate.

There's the Lean/Agile exemplar: that almost any practical technique/tactic/process could be "Optimal" in any given context...of time plus team plus cost plus organization plus decision environment plus team-members-on-the-given-day. Optimal is not static and because it isn't static, push-button management using the discovered optimum, even though it fails less often that other choices will in composite, may underperform many other possibilities in the specific context.

La Russa's is a classic innovator. His ability to reject the intentional walk, while still being able to use in in contexts he finds beneficial for it is an elegant example of Agile/Lean management practice.

La Russa is an advocate for a variant of the IBB that he can use in lieu of it -- in certain contexts: The Unintentional Intentional Walk.

Here's his book on the replacement technique:

One of the strategies we employed a lot was the "unintentional" intentional walk, and it was particularly effective in the National League, where the pitcher hits. Often the eighth-place hitter comes up with two outs, a runner in scoring position and the pitcher on deck. If you walk the eighth-place hitter and the pitcher makes an out, then the leadoff man comes up to start the next inning (NOTE: and this is an incremental advantage to La Russa's opposition). We'd rather try to use an eighth-hitter's aggressiveness against him, hoping he'll swing at a borderlibne pitch and get himself out.

We had a sign for it. {SNIP} You're telling the pitcher he's going to pitch at the edge of the strike zone or off the edge. (NOTE: which are harder to hit well){SNIP} What you have going for you is that the hitter, since there's an RBI situation, wants to get that bat going. And he may be more apt to chase a pitch out of the strike zone. And sometimes you might get a call from the umprire on one of those "edge" pitches, which changes the count (NOTE: to a count more favorable to La Russa's team).

It's a classic innovation a la H.G. Barnett. Take a standard procedure and see what you can take into account from the context to see if there's a change that gives you a small edge either in the general case or for the specific context you're about to face.

There are side benefits, even when it doesn't work. The team gets used to applying it so when it comes up again, they will have more experience with the tactic. The opposition number-eight batter also knows you may use this subtle technique, and may use brain cycles thinking about it even when you're not. That divides her attention, and that's a benefit to you, too.

It's not without risk of failure (and to be realistic, almost nothing is, unless you're Goldman Sachs and you carry in your pocket or perhaps on your payroll both main political parties' choices for Treasury department- and S.E.C. leadership). The difference between being a good manager and a great one is picking your spots for innovation/riffing off of the Optimal choice.

The majority of managers gravitate towards what I call Binary Thinking (simple dualities such as "good v. bad"). Binary Thinking makes it hard to apply things the way one must to be truly good as a manager of the First Base skill set, that is, stochastically. The optimal model is stochastic, neither random (investing an equal amount in any eventuality no matter how likely or unlikely) or deterministic (invest in the likeliest n options only until there is no more to invest), because evolution is stochastic (not exactly hitting the bull's eye of potentiality every time but scattering arrows clustered variably around the bull's eye, getting sparser with distance from the bull's eye).

For you to have any chance being successful in a system that at least pays lip-service to Lean or Agile, one needs to be willing to riff off all but the most static systems' Optimal choices. Follow La Russa here. Never stop experimenting with stochastically distributed variants off the Optimal.

This page is powered by Blogger. Isn't yours?

free website counter