THCB

Optimal Positioning Strategy and the “Quantified Relationship” in Baseball and Health Care

Screen Shot 2014-05-30 at 6.35.12 PM

Strategy in baseball used to be a fairly straightforward matter. A few strategy rules – a right handed pitcher was more successful against a right handed batter, lefty against lefty, no left handers at infield positions except for first base, don’t hold the runner at first with two outs and a left handed batter, and sacrifice bunt to move a runner at first with less than two outs- were taken as gospel and practiced by the community.

It was baseball’s version of the 10 commandments, written in stone and for the first century of baseball, unchangeable.

The world changed, though few knew it, in 1946 when Cleveland manager Lou Boudreau moved his shortstop to the right of second base against the legendary dead pull hitter Ted Williams of the Boston Red Sox.

However, like many innovations, it is only with the advent of large data sets that the revolution that started that July day in Cleveland impacted day to day strategy.

A players position on the field is no longer the result of the manager’s intuitive hunch, or even the result of consulting a written document of the past several encounters between a particular pitcher and a particular batter- a scatter gram of where this batter is likely to hit the ball. Instead, major league teams are increasingly relying on sophisticated, large data sets that are housed on remote servers.

These data sets run complex algorithms predicting the best solution for a particular ecosystem- elements of which include – batter, pitcher and all the defensive players and their particular gifts, skills and tendencies- and even the weather and time of day.

Despite what our heart may tell us Casey Stengel is no match when compared to the wisdom of regression analysis and multi-factorial modeling when applied to predicting outcomes.

While the revolution may not be televised, it certainly is being written about- note the two pieces, both appearing on March 28, 2014 one,’ Deee-fense: Baseball’s Big Shift’ in the Wall Street Journal and the other in the Washington Post.

Both articles describe how the game has dramatically moved toward predictive modeling to assist in defensive placement of players. One reflection of this change is the rapid emergence of a new coaching position titles such as “defensive coordinator” and “Director of Systems Development.”

The Director of Systems development is responsible for the creation and maintenance of the database(s) which the defensive coordinators can then tap into in order to achieve what is referred to as ‘optimal positioning’ with the goal being to maximize the return when ‘defensive runs saved’ are calculated.

Optimal positioning is putting every defensive player in the most likely position for them to be successful based on the pitcher, the batter, the defensive players skill set, and the game situation. This is no longer just shifting the shortstop or second baseman based on whether it is a right or left handed batter!

This data driven perspective, in addition to generating almost a 350% increase in defensive shifts from 2012 to 2013, has also led to a burgeoning consulting opportunities, with Baseball Info Solutions, founded by the prophet of the field, Bill James, being the first of many.

So what does this have to do with health care?

Everything!

Both are entering the big data era.

Let’s just play a bit with a few of the comparisons.

Screen Shot 2014-05-30 at 6.46.04 PM

So the Director of (Systems Development/Health Care Analytics) creates, maintains and updates the analytic engines/ algorithms that the (Defensive/Case Management) Coordinators use to position the players/providers in order to obtain optimal positioning/right care, right time, right place. And of course the consultants continually update the data bases to continue, enrich, and deepen the learning cycle.

I want to suggest that baseball has a more detailed and therefore more powerful database than health care. Health care informatics would be doing well to catch up baseball informatics.

In order for health care to catch up with baseball and move closer to optimal positioning, I want to close by point out the need to have the type of data that baseball gathers with every pitch- what type of pitch- fastball, curve, slider, splitter? How fast was the pitch? What was it’s location- high, low, inside, outside? And what did the batter do- swing, miss, foul it off, ground ball, fly ball. What direction was it hit? How hard was it hit? Was it caught?

So much data!

Health care’s next data frontier is to find and quantify the healthcare equivalent that duplicates the data generated in each unique batter-pitcher relationship.

I want to suggest that the healthcare equivalent is the doctor-patient relationship.

We need to find ways to quantify those aspects of the relationship that can be easily measured, quantified and most importantly are strongly correlated with good outcomes.

The “quantified self”, in healthcare, will become the “quantified relationship”.

In the Major Leagues they say, “Its baseball, you have to play the percentages.”

In health care its time we said the same thing.

Lets play ball!

Charlie Gross, PhD is a director of clinical programs and behavioral health at WellPoint.

Livongo’s Post Ad Banner 728*90

2
Leave a Reply

1 Comment threads
1 Thread replies
0 Followers
 
Most reacted comment
Hottest comment thread
2 Comment authors
Charlie GrossBubba For President Recent comment authors
newest oldest most voted
Bubba For President
Guest
Bubba For President

Holy smokes! This is actually a very insightful post. Suggests that we need new metrics that more accurately represent performance ..

How do we come up with smarter metrics that let us calculate who the best offensive and best defensive players are?

What stats are we currently overvaluing (i.e. home runs, batting average, stolen bases) and what are we undervaluing (bases on balls, game winning plays,) – what do our current definitions imply?

How do we calculate fielding errors? Good defenders with better range may make more “errors” than average defenders who don’t try to get to the ball.

Charlie Gross
Guest
Charlie Gross

Bubba
Great name, thoughtful comments
I particularly was struck by the last one – “good defenders with better range may make more errors”

In healthcare and in innovation more generally, the same may hold true.
The greater our reach, the more potential for “errors” or “failure”

How does one account for that when evaluating ones efforts?

Or more generally, how does one incorporate that sort of “error/failure” into a learning loop, that iteratively leads to better outcomes?