Monday, August 31, 2009

Mike Rizzo and Sabermetrics

There were a lot of questions directed at Mike Rizzo in the days around his hiring about the role of statistical analysis in his new job. During his introductory news conference, he said:
It's another piece of the puzzle. I always err toward my own two eyes rather than a computation, but as we've seen with our great young staff, the guys that grind it out with me every day 24-7, is that we do utilize all facets of scouting and player information.
And in a Tom Boswell column a few weeks back, we got this:
For the last decade, baseball has had a debate between those baseball lifers with an eye for the game -- like Rizzo's dad and Boras -- and the younger, polished, generally more educated "Moneyball" types.

"I'm a hybrid," says Rizzo, who then does his ode-to-Sabermetrics riff about the value of knowing Value Over Replacement Player and WHIP. "My dad still doesn't buy it. But there's a place for it. Why not use all the tools?"

Then Mike Rizzo puts his hand over his mouth like he's about to tell you a secret. Remember, this is a man who thinks that it's nothing to drive the extra 200 miles, or figure out a 21-year-old's personality, then bet his team's future on it.

"Besides," he whispers, "it's not that tough."
Rizzo's exactly right that scouting and saber are complimentary, and the best saber folks have never argued otherwise. It's almost always people who know little or nothing about either scouting or saber who act like there's some running war between the two.

I wondered, though, about the "it's not that tough" line. Because if he really thinks good statistical analysis is easy, then I'm not sure how much he really does get it.

So on blogger day, I dropped what was admittedly a "gotcha" question, asking him to give us a couple specific examples of how he uses statistical analysis. Nats 320 has the full transcript, but the examples he cited were groundball rates for pitchers and zone ratings for fielders, specifically pointing to Nyjer Morgan as a player they evaluated with fielding stats.

These were fine answers, and they jive with his actual moves. But the red flag that was raised at least for me is that these were really pretty basic examples,
and it reminded me of that "it's not so tough" line.

What I'd like to see is the Nationals front office to get on the forefront of breakthrough statistical analysis, figuring out trends and probabilities that no one else has figured out before in order to gain an upper hand.

It wasn't really all that long ago that Voros McCracken did his groundbreaking research on batted balls, or Mitchel Lichtman's work on fielding, or Rany Jazerlyi's and Tom Verducci's work on pitcher usage.

There are so many remaining areas of inquiry. Fielding metrics remain a work in progress. Reliever usage league-wide is stuck in an idea Tony LaRussa had 20 years ago. Is infield flyball rate a repeatable skill? Game theory and bluffing remains a much underexamined part of the game (e.g., how many bunts must a player drop down before the other team brings in the corner infielders, and how many additional hits are created by the repositioned fielders?). Why do some pitchers seem to consistently exceed (John Lannan) or fall short of (Javy Vazquez) their fielding-independent pitching projections. Pitch F/X alone could provide several books worth of insight and analysis.

The main point of Moneyball wasn't so much about how Billy Beane used statheads to gain an advantage. It's that he found undervalued commodities in the market to build a better team. As the game evolves, there will always be new and different undervalued commodities. The challenge is to stay ahead of the opposition. That's tough to do in every facet of the game, and the saber side is no exception.

Here's hoping that Rizzo can go beyond merely implementing established saber principles and bring in people able to do original, proprietary analysis to give the Nationals a true leg-up.


JayB said...

The way I read it was that eye ball scouting was not that tough for his dad.

Hendo said...

Actually, the sustainability of groundball rate for pitchers is a fairly cutting-edge topic and a fairly controversial one. There are and have been murmurs that some pitchers with low strikeout rates enjoy relative success (as John Lannan does now, which you point out, and Chien-Ming Wang did back in the day) because of their ability to induce ground balls as opposed to line drives. This pretty sharply contradicts McCracken's thesis, which has stood up relatively well to challenges over the last decade.

I've been a skeptic in the matter, but when you look at Lannan vs. Hanrahan over the first half of the season, it's tough not to be tempted to attribute a skill (or lack thereof) that may or may not exist. I suspect the analysis and discussion will continue -- and, since the Nats are avowedly a pitch-to-contact team, they'll be responsible for no small part of the body of accumulated evidence.

James Bjork said...

At this point, I'd settle for some serious COMPETENCE in the GM seat. I think there's ample room to build a contender with astute consideration of the metrics already recently developed, provided the Lerners pony up when needed.

I don't know that Rizzo et al coming up with some new, secret formula to gain an edge, like the "Ultimate-donut-zone-rating-of-
pitching-under-lights" metric. It would be great if the Nats' new brain trust excelled at some kind of new and superior pattern recognition, but I'm not holding my breath.

Avoiding overpaying for vets in decline is easy step number one that a lot of GMs still make. At least Rizzo's past that one.

Kevin said...

I thought this was the most telling part of the exchange:

"So, we bring the statistical analysis into it, although I trust more what I see than what I read. But, it is always nice when what I read corresponds with what I see."

flippin said...

"...bring in people able to do original, proprietary analysis to give the Nationals a true leg-up."

If Rizzo was looking to hire a cutting edge analyst(s), why would he tell a bunch of bloggers? Or, why would he share his methods and practices? Unlike illicit interrogation techniques, I am perfectly happy with secrets. I don't need to know how, I just need results.

Steven said...

He wouldn't tell us his analysis, but he could tell us that he'd hired Bill James like the Red Sox did or Tom Tango like the Mariners did. The Indians have Keith Woolner, the Rays hired Rays James Click, the Pirates hired Dan Fox, and Josh Kalk got hired by someone in the AL.

These guys aren't just sitting around looking at Fangraphs. They're creating new and revealing metrics that no one else has, and it's not necessarily the case that they are coming up with stuff of diminishing value.

James Bjork said...

Maybe I'm too pessimistic, but I wonder if the variance in player outcomes due to sheer randomness (noise) or dumb luck (injury) isn't going to blow away that of any other new metrics.

The field has already put out to pasture (except for in the minds of old-timey columnists) the role of raw ERA, or plain old batting average, in place of metrics that seem far more likely to be prognostic of success irrespective of luck.

I wonder if sabermetrics will hit an asymptote caused by the vagaries of chance, especially with pitchers. That said, if Rizzo hires a stat-head, it probably can't hurt- especially if Rizzo applies his watchful scouting eye.

miguel said...

i will tell you if yogi berra was to be draft in 2010 no one will take him the same duck medwick and almost all those old timers in the halll fame, that were small and with maybe two tools players and the mechanics for pichers they will never make it in todays world, but yet some one a scout saw talent,yes you can use all the computer you want,but nothing replace a good scouting eye period