Duquette on Statistical Analysis & Decision Making
By Mike Silva ~ August 14th, 2009. Filed under: Mike Silva.
Readers of the site know about some of the spirited debate going on here regarding statistical analysis. I questioned our good friends at Amazin Avenue, who tend to look at the game from a very numbers centric perspective. The best example is the Jeff Francoeur/Ryan Church trade, which continuously is analyzed by comparing numbers such as WAR, OPS, and UZR. I like Frenchy because of the power and run production he has supplied. We didn’t see this from Church for well over a calendar year. Of course, when you compare OPS it’s very similar because of the amount of walks Church has generated. A great deal of his production has come in just two games where he had five hits, a homer, and four ribbies. Of course you wouldn’t know that unless you looked at his game log/or watched him play on a consistent basis. Francoeur simply has been a better fit for this Mets team than Ryan Church.
After coming under fire for my assertions that statistical analysis can’t be a major part of the decision making process, I decided to reach out to someone that could give me a front office perspective. Who better than former Mets GM Jim Duquette, who recently appeared on NYBD Radio, to give us a firsthand look at statistical analysis in professional baseball.
When I asked Duquette his thoughts on a front office executive using numbers like WAR, UZR, etc., and how big of a role it players in personnel decisions, his response was as follows:
I like the stats (WAR and UZR) are the most reliable in my opinion.. I also like RC27 as well.. but if you solely build a club on this type of basis.. you’re doomed to fail.. you can’t take the human element out of the game.. no matter what the stat guys think.. in fact, the stats play a no more than a third of the total eval of a player..maybe even less, for me..
We have discussed WAR and UZR, but not RC27. The “RC” stands for “Runs Created” which is a statistical attempt to measure a players overall offensive contribution to his team. It involves taking players OBP and multiplying it by Total Bases. The premise is that you need to get on base and drive runners in to actually score. Players that do both at the highest level have the most value. RC27 takes that concept and adds another element: measure how many runs a lineup of nine of the same individual would score in a game. I have seen where this number has also been adjusted for ballpark as well. Again, it is trying to determine a player’s value versus the “replacement level”.
Although Duquette and I didn’t have an extensive conversation, I believe he shed light on the topic. When he stated that, for him, he uses this type of analysis for a “third” of player evaluation because, if you “take the human element out of the game”, you are “doomed to fail”, my original premise, to an extent, was validated.
I am surprised that even a third of a player analysis would be considered. I still believe “replacement level” can take on many different meanings. I have no problem looking at a players “Runs Created” (RC), but not quite sure you learn any more from RC27. Actually, I might give it a bit more credence, than WAR, because you don’t have the ambiguous defensive metrics involved.
Again, and this goes back to my original claim, I have no problem with any statistical measurement. As Jim Duquette pointed out, this stuff is being looked at in Major League Baseball. Of course, some take it more serious than others. If I ran a professional ball club I would rely more on scouting reports and the opinions of long time “baseball people” versus the “number crunchers”. Anytime we cite Bill James I still think back to his “anyone can be a closer” statement. That worked so well for the Red Sox that they needed to get Keith Foulke a year later, and, well, we know what happened. With all that said, there continues to be many that tell me there is value in this analysis. A friend of NYBD, who is also an engineer, took a look at some of the defensive metrics cited the other day. He agreed that there are issues with each analysis, but felt that it was solid enough to have merit and be taken seriously.
The best type of analysis uses statistics for support and, based on Duquette’s response, that is what he did during his days in the front office. If I were running a ballclub, that is how I would approach the situation.
August 14th, 2009 at 8:29 am
Jim Duquette used his extensive player evaluation credentials to trade Scott Kazmir for Victor Zambrano. That is all.
August 14th, 2009 at 10:04 am
The idea was never that “anybody could be a closer”, it’s that “any good pitcher can be a closer”, and it’s been demonstrated time and again . . . and the 2003 Boston Red Sox? The closer by commitee idea killed them all the way to 95 wins and a better record in 1-run games than the 2004 Red Sox.
August 14th, 2009 at 10:43 am
Why did you ask Duquette, Mike? Was Steve Phillips not available for comment?
“I am surprised that even a third of a player analysis would be considered.”
Why do you keep implying that the uses of sabermetrics and scouting are mutually exclusive?
August 14th, 2009 at 11:33 am
Couldn’t agree more JE. Did anyone on here say that there should be mass firings across major league scouting departments, replacing old time scouts with spreadsheets??
Mike, I think the general argument (which you seem to be missing the point on) is that an aversion to “new-fangled” statistical analysis shows a lack of comprehension of the basic concepts. The examples you have cited as to why SABR type analysis doesn’t work have been piss poor at best (see comment from Linus above). The point of these new statistics is not to put someone out of a job, but to find as many ways as possible to evaluate a player’s performance. To ignore these things is just plain closed minded.
Also, I probably wouldn’t use Jim Duquette as my primary source for a post about how front offices should be run.
August 14th, 2009 at 11:53 am
http://www.baseballthinkfactory.org/files/newsstand/discussion/nybd_silva_duquette_on_statistical_analysis_decision_making/
Read the comment at #10 by Ron Johnson. Once again, you’re criticizing something that you do not understand, and once again you sound silly for it.
‘Anytime we cite Bill James I still think back to his “anyone can be a closer” statement’
Can you produce evidence that James said that? Because he didn’t, and your reporting that he did is unfair and ridiculous.
August 14th, 2009 at 12:02 pm
I clicked on the comments to make the following points:
1) Duquette gave Kazmir away for Zambrano (among many other egregious moves); and
2) I’m assuming Steve Phillips was unavailable.
It looks like I was beaten to the punch.
Up next, I’m assuming we’re going to get a post about investing from Bernie Madoff, a post about journalistic ethics from Jason Blair or a post about the art of slugging from Rey Ordonez.
Honestly, how do you use Jim Duquette (in New York no less) as your source for how to run a front office and not expect to be ripped a new one?
August 14th, 2009 at 2:45 pm
Why would you use Duquette as your guy? He is an idiot.
http://mvn.com/aroundthemajors/2009/08/revisiting-scott-kazmir-for-victor-zambrano.html
He clearly did not use stats and focuses too much on the human element in the ridiculous Scott Kazmir trade. That is like going to Kramer for advice on race relations.
August 14th, 2009 at 6:47 pm
Hey Mike, did you ever read an of the Baseball Abstracts?
August 14th, 2009 at 9:49 pm
You are, quite simply, an idiot. Replacement level is replacement level. It means exactly that. What’s the average scrub going to produce. It’s not complicating. We’re not saying you have a megaprospect. We’re saying it is your average AAA player coming up to the MLB. That’s WAR. That’s it. It’s easy.
August 15th, 2009 at 2:25 am
It’s ironic that you say statistics should be used for support, because that’s the one way they should not be used. As the famous saying goes, that’s akin to the way drunks use lamp posts. When statistics are used properly – not in support of a conclusion already reached – they provide illumination.
August 15th, 2009 at 7:31 am
Greg
I actually used that quote earlier in the week. I believe that decision making can be complemented by stats, that is what I meant regarding support. I used the Alex Cora/ Anderson Hernandez example when discussing who the Mets should use as a back SS next year. Do the stats illuminate or provide support (depending on my method of decision making). Can I use WAR as a secondary support measure? I think all that is fair game.
August 16th, 2009 at 5:05 pm
Great guy to ask…a Former GM who is out of a job due to his complete inability to build a quality team or farm system.