World Famous StatSpeak Roundtable: August 11

StatSpeak is pleased today to welcome Paul DePodesta to the roundtable.  Paul is currently working as a Special Assistant for Baseball Operations (when I asked Paul what his official title was, he said “Ballboy is good enough”) for the San Diego Padres and writing random musings in his blog It Might Be Dangerous… You Go First.  He has previously worked as an assistant GM with the Oakland A’s, GM with the Los Angeles Dodgers, and most importantly, he started out his baseball career with the Cleveland Indians.  Paul shares his thoughts with Eric and Pizza concerning the trading deadline, the future of Sabermetrics, and batter/pitcher matchups.  Random coincidence: Paul also shares a birthday with both Eric and Pizza.  How random is that?
Question #1: Are batter/pitcher matchups useful?
Paul DePodesta: With the amount of roster movement, the emphasis on young players, and Interleague play, a single batter-pitcher matchup rarely accumulates enough events to become really useful.  If a significant number of individual matchups does exist (though that number is probably much higher than we assume), many of them likely occurred years and years ago thereby making them less pertinent.  Because of these reasons, I don’t know that many Managers rely on the data, which is probably the right move.  More than anything, the matchups provide a convenient reason, not necessarily a good reason, for a decision.
Eric Seidman: They are and they aren’t useful, depending how you look at it.  For instance, in watching Paul’s Padres take on the Rockies Saturday night I saw Brian Giles come up to bat against Luis Vizcaino and the following graphic (or something like it surfaced): “Giles vs. Vizcaino: 3-4, 2B, K.”  Sure, something like that offers viewers and announcers some fodder to think over but honestly, I don’t care.  I don’t care if Giles is 3-4 off of Vizcaino. 
This was actually my favorite chapter in The Book: Playing the Percentages in Baseball.  Without giving away all the gory details, they found that specific batter/pitcher matchups had just about no predictive ability moving forward; if Giles was 13-22 in two years against Vizcaino it did not necessarily mean, given the wide array of data, that he was more likely to perform well against Luis the following year.  Going further, I had long been of the belief that perhaps these specialized matchups aren’t important but special subgroupings would be.  For example. instead of Giles/Vizcano, look at Giles vs. everyone who throws very similarly to Vizcaino.  Once again, Tango/MGL/Dolphin showed that the 60-70 PA in this sample still had less predictive ability moving forward than the player’s true talent level, his weighted three previous years.
All told, I do not think batter/pitcher matchups are useful for determining anything significant with regards to the matchup.  Is it interesting to me that Rollins is 12-19 off of Snell with 4 triples?  Yes… but I know in my head it means absolutely nothing with regards to whether or not he will continue to “own” him.
Pizza Cutter: Yes.  Play-by-play announcers have to say something to fill the time.  The knee-jerk reaction of the statistician in me when I hear “Jones is 2-for-4 lifetime against Smith with a double and a strikeout” is to shrug and say “small sample size.”  However, there’s something to be said for the possible utility of batted vs. class of pitcher matchups.  With the advent of Pitch F/X, it’s now very possible to group pitchers together who have common arsenals.  A lefty sinker/slider guy requires a different approach from a right-handed fireballer with a devastating change, and with cluster analysis, it’s just a matter of plugging in the right input to figure out which pitchers might be considered in the same class.  We can be a little more empirical in grouping those pitchers together, and we can say things like “Jones is 32-for-297 against this grouping of pitchers who are all basically lefty junkballers.  Maybe Jones needs a day off against Jamie Moyer.  The downside: It doesn’t sound as cool on play-by-play.  People tend to dose off when I start talking about cluster analysis.

Question #2: Is the trading deadline over-rated?
Paul DePodesta: Despite how much emphasis we put on it, I have to look in the mirror here and say yes.  Whenever there’s a flurry of activity over the winter, we always remind ourselves that pennants aren’t won in December.  The same can certainly be said for July 31st, even though we tend to get caught up in the psychology of the competition.  I do believe that certain moves can have an impact, for instance, being in a tight race and acquiring a dominant player in a position where you’ve been getting replacement level performance or even acquiring a player that will be with your team for multiple seasons.  Though many of the transactions do make some sense for the Clubs involved, as the cost may be minimal, the reality is that the majority of moves won’t materially influence the pennant race. 
Eric Seidman: Yes, the trade deadline is overrated.  For starters, by July 31st, with just two months and 50-55 games left on the season, a player is only going to offer an upgrade of maybe one win over his replacement.  Getting Adam Dunn to replace Chad Tracy/Conor Jackson, or whomever is playing left-field these days in Arizona would be an upgrade, but it wouldn’t be the different between 88 wins and 94 wins or anything like that.  Secondly, whether or not it is true, a perception exists that teams need to scurry to make moves to enhance their chances.  Well, if everyone is scurrying then they don’t have as much leverage and are going to appear more desperate; therefore, they may give up more assets to receive a product not necessarily worth the return they expected. 
A team mortgaging the farm to get a guy like Ron Mahay for under two months of play is a pretty ridiculous notion, but that is the kind of frenzy that this deadline catalyzes.  If you look at all deadline deals, as Jayson Stark did in an article not too long ago, you can see that very, very few have huge impacts.  Overrated refers to something garnering a huge reputation that doesn’t actually live up to it; that essentially sums up the trade deadline.
Pizza Cutter: As a professor, I always appreciated the students who did things before the deadline.  In fact, I think that the smartest trade of the season was the way in which the Brewers pulled off the C.C. Sabathia deal.  Before anyone else could do it, the Brewers ponied up and got their man and got an extra month’s worth of use out of him.  Plus… they got C.C. Sabathia.  Is the trading deadline over-rated?  I understand that some teams are waiting until the last moment to see whether they have a good enough chance to take a shot, but the flurry of activity on the last day always makes me wonder.  Were your chances really that much better or worse on the 28th?  I have to imagine that the entire week of the deadline, or perhaps the entire month of July is filled with discussions between teams and some ideas being passed back and forth.  The thing that worries me is that those discussions should, in theory, produce a steady stream of trades throughout the month.  It seems that the majority of the moves happen right before the deadline.  As I’ve pointed out before, people make bad decisions on a deadline, especially an imminent deadline, because they panic and feel the need to “do something.”  That makes me wonder how many trades are made/agreed to out of that sense of panic rather than a rational cost/benefit analysis.
Question #3: Pitch F/X has been heralded as the next great frontier for baseball analysis, and indeed is already producing some interesting results.  Looking ahead, what do you think will be the next great frontier after that in baseball analyses?
Paul DePodesta: I can’t tell you that!  In all seriousness, I don’t believe there’s even one area that we’ve really mastered yet.  We, as an industry, continue to get better every year in our understanding of the game, but we’ll never “figure it out” so to speak.  Even though a large part of my job entails getting our arms around the uncertainty, we’re well aware that it’s a never ending journey.  That’s part of the fun – living with the uncertainty.  When I was in Oakland, a fan once said to me, “The reason I love baseball is because it’s a drama without a script.  It’s the original reality TV.”  I couldn’t agree with him more.   
Eric Seidman: Hmm.  Well, if the Cardinals would stop being so selfish and let MGL give us his UZR we would have an excellent defensive metric available and that would be covered… pitch f/x and hit f/x are already in the works and allowing us to study a whole new aspect of the sport.  Ultimately, I think the next great wave will be some way to quantify the mental aspect… the aspect previously/currently thought of as impossible to quantify.  For instance, what if we could really measure the intention of a pitch.  In the pitch f/x data, we don’t know how accurate a pitcher is in hitting spots because we don’t know where the catcher set up.  But what if we did?  Though it isn’t truly feasible for now, especially after speaking to Cory Schwartz of MLB Advanced Media, this would allow us to study which pitchers were better at hitting their spots and truly deducing their approaches.  Do “missed spots” matter as much as we are led to believe?  Are umpire really more likely to call a strike a ball if it’s over the plate but the catcher set up really inside and had to reach?
On top of that, after speaking with DBacks rookie Max Scherzer and doing some personal studies for him, something he is really interested in is quantifying what he knows, in his head, are his good or bad pitches.  We discussed some method wherein he would keep a mental tab or something between innings with regards to which pitches he felt really good on in terms of release point, how it moved, etc, and then we could look in the data and search for correlations or meaning.  This is an area thought of strictly as mental that could be quantified.
Pizza Cutter: Oddly enough, I think that the next great frontier in Sabermetrics is… scouting?  Scouts have often loudly complained that Sabermetric folk don’t respect the fact that they’ve been around the game for years and years and years and can often pick a guy out of a crowd that’s major league material.  Sabermetricians, on the other hand have often loudly complained that scouts, on account of being humans, are going to over-value some information and under-value other information when watching a player.  The thing is that there are likely things that scouts really are very good at measuring with their eyes.  Sabermetrics comes into the equations in two places.  One is figuring out, empirically, what the scouts are good at observing and what they are not.  When a scout says “excellent bat speed,” does that usually bear out?  When a scout says that a player will probably end up as a regular 3B somewhere, does that usually come true?  Are some scouts better than others at predicting the future?  There has to be a drawer full of old scouting reports out there somewhere.  Open it up and analyze them!  Find the ones from five years ago and see how those guys are doing now.
The other piece, and I have to assume that this one is going on at least in some places, is that Sabermetricians can tip scouts off on what skills to actually look for, perhaps creating a new breed of Saber-savvy scouts.  A good Sabermetric look at the actual scouting report itself may even be in order.  In my field of psychology, we have standardized assessment forms that have been validated as being able to pick out people suffering with depression or anxiety or attention problems.  Could we not do the same for diagnosing who would be a good baseball player?


7 Responses to World Famous StatSpeak Roundtable: August 11

  1. Wow, nice catch with DePo. Great stuff, guys.

  2. dan says:

    I could be a couple days off, but I think the 16th is the day the Mayan long-count calendar ends (history channel documentary was a couple weeks ago, and my memory’s a little fuzzy at 2:20 in the morning)

  3. Dan, me, Pizza, and Paul all planned that. We asked our parents to have us on the Mayan long-count calendar year end.

  4. Lisa Gray says:

    PC –
    actually, that is an interesting idea about evaluating scouts. of course, problem is that once the ballplayer is in the Organization, no telling WHAT they will do with the guy. they make catchers into shortstops, shortstops into catchers, starters into relievers, hitters into pitchers – and many times, for apparently no good reason.
    which, of course, isn’t exactly the scout’s fault.
    so if you take a guy who the scout says pitches like pedro and the Organization decides he’s too short to be a pitcher and turns him into a lousy hitting SS who can’t hit his way out of A ball, doesn’t mean the scout was WRONG…

  5. Colin Wyers says:

    Lisa –
    As with everything, it comes down to sample size. You don’t judge a scout on one player he’s evaluated, but on hundreds or thousands of players he’s evaluated.
    The most exciting parts of sabermetrics these days (Pitch F/X and PBP defensive analysis) both really blur the line between “statistics” and aggregated scouting reports. The problem with scouting data (at least, for those of us on the outside) is that it’s difficult to obtain and not collated well into a database.

  6. Colin, yeah exactly. I just finished the book “Fantasyland” and the writer had a statistical analyst and a scout to aid in his fantasy decisions, and the scout had developed a database filled with information regarding the players that the analyst (Sig Mejdal, now of the Cardinals) could use to quantify results. Things like, all the players in the last 20 years who got married in the off-season, did it translate to anything in the stats.
    This was a pretty interesting thing to read because they were using detailed scouting to rate players and then compare that to the stats. A guy like Jacque Jones rated about average with statistics (this is 2005 or so) but much higher on the scouting side.

  7. Peter Bendix says:

    I think you hit the nail on the head, Pizza Cutter: quantifying scouting, and developing scouting-saber hybrids, seems like it could be a huge competitive advantage.
    Most sabermetricians understand the important of scouting, but are simply unable to do it themselves. If you take a sabermetrician and teach him/her scouting, you have one person who fully understands both aspects, and can give a much more thorough opinion of players. This is one of the reasons I respect Keith Law so much.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: