World Famous StatSpeak Roundtable: October 1

Just to be on the safe side, I’ve sent Dane Cook on a fact-finding mission to Slevetika.† But, welcome to the most wonderful month of the year.† We begin the month, appropriately enough with a roundtable featuring Matthew Carruth, of Lookout Landing, StatCorner, The Hardball Times, and FanGraphs.† Not happy with all that, he’s joins us at the big StatSpeak table to talk about the strike zone, the playoffs, the schedule, and the best. closer. EVER.
Question #1: Ignoring the specifics of how it would be accomplished, would you be in favor of ball/strike calls being made by Pitch F/X instead of an umpire?”
Matthew Carruth: Yes. Unquestionably yes. I love baseball and I love the history of baseball. To me, that history does not extend to umpires and umpiring calls. I think the team that plays the best should get the win and that officials of all kinds, in all sports, should never be noticed. That’s the ideal and any step we can take toward that ideal we should take with gusto. I am pretty confident that pitch F/X can outdo an umpire right now in terms of accuracy and I am positive it can outperform umpires in terms of consistency, and really from a strikezone perspective, that’s what I think is most important.
Brian Cartwright: I donít think so. I do consider myself a traditionalist. Hate the DH. I think it would take too much of the human element out of the game. If the batter doesnít like the call, who is he going to yell at? I know we have the technology already in place at the major league level, but not at any other, professional, college, scholastic or amateur. It would create a wide distinction in how the game is played. Also, the home plate ump is the top spot on the umpiring crew, getting hundreds of calls a game in the twin roles of calling the pitches and also calling plays at the plate. Without calling pitches, this would be reversed, as the plate ump would then have the fewest calls per game.
Colin Wyers: I’ll be honest and admit that I’m still a little miffed at the way the umpires handled the addition of instant replay – their arguement basically boiled down to, “But if you implement instant replay, the asinine restrictions we insisted on as part of our negotiations with MLB will slow down the pace of the game!” Well, no kidding! Maybe I’ll cool off in a few weeks, but right this second? I’d replace the whole lot of them with motion sensors and animatronic ball-and-strike-calling robots from the Wonderful World of Disney in a heartbeat.
Eric Seidman: In theory, yes, it would be great to always get balls and strikes correct, but I personally feel that it is far more interesting for certain pitchers to test the strike zones of certain umpires and adapt.† For instance, if we made it a Pitch F/X strike zone, would it just be the proper strike zone, or would it be a certain amount on each corner?† Certain pitchers need questionable calls on the zones to succeed, and this would weed them out, which could be good or bad depending how you look at it, but ultimately, I don’t care enough about this issue to get worked up over it.
Pizza Cutter:†Remember the old NES game Base Wars?† (Still the single coolest baseball-related game ever made.† If they updated it and re-released it, I’d so buy it.)† Let’s replace everything with robots and computers!† (And… laser guns.)
The statistical geek in me says, “Well, of course.† We could probably put together a system that could check balls and strikes more accurately than a human†umpire.”† (Perhaps we’ll call it QuesTec.)† Accuracy is good.† But half the fun of a baseball game is grunting when the umpire either completely blows a call or calls the opposite of what you thought on a close pitch, and then complaining about the ump.† We stat geeks have to stop somewhere… right?

Question #2: Despite claiming to have read “The Book“, a couple weeks ago Ned Yost was fired shortly after having Brian Shouse intentionally walk Ryan Howard to get to Pat Burrell. Do you have ideas on how we might analyze play by play to rate a manager’s in game strategies?
Matthew Carruth: Well, we could certainly use win expectancy here, adding up the changes in WE over the course of the season on plays that the manager calls for. The issue would be deciding on those plays. Not all bunts are called for, some steal attempts are done by the player’s initiative, some are failed hit and runs, etc. But if you could resolve those questions you would have a decent first half of a method. For the second half, bullpen usage, I would propose a weighting of the relievers effectiveness (by your measurement of choice) against the leverage index of the situations in which he was brought in. A good manager will maximize his good relievers in high leverage spots, so there’s room to come up with something there.
Brian Cartwright: There are two basic categories to be modeled: moves made, and those not made. For each game situation, we would need to calculate what is the run expectancy, or WPA/LI, of the choice the manager made (even if it is do nothing Ė let play proceed) but also of the highest value alternative. Whatís the difference between bunting or not bunting with this batter. Leave the pitcher in, or bring in the LOOGY, etc, but we might be able to add up the values of all the choices. Make it so that low leverage situations donít really matter, but high leverage ones do.
Colin Wyers: I think you really have to take a step back first and start to model what a manager should be doing in the first place, and once you have a well-validated model, then you test. Using win expectancy and leverage we can get a pretty good idea of the appropriate thing to do, assuming only that play matters.
But for stuff like bullpen usage, what we first want to be able to articulate is a whole-team model of reliever usage. How often can a guy pitch? How often do we want him to pitch? How often does he absolutely have to pitch in order to be effective? Managers are tracking a lot more variables than our current frameworks dictate – that doesn’t mean we can’t critique some particularly egregious things, like the Eckersly model of a closer, but it does mean that we need to be careful in understanding that we are viewing things through a darkened glass.
Eric Seidman: Something like this would first call for an understanding of what the correct decision would be, which is somewhat up in the air for certain decisions or moves to†begin with.† Chris Jaffe is writing a book in which he evaluates managers and I am extremely interested in how it turns out.† For me, the outcome is not necessarily as important as the decision made.† For instance, bringing in a lefty to face Howard, when he has well-documented struggles against them but feasts on righties, is the right decision.† If he happens to hit†a home run, it does not negate the fact.† It would be very difficult to ignore the outcomes but we would have to, to some extent.
Pizza Cutter: Oddly enough, I had actually sketched out a few ideas on the subject a few weeks ago, most of them were pretty standard (SB success rate, bunt success rate, etc.), although the one that I was most proud of was to look at lineup construction.† How often does the following happen:†of the nine guys in the starting lineup, one of the worst four in OBP is located in either the leadoff or second spot in the order.† I also thought about looking at games which were tied at the start of the seventh inning to see how good a manager was at “pushing buttons” but that gets into a host of problems in terms of player personnel.
Question #3: The first round of the playoffs is ready to start, and the rules are simple: two teams enter, one team leaves. Which teams from the first round will move on to the second round of the playoffs?
Matthew Carruth: Boston takes out the Angels. The Red Sox are the superior team and have the better pitching both at the top of the rotation and in the back of the bullpen. The Rays move past whomever because it’s not time for the roller coaster to end and the Rays will have an advantage with a set rotation against either Minnesota or Chicago who have had to burn pitchers getting in.
The Brewers and Phillies are exceedingly well matched, but with Ben Sheets ailing and Yovani Gallardo still an unknown, the edge right now goes to the Phils who have done a good job making over their early season rotation woes. The Dodgers are better than most people give them credit for, but the Cubs are a legit elite team and can match the Dodgers’ pitching. The Cubs bats should win out.
Brian Cartwright: I really have to confess both my homerism and stats geekiness. I follow the Pirates, and the stats of individual players, but sometimes get in a fog evaluating other teams. Just not in the habit of looking at it that way.
Anyway, Iíll take a few WAGs. Two Chicago teams, two LA teams (well, so they say). The TV people ought to love this. I already expressed my admiration of the Red Sox, so letís go with them over the Angels. White Sox over Tampa, Phillies over the Brewers, Cubs over the Dodgers, but I am pretty undecided about both the NL series, I think theyíre pretty close. Expect them to go the maximum.
Colin Wyers: I’ll take the Cubs over the Dodgers in four games. I’ve been up to my eyeballs in Cubs/Dodgers stuff ever since the Brewers won the wild card – the Dodgers with Manny are probably a better team than their record, but the Cubs are probably a better team than the†Dodgers.†In the other NL matchup, I’ll say the Brewers over the Phillies. The Phillies have the sort of rotation that doesn’t feel like a playoff rotation, whereas the Brewers don’t really have a rotation at all, just one man on a mission.
For the AL – let’s say Tampa over the White Sox (please) and the Red Sox over the Angels. Red Sox – Angels might be the closest matchup in the first round. I think the International Union of Sabermetricians required me to mumble something about Pythag here, but I think Rally has taken enough abuse recently as-is over at Baseball Think Factory so I’ll leave it alone.
Eric Seidman: Phillies over Brewers (at least I hope).† Red Sox over Angels, Rays over White Sox, Cubs over Dodgers.† I really feel the Red Sox are the best team in the playoffs, and that the Phillies have learned from their playoff “run” last year.† The White Sox have the momentum of yesterday’s win, but the Rays have been riding†momentum all season long.† Plus, they’re such a great story!† Who cares about who wins, though?† I just want ads for Frank TV!!
Pizza Cutter:†I’m told C.C. Sabathia will actually pitch all five games of the Brewers/Phillies series.† The Brewers have a better shot†than the joke(?) would indicate,†but I still pick the Phillies there.† The other series is being played by the Dodgers and a team whom I am forbidden from mentioning, because I still have a lot of friends on the North Side who would kill me if I jinxed the Cu… errrr, that team.† Because jinxes are what cause the Cubbies to lose.† I see the Dodgers winning fewer than three games in this series.† (Side note: I lived in Wrigleyville up until July of this year… can I still claim the Cubs as my “hometown” team and not look like a total bandwagon jumper?)† In the AL, Boston against the Los Angeles California Angels of Anaheim California which is near Los Angeles is a 50/50 shot.† LACAoACwinLA has home field advantage, but Boston has more people in their lineup who can hit baseballs.† And if you’re not cheering for Tampa Bay, you must like torturing orphaned puppies.
Question #4: We’ve heard a lot about how, regardless of the gaudy saves total, Frankie Rodriguez is not the best closer in baseball this year.† Is there anyway, using stats other than saves, to show he was?† Or do we have to manipulate data to even make him a case?
Matthew Carruth: Well, according to FanGraphs, K-Rod had the highest leverage index of qualified relievers when entering a game, so he didn’t get all his saves cheaply. But after the whole entering part was over with, he certainly didn’t pitch all that superbly no matter what measurement you go by. Strikeouts, walks, FIP, ERA, tRA, WPA/LI, take your pick. He had a fine year, but his save totals were a product of his team, not his performance level. And with the market set by Mariano Rivera and Francisco Cordero last winter and the Kyle Lohse contract already this winter, it could be mind-boggling what Rodriguez ends up signing for, but it almost assuredly will not pay dividends.
Brian Cartwright: I would prefer expressing it as a rate. Ichiro has the most hits, but we give the batting title to the player with the highest batting average. As I showed a few weeks ago, you have to include holds in both the numerator and denominator. Then regress it. And probably weight past performance, so now we have a Marcel. I havenít run the save numbers this way (yet), but I may lean to the previously oft-maligned Brad Lidge (that should make Eric happy).
If you look at it from WPA (Win Probability Added), I’ll use Baseball Prospectus’ WXRL, and there Lidge is tops with 7.59, Mariano Rivera 2nd with 6.25, and then Rodriguez 3rd with 5.66
Colin Wyers: I don’t know if you can end up saying anyone was the best closer without looking at save totals, or save percentage, or what have you. And that’s really the problem. What is a catcher’s role on a team? He catches the baseball when the pitcher throws it over the plate. What’s a hitter’s job? To hit the baseball when the pitcher throws it over the plate? What’s the closer’s job? To get the save. Pretty much by definition, the save is the best way to evaluate the closer’s performance.
Maybe I’m being a bit too literal, but the question really boils down to, “Who was the best relief pitcher who lead†his team in saves?” Simply by asking the question you’ve at least bought into some of the basic assumptions of the save and the fact that managers manage to the rule rather than to win the game in front of them.
Eric Seidman: If you are making a case for him as the best, you could mention that he had a 2.57 Leverage Index, almost 3/10 higher than any other qualifying reliever, so not only was he used most often in critical situations, they were pretty darn critical!† Unfortunately, this is the only legit statistic to shine him in a best-closer light.† His K/9 is down, his velocity is down, his WHIP and FIP were higher, his WPA was 6th, behind Lidge, Rivera, Soria, Nathan, and Marmol.† His WPA/LI was right around one win above average, which paled in comparison to others.† Yes, he recorded 62 saves, and yes he had a ton of save opportunities, so he did his job, but that doesn’t make him the best at the broad spectrum of players that his job falls under.
Pizza Cutter:†If you torture a data set long enough, it will say what you want it to say.† So, why not make the opposite (and equally specious) argument that K-Rod was the worst reliever in the AL last year.† After all, he blew 7 saves (3rd in the AL).† Among those with at least 20 saves, he tied for the second-lowest number of games won (he went 2-3, tying him with C.J. Wilson, B.J. Ryan, and Troy Percival, and behind Joe Nathan with one win.)† And we all know, it’s all about wins.† And since we’re only looking at “fantasy” stats, among that same group of guys who saved 20+ games, K-Rod had the third worst WHIP (behind George Sherrill and C.J. Wilson).† So, by these measures, K-Rod was actually among the worst relievers in the American League.† Statistics never lie.† Liars use statistics.
Question #5: Is it time to re-visit the schedule format in MLB?
Matthew Carruth: Personally, I would like to see not just the schedule reformed, but the whole division and leagues realigned in order to create more regional rivalries and help cut down on the amount of travel teams have to go through. But that’s not going to happen any time soon, so revisiting the schedule would be a satisfactory first step. I’ll toss this out as a first suggestion; have five game series be the norm. You maintain the balance of having an odd number but you expand it out to cover a full go through a team’s rotation. I think that has the possibility of making things more balanced.
Brian Cartwright: Without reducing the number of opponents, I donít think itís possible. I do wish it was better. I look at the current schedule and see the Pirates ending the season in San Diego. What? And they only make one trip to Philly. Back in the good old days (1970ís for you younger guys), teams would open the season with a home and away with each team in their division. Then the other division, your own, the other, then September would again only be against division rivals. It was all so predictable.
Colin Wyers: I really don’t know what the answer is, but I really believe that MLB needs to work on getting a more balanced schedule. The idea that a team’s season could hinge upon the fact that, say, the Cubs’ “natural rival” is the White Sox and the Cardinals’ “natural rival” is the stinkin’ Royals is rather offputting.
Eric Seidman: I’m not bothered by divisional play, because the point of having divisions is that you are going to play the teams grouped with you more often.† What I dislike is inter-league play in the format it is in.† I don’t like how natural rivals are picked, or why the Phillies are seemingly natural rivals with both the Red Sox and Orioles.† Of course, then again, the only way to truly avoid this is to cancel inter-league play, which would just be blasphemous to even mention.† What if we just dismissed the separation of divisions and leagues, put all 30 teams into the same grouping, and had everyone play everyone an equal or close to equal amount of time?† Personally, I’d rather just keep the current format, but divide interleague games up amongst the rest of the league.
Pizza Cutter: I’d get rid of inter-league play, not on traditionalist grounds, but on pragmatic grounds.† As much as I’ve always wanted to see the Twins play the Padres, if we’re going to have a wild card that means anything, then those 15 games need to be put back against league opponents to balance out the strength of schedule issues that are sure to follow.† Maybe, for marketing purposes, have one or two series per year (for some reason, we have to put series together based on geographic puns, which is why the Rays play the Marlins home-and-home in the “Battle of Florida.”)† In fact, to make the Wild Card meaningful (and for that matter, the division races), I say go back to the 1994-96 model of a league-wide balanced schedule.† The regular season is first-and-foremost how baseball sorts out the post-season entries.† All of the theories on how to schedule should flow from preserving maximum integrity of that process.† And if it makes them feel better, they can load the division rivalry games into late August and September.

Advertisements

2 Responses to World Famous StatSpeak Roundtable: October 1

  1. kevin says:

    The question of rating managers’ performance using objective stats seems to be hugely complex.
    Example. Let’s say you are rating a manager’s bullpen usage by assessing if he uses his best reliever in the highest-leverage situation. So, it’s June 1, and the manager has a rookie with 25 innings and excellent peripherals, and he has a veteran setup man who has mediocre stats. There’s two men on, nobody out, tied game in the eigth. The manager uses the veteran. Should have used the rooke, right? Now, it’s August 1, and the rookie has come back to earth, with an ERA of 4.80 and peripherals to match. The veteran still bumping along decently. Same game situation again. Suddenly, the manager’s decision on June 1 doesn’t look so bad. He didn’t know enough about the rookie’s capabilities, so he went with the veteran, even though it could be argued the rookie was better. Then, two months later, it seems as if the veteran truly is better — and was back in June, too. Do you revise your rating of the manager’s June decision?
    Shorter example. It’s one thing to quantify what the manager should do, but how do you rate the manager based on the actual outcome? If the manager makes the “wrong” or sub-optimal decision, but it works, that should carry some value. Half-credit? Quarter-credit? What?

  2. Colin Wyers says:

    That’s why I’d use projections as a basis for true talent level, not season-to-date stats.
    Honestly with reliever usage, it’d probably be better to assess a manager from a seasonal perspective – simply look at how he alloted high-leverage situations during the season.

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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: