Rob Arthur finds data where no data was previously known to exist. Ask him about the crack of the bat, or other frontiers of sabermetric research.
Rob Arthur: Ask me questions, but be forewarned, I know nothing of prospects or fantasy baseball. If you take my advice, you are probably doomed.
DavidPaton (Toronto): So where does the line-up start for people apologizing to Dayton Moore, for all the "bad" moves he made?
Rob Arthur: I guess somewhere in Kansas City? In all seriousness: in baseball, good results do not necessarily come from good decisions. That said, I think I (personally) underestimated some of his moves, and I probably wasn't alone among sabermetricians in that.
Rick (KS): How worried do I need to be about what I perceive to be overuse of Ventura? More broadly what data is out there on rookie pitchers who are building/stretching their innings limit? Is there a way to tell if a team is handling a young pitcher right?
Rob Arthur: I'm going to answer these in reverse-order. No, there's no way to tell if a team is handling a young pitcher right, just because they know so much more about their pitchers than we do. Hence the saber metric community's general shrug about the weird usage of Wacha and Duffy, for instance-we just assume that the teams know what they are doing.
There's no data, or very little. I'd recommend this piece from Ben Lindbergh, which I think shows the depth of our current ignorance: http://www.baseballprospectus.com/article.php?articleid=17517
And finally, I would say you should be moderately worried, because if ever there was a time where a team and a pitcher might choose to forgo caution and push the limits of a pitcher's endurance, it's during a championship race for a franchise which gets into the WS about once a generation.
Dragonbreath (Gurnee, IL): zDo you see Hunter pence maintaining his consistent numbers into 2015,16, and 17,?
Rob Arthur: Nope, because aging. PECOTA agrees, and has him down for drops of 10 points of TAv in each of the next two years. It might be more like 0, then 20 points of dropoff, but I don't think he'll be performing at his current excellent level for much longer.
golfer Nacho Elvira (the links): Help us all, Rob; we need a new metric to evaluate pitchers. All the variations on DIPS exclude several repeatable skills that some hurlers possess in much more abundance than others -- e.g., holding and/or picking off runners, pitching from the stretch and/or with a slidestep, inducing popups and/or GIDPs, deftly fielding batted balls, and most importantly suppressing both BABIP & non-HR ISO by getting ahead in the count and exploiting that advantage to elicit weaker contact. These are all included in ERA and ERA+ of course, but those also include randomness & team defense. Help!
Rob Arthur: I hear you: we do need a new and improved metric, or maybe set of metrics. Right now we have sort of an ad hoc approach to pitcher evaluation, using a set of partially independent, but flawed statistics (FIP, BABIP) and kind of eyeballing it. That's not ideal, but so far no one has come up with a better way. Maybe when I'm done digitizing bat cracks I'll take a stab at it. But it's not easy.
Dragonbreath (Gurnee, IL): Can Gbiancaro repeat his 2014, next season?
Rob Arthur: (Assuming that's "Giancarlo") Always bet on regression. That said, he's got as good a chance as any to perform at an MVP-type level.
Alex (Anaheim): Can we predict the outcome of a short baseball series with any degree of certainty at all?
Rob Arthur: Not very much certainty, no. As I understand it, PECOTA odds for short series get as high as ~70% in favor of the better team. In some hypothetical future in which we understand baseball perfectly (or nearly perfectly), I could see the percentage maybe getting as high as 80 or so. Baseball is just very stochastic, which is part of why it's fun to watch.
OB1KENOBI (Tampa): who would you count on for this coming year here's a look at pitchers(how do they rank)Is there a Jose Fernandez hiding here.
1-Kennedy 2- Lackey 3- McHugh 4-C.Anderson 5- A,Sanchez Worley Walker
Philips, House, Carrasco, Hultzen, Guerrieri. I been told in points league Ill have a problem with this staff.
Thank you very much for your time
Rob Arthur: There's no Fernandez there, because Fernandez was both 1) incredibly, historically good, and 2) very, very surprising.
William (Spokane): Do you think the eight games the Royals have played are more significant than any other eight games? Should we talk about them as good in a way we might not have in September, or is it all BS?
Rob Arthur: Yes, they are more significant, because the playoffs ARE different from the regular season. Witness, for example, the fact that the Royals ace reliever trio has pitched more than twice as much in the postseason (on a % basis) as during the regular season. However, I don't think they are nearly as good of a team as they have played this October.
I have an article coming out soon that addresses another reason why the Royals are playing better than we expect.
bachlaw (Chicago): Hi Rob,
Would you mind sharing which regression methods you tend to find the most useful when taking a first pass at a new series of data, and you are just trying to get a sense of where you should go with it? Thanks very much.
Rob Arthur: This is tough, because it depends a lot on the questions being asked. I do a lot of correlation analysis, especially using the Spearman correlation coefficient, which is much more flexible in terms of the relationships it can detect. Besides that, just a bunch of linear and logistic regression, I guess.
Matt (Portland): After 2014, who wins another World Series first, the Royals, the Giants, or the Yankees?
Rob Arthur: Hmm, I'm with Ben and Sam on this, I think. Despite the flattening of the playing field in MLB recently, money is the most consistent ingredient for winning, I think.
Matt (Michigan): I keep reading that now the Royals might be able to re-sign James Shields this winter. If it happens, will that ultimately go down as a feature or a bug of this postseason run?
Rob Arthur: That's a great question. Teams that win the championship often resign their veterans at reduced rates, but whether those are good moves really depends on the player. In the case of Shields, I would say that they are better off letting him go, so that would be a bug, unless he's willing to sign for an absurdly low number.
Andrew (St. Louis): Assuming you don't have the world's best, worst, or most handsome manager, is there any single move a manager could make that you would considerable fireable all by itself? For instance: Matheny/Wacha, or Matheny and his "can't use the closer in a tie" comment?
Rob Arthur: Oh yeah, tons of stuff. Punching players. Saying something racist/sexist/homophobic in a postgame press conference.
I get the feeling you are talking about in-game strategic moves, though. Besides insane stuff (I dunno, putting a position player in as pitcher with a 3-run lead or something), there's no single move, I don't think, that would make me want to fire them. Managers just don't have enough influence on the outcome of the game for a single boneheaded bit of tactics to be fireable.
Cal Guy (Cal): Hi Rob, Are you a proponent of using K/BB in evaluating both hitters and pitchers, and if so, what is the threshold value for K/BB for hitters and for pitchers where you start to pay attention?
Rob Arthur: Yes, but moreso for pitchers than hitters. And I would use K%-BB%, rather than the fractions, because the denominator can radically sway the metric (especially if you are dealing with a contact hitter).
No single threshold value, though, I wouldn't say. It's all on a continuum, and should be combined with other data that you have (e.g. batted ball profile).
Norm (Los Angeles): If i played you 50 bat cracks, do you think you could do a better job of telling me which 15 were hits than random chance?
Rob Arthur: Yes, I think I could, marginally. I assume you are talking about my ears? I'm less confident that my ears could than I am that my computer could (I know my computer can do that, because I've tried it).
Aaron Boy (St. Louis): If Dayton Moore were hired to be your team's GM, you would open a bottle of a) champagne b) sleeping pills
Rob Arthur: My teams have pretty good GMs already, so I guess sleeping pills? But really I wouldn't be all that disappointed.
jw (Glentown): Rob, do you think having two first names is im/amoral? I do. I'm curious to hear how you can defend this practice.
Rob Arthur: I don't know about the morality of it, but I can tell you it's a pain in the ass. I can't count the number of times I've had to explain that my first name is Rob, not Arthur.
Matt (Portland): What knowledge would you want to add to get that playoff prediction certainty up to 80 percent?
Rob Arthur: Oh, everything. Bat speed, HitF/X, Statcast, personal, recent psychological profiles. The whole shebang, and then we'd be at 80.
A lot of it, I'd guess, would be injuries, and really knowing for sure when a player has some nagging problem that's not sufficient to put them out of the game but explains why they are doing a little worse than usual.
Dragonbreath (Gurnee, IL): Joc Pederson figures it ot in the bigs in 2015, and plays like the stud he is projected to be?
Rob Arthur: Maybe not 2015, maybe later. But I think he'll be good.
Thom (La Mesa): Unintended consequences of StatCast data?
Rob Arthur: Hmm. Maybe injury information--if a player's speed is normally 18mph, and then they are suddenly running the bases or the field at 16mph, maybe we can detect those minor nagging injuries that I just mentioned.
jw (Glentown): Who are your: (a) baseball hero; (b) sabermetric hero; (c) real-life hero; (d) hero sandwich.
Rob Arthur: A) I really liked Roy Halladay, and I was sad to see his career end so abruptly.
B) The standard ones, I guess: Bill James and Rob Neyer and Nate Silver and so on.
C) E.O. Wilson, the biologist.
D) Like my favorite sandwich? Or do I like hero sandwiches? I don't like hero sandwiches that much, but maybe I've just not had a great one before.
John (Chicago): What % does the success rate need to be for stolen bases to make it worth the risk?
Rob Arthur: Depends on a lot of things, but I can recommend some articles, for example: http://www.beyondtheboxscore.com/2013/2/20/3980504/individual-stolen-base-break-even-rates
William (Spokane): You can have an average manager and an elite pitching coach, or an elite manager and an average pitching coach. Which would you choose? Would your answer change if I gave you no assurance that your assessments of each person would be accurate?
Rob Arthur: Elite pitching coach, I think. Yes, it would--it would make them more even, if I didn't know which one was really good at their job.
HW (MO): If teams can carry 13 pitchers in the regular season, why can't someone carry 12 plus Terrance Gore?
Rob Arthur: They could, but I think it's perceived that having the additional pitcher is a better use of the roster spot. Whether that is correct or not is perhaps an interesting question for future study.
Andrew (Warm Kitten): Madison Bumgarner is throwing 1 mph harder than he did in the regular season. His ERA forecast is presumably something like 3.10 generally. Armed with the new velocity knowledge, what would you forecast his ERA to be in his next couple starts? Is this what you mean when you say forecast ability for a series might approach 80 percent?
Rob Arthur: A little bit lower, although exactly how much, I don't know off the top of my head. And yes, that's what I mean--if I knew for certain that he was going to throw a little harder/pitch a little better, I could incorporate that into my forecast.
justarobert (Santa Clara, CA): How do you feel about the potential invasiveness of more detailed data collection about players, and is this a likely impending risk for baseball? (Compare the monitoring patches that some football and basketball programs are using.)
Rob Arthur: It certainly is an impending risk. It will be interesting to see what the player's union does about things like this. Wait til they start sequencing player genomes and using them for scouting (I think this is only a matter of time).
Emile Waldteufel (Home of Tim Teufel): Thanks for chatting, Rob! Mike Matheny's averaged 92 wins a year as Cardinal manager -- more than they won under LaRussa in 2011 or 2010 or any other year since 2005. Matheny's also won an average of 7 playoff games per year, which is *twice* the rate of Francona with the Red Sox or Girardi with the Yankees. So despite what Twitter says, I think he is by definition an elite manager. Thoughts?
Rob Arthur: Yes, I have seen this defense made as well. And I think there might be something to it, in that much of a manager's job is presumably getting players to work hard and cooperate with each other and spend the extra time to become better players. Matheny might be doing that really well, in order to get his team so much success.
Unfortunately, we can't see any of those managerial duties--they are basically invisible to us. So we go with what we can see, which, for Matheny, is dumb blunders in close playoff games.
Sam (Bay Area): Any hypothesis for why home field advantage doesn't get bigger in the postseason? Seems like any factor that would be causing HFA would be exaggerated in postseason, other than maybe park familiarity (though even that would in the WS).
Rob Arthur: Maybe because players are more rested in the postseason (more days off)? And so they are less susceptible to the stress of travelling, being away from home, etc. That's all I've got.
RMR (Guelph): How might we better tease out failures of scouting and failures of development?
Rob Arthur: So, trying to figure out whether a player was not as good as we thought versus being unable to realize their potential?
Tough question. I think some of it comes down to having more information about how teams actually develop players--what do they advise players to do or not do? Then we can look at whether those pieces of advice are useful, and why or why not. The trouble is, we don't often get to see what the teams advice or development plan is.
therealn0d (arizona): Can you think of any good way to explain the Giants recent run of success analytically? What are we missing?
Rob Arthur: Nope, but not everything IS explainable. There's always luck. We might not be missing anything. Sorry, that's a cop-out answer, I'll think about it more.
William (Spokane): If you wanted to bet at a sports book on a baseball team to win the 2015 World Series, would you rather use the odds set before the offseason or after?
Rob Arthur: After.
justarobert (Santa Clara, CA): Are you familiar with the computer science research on tracking basketball and hockey players from broadcast video (i.e. uncalibrated moving cameras)? It doesn't produce as accurate data as Statcast does, but public-sphere analysis may or may not get that raw data...
Rob Arthur: Yep, I have thought about this, and I plan to look into how feasible it is in the offseason. I'm not familiar with the literature though, maybe there's something already out there I can use.
Davis (Here): What are you three favorite baseball books?
Rob Arthur: To be honest, I don't read that many. I guess the standards: the Book, Moneyball, the Extra 2%.
Parrothead (St. Somewhere): How can Pecota possibly justify a 70-80% favorite for any single game when the best team in baseball can only defeat the average opponent 60-65% of the time?
I could understand 70-80% if Kershaw was slated to face Bollsinger in a July matchup. But such long odds seem unrealistic for any post-season game.
Rob Arthur: 70-80 for a SERIES, not a game. A 65% favorite over 5 or 7 games gets up to 70-80, I believe (someone check my math).
Dragonbreath (Gurnee, IL): Gregory Polanco 290 25/35 85, 85 in 2015 sound about right?
Rob Arthur: Less than that; close to PECOTA's forecast, but maybe a little higher.
hduumgpc (1): 1
Rob Arthur: 2, 3, 4. And on that scintillating note, I will end today's chat. Thanks everybody for your questions, I appreciate it, and I apologize if I didn't answer your question properly/at all.
Rob Arthur: See you next time.