Keith Woolner is an author of Baseball Prospectus.
Keith Woolner: Hi everyone, let's get chatting...
Joe torre (lights on/nobody home): Sorry this question came so late, but I justified letting Embree pitch to Konerko (and losing the game) by pointing out Konerko's 2005 splits, why can't I see the fallciy in one year splits?
Keith Woolner: While the sabermetric perspective certainly believes that one year splits are too small a sample to rely on, that's a fairly sophisticated mathematical argument to make.
Furthermore, if you are a manager who believes that batters can evolve radically over time -- adjusting their approaches at the plate that make their past history less relevant, it's easier to understand.
Doesn't make it right, but somewhat more tolerable.
At least there was *some* objective evidence being used in the decision-making, rather than "I was playing a hunch, and had a feeling."
Mike (record holder rutgers): Next year, do you think you could keep each team's expected runs matrix, and make them available to put into spreadsheets?
Keith Woolner: That's an interesting idea -- and I'm always keen to get suggestions for how to improve the statistics on the site.
I'm not really sure that a single season for one team is enough data to make the expected runs matrix useful for analysis.
The same problem exists to some degree with the expected wins matrix -- whereas there are only a few cells in the exp. runs matrix, there are dozens of cells in the expected wins matrix (all combinations of inning, outs, baserunner states, and run differential, plus home/away and which team is leading), which render the sample size for many of the cells to a vanishingly small level. It's quite easy to find examples of the "expected win" for a small lead to be lower than a the same situation with a bigger lead, which is counterintuitive.
That's a large part of the reason I introduced the Win Expectation framework in BP2005. The expected runs matrix for a team would benefit from a similar treatment -- perhaps presenting a matrix for the exp. runs given the team's overall offensive production.
Dave (San Francisco): Do you think that the Giants bullpen is the best in the NL West? --> Between Scott Eyre, Scott Munter, Tyler Walker, Latroy Hawkins, and Jason Christiansen, it is the one of the biggest reasons the Giants are [almost] in contention.
Keith Woolner: It's not as clear cut as you think. Eyre has been the best reliever in preventing runs in the NLW with 19.1, but Munter is the only other one over 5 with 8.1
Contrast that with the Padres -- Otsuka (15.0), Hammon (13.0), Linebrink (11.5), Seanez (9.3) and Hoffman (5.2)
dmaybee (Virginia): Ever wonder about how significant stats would change when great hitters are on the same team as great pitchers and therefore do not have to face each other. For example (Pujos vs. Carpenter or Derek Lee vs. Prior) These matchups will not occur this year but I wonder over the course of a year if their individual stats were signficantly benefitted by not having to face each other. What do you think?
Keith Woolner: You can look at the batter's quality of opposition report to get a sense for how big the effect actually plays out.
That report takes every plate appearance by a batter, looks at the opposing pitcher, and what an average PA against that pitcher looked like, and composes the profile of what an average hitter would have done against the same mix of opposing pitchers. It's a little hard to get your mind around a textual explanation like that, but just remember that high OBP/SLG on that report means the batter faced an easier quality of pitching than a lower OBP/SLG.
Gavin (San Francisco): Voros has taught us that the outcome of balls in play is largely not controllable. For that, and other reasons, K-rate and K/BB ratio are very important pitcher stats. Why would the same not be true for batters? Given that putting the ball in play results in a league average of about 0.290-0.300 (or whatever) wouldn't this be a high incentive to build a team with a low K and K/BB rate (just as the opposite is true of the ideal pitching staff)? And isn't this exactly what Billy Beane has done this year? The A's have the fewest strikeouts and the lowest K/BB ratio in the AL...
Keith Woolner: Voros McCracken taught us nothing of the sort, and his article on the relationship between balls in play across pitchers is one of the most misunderstood of recent years.
McCracken's work (and the followup work by myself here and here, and Tom Tippett and others) has shown that the range of abilities for pitchers to prevent hits on balls in play is much narrower than was previously thought, and that to a first approximation you can get away with saying that major league pitchers don't have *any* difference in ability.
That's far from saying that the outcome of batted balls is totally random or uncontrollable. First of all, some pitchers *do* have some demonstrable ability (as shown in the above articles), but the range of ability is very narrow, and shows a fairly weak, but definitely present, correlation over time. Also, pitchers definitely differ in their ability to create ground balls vs. fly balls, even after removing home runs, and so their opposing SLG or isolated slugging will vary directly with the pitcher.
But more importantly, the research says nothing about whether *batters* affect ball-in-play hit rates. In fact, I looked at this years ago, and found that they have a large effect on the outcome of a batted ball. See the article for details.
Fonzie (Aaaay): In a recent chat Joe Morgan said "The greatest defensive CF of the last 15 years is either Andruw Jones or Ken Griffey Jr." Please comment.
Keith Woolner: This is absolutely true, in much the same way that you could say that the greatest hitter of the last 15 years is either Barry Bonds or Keith Woolner.
Dennis (Newark): Can the theory behind "relief aces" be applied to hitting? I've always contended that the Barry Bonds should start the game on the bench. If you assume that the team has interchangable outfielders (and maybe one who can play 1st base too), then you could pinch hit Bonds when one of them comes up in a high leverage situation. In a normal game, Bonds could be pitched around the entire time and never come up with as many RBI opportunities as that one at bat. If the game becomes a blow out before Bonds gets in, then you can give him the whole night off.
I think this type of scenerio would allow the Giants to maximize the innings they can get out of Bonds in a season.
Keith Woolner: There's a big difference -- an "ace" pinch hitter can only be used in one guaranteed high-leverage situation. A relief ace faces consecutive batters in the same high leverage situation, and so has more opportunity to affect the game's outcome.
Given that relief aces average a leverage of about 2.00 over the course of a season, I can safely say that I'd rather have 5-6 plate appearances of Barry Bonds, even with the bases empty, than knowing I can use him in one pinch hit appearance in a clutch situation that's worth double the usual.
Your best hitters can help keep the game from being close enough to worry about pinch hitters or relief aces. Get them as many PA as you can.
johnpark99 (Boston): Have you or has BP ever done much work analyzing situational LOB? I wonder if the LOB stat that is widely used can be improved on. For instance, it is likely (and perhaps reasonable) to expect a guy who hits a single with 2 outs to remain LOB at inning's end, while it should be considered travesty for a guy who hits a leadoff triple to get stranded. Do you think that a statistic that accounts for such things would be valuable?
Keith Woolner: This is essentially the same as looking at a slice of the expected runs matrix before and after a hitter's plate appearance. You charge him with losing more runs for creating an out where the expected runs total is high than if there are no runners to advance.
The Leverage stat that we compute for relievers is equally applicable to batters, and I will probably introduce such a measure in the next iteration of our statistics database.
I think such a statistic is useful for certain purposes. I wouldn't limit it to LOB situations, though, and would use it to rate batter performance across all his PA.
Velez (Miami): If Victor Diaz starts the year at first base, would the Mets be any closer to the wild card?
Keith Woolner: They'd probably a couple of games closer. Diaz has been about .130 MLVr better than Mientkieciz, which is worth about 20 runs over 150 games. But Mientkiewicz hasn't been *that* much worse than Diaz so far this year.
But it hasn't been a good idea to play Jose Offerman at first base since the 20th century, if it was ever a good idea to begin with.
Ross (England): No articles since March? what have you been up to?
Keith Woolner: Guilty as charged. I've been spending a lot of time behind the scenes, working on various projects that don't have bylines -- we started working with a new statistics provider this year, and that required a lot of attention early in the season. We also introduced the Sortable Statistics reports, which Ben Murphy and I mostly developed. I've been working with Dave Metz on setting up the BP Library that organizes many of our best research articles into topics, and makes our back catalog of articles much more accessible to readers who haven't been with us since the early days. And lately I've been working on a new overhaul of our statistics processing that will allow us to present more detailed and flexible player, team, positional, and league statistics than we've been able to do before.
dangor (new york): The main complaint that I hear from casual baseball fans is the slowness of the game. Why won't they enforce some speeding of the game? 10 seconds per pitch, 60 seconds between innings, max pickoff attempts. It seems so easy to do.
Keith Woolner: I'm all for speeding up the games, too. However, there are some economic realities to face. The time between innings is dictated by the television contracts allowing them to have commercial breaks, for example.
It should be possible to cut down on the time between pitches by not granting time as readily to batters who wander away from the plate, and enforcing a time limit (30 seconds?) between the end of a play, and the next batter being in the box ready to hit.
My personal pet peeve is the number of times a manager changes pitchers mid-inning. I'd like to see a rule that a pitcher who did not start the inning needs to face a minimum of three batters, or finish the inning, before being replaced.
alappin (Needham MA): How do you calculate MLV? The link in the glossary declares that it takes a look at the player's AVG, OBP, and SLG; the league's AVG, OBP, and SLG, and the number of games in the season and number of outs per game. But that ignores SB, CS and double plays, things that have to have an effect on a player's true value. Is this a limitation of the metric, or is there a newer version of MLV that addresses this?
Keith Woolner: The most complete writeup of MLV I've done is actually on my old (and neglected) Stathead.com site -- here's the link.
MLV addresses solely a player's prowess as a hitter, not as a baserunner (thus doesn't include SB and CS). However, even then it doesn't include double plays, as you rightly note. It's also based on the simple Runs Created model, which is a bit archaic and prone to some level of inaccuracy versus some other alternatives we have today.
Updating MLV to account to use a better model and account for double plays has been on my list for a long time. The baseball world doesn't really need another run-scoring model (RC, Linear Weights, XR, BaseRuns and others all do a decent job), but I'll probably end up coming up with something that satisfies my own aesthetic quirks. In particular, I am not comfortable with a run scoring model that charges a batter for double plays without somehow accounting for the double-play opportunities in the process. We know from reports like this one that batters see a vastly different percentage of PA where a double play is possible, yet none of the major run production models seem to account for it.
Steve (Manalapan): I noticed that a lot of the authors at BP were harsh on Michael Young before the season and in the past. Now that he's had another great year, is it time to come around on him?
Keith Woolner: To be fair, Young didn't post a .350 OBP until 2004, and even then it was only .353. But he has good power for a middle infielder, and has been very durable. That's tremendously valuable.
BP authors have their own biases, even though we try to be objective, and it's sometimes hard to look past the fact that a player has never drawn 50 walks in a season, even if he's getting on base 200 times other ways. And when we do, it's perfectly legit to call us on it.
PJ (Parsippany): Why don't starters come in for an inning of relief on their "throw days"? If they are throwing at full strength on the side, why not capitalize on it? You can get 40 more innings from a starter throughout the season that way.
Keith Woolner: I believe that this practice was more common in the olden days that Steven Goldman likes to write about.
There are risks and problems, though. A pitcher throwing on their off day can take it at whatever pace is comfortable, work on specific pitches, focus on their mechanics, and otherwise work on the craft of pitching without having to take the game situation into account.
Throwing pitches in a game may be more stressful on the arm than regular side work too. If the pitcher gets into a jam, and ends up throwing 40 pitches to get through 4 batters, you risk burning him too much before his next start.
I do think that it would be a creative way for a team to get a more effective bullpen, and it should probably be used more than it is.
Keith Woolner: That's all we've got time for tonight! Thanks for the questions everyone. Next chat is with James Click on the 22nd, so make sure to throw some hard questions his way. :-)