Run Values
I got my copy (okay, copies, since I’m a geek who plans on giving out a couple to friends…) of the Hardball Times 2006 Annual today. I’ve mentioned before that I have an article in the book, so there’s little objectivity involved when I tell you to buy one. Actually, buy more than one. Buy as many as you’d like.
But, don’t worry, this isn’t just a plug, though it obviously is that too. I’ve only had a chance to read a couple of the articles so far, and because I’m fascinated by the batted ball data that they provide, I jumped right into one of the articles by Dave Studenmund called “What’s a Batted Ball Worth?” Since they purchase exclusive batted ball type data from Baseball Info Solutions, they have some really interesting numbers that you can’t get anywhere else. And this article is full of those types of numbers.
There’s one table in particular that jumped out at me, and I think it illustrates pretty well why we believe some of the things that we do and are often questioned about. Below are the average run values for each type of possible outcome.
Line Drive: .356
Hit By Pitch: .342
Non-Intentional Walk: .315
Intentional Walk: .176
Outfield Fly: .035
Groundball: -.101
Bunts: -.103
Infield Fly: -.243
Strikeout: -.287
This is particularly of interest as it relates to pitchers. Strikeouts are clearly the best outcome they can hope for in an average at-bat, though obviously, there are circumstances where a ground ball (which could lead to a double play) would be more beneficial. Walks are abysmal for pitchers. Inducing a groundball is better than giving up a flyball, even though the flyball has more chancees of being caught.
So, looking at the run value chart, the ideal pitcher would be a extreme groundballer who also strikes out a lot of batters and gives up almost no line drives while limiting his walks. Like, say, King Felix.
Just for fun, I took the percentage outcomes for these four outcomes and created a quick and dirty formula to create one “effective percentage”. (GB%+K%)-(LD%+BB%). Basically, how many more groundballs and strikeouts do they get on a per-batter ratio then they give up in walks and line drives. Here’s the list of the top 40, using a minimum of 300 batters faced, and I’ve highlited some of the names of interest to M’s fans:
Hernandez F. 52% Halladay R. 47% Rivera M. 47% Lowe D. 46% Webb B. 45% Westbrook J. 42% Williams T. 42% Wang C. 42% Carpenter C. 41% Gordon T. 41% Shields S. 39% Burnett A. 39% Cook A. 38% Loe K. 38% Santana J. 37% Johnson R. 36% Duchscherer 36% Qualls C. 36% Mulder M. 36% Reitsma C. 35% Peavy J. 35% Clemens R. 35% Pettitte A. 35% Silva C. 34% Hudson T. 33% Maddux G. 33% Oswalt R. 33% Street H. 33% Rincon J. 33% Martinez P. 33% Dempster R. 32% Sabathia C. 32% Johnson J. 32% Colon B. 32% Haren D. 32% Pavano C. 32% Frasor J. 32% Mussina M. 32% Wells D. 31%
By evaluating batted ball type rather than outcome, we’re evaluating more what the pitcher could control and less what is influenced by his fielders. I’m not offering this as anything other than a fun toy, but I do think the list is interesting, if nothing else.
Comments
44 Responses to “Run Values”
Wow! Felix tops a pretty good list here. Nice work Dave.
And yet people seem to forget that Halladay is a dominant pitcher…
Nice Felix numbers, there. I’d love to see both Roy and Felix give us full seasons so we can compare them come Cy Young voting time.
Nice work, Dave–
Shouldn’t Wiiliams and Qualls be disqualified from your list for not enough batters faced?
Any explanation why Lowe hasn’t been more successful than his ranking here would predict?
I’m fascinated at the different values. That a HBP would be worth that much more than a straight walk… my mind is a-flurry with theories.
I’m interested in what this means for the prevailing sabermetric thought that strikeouts are not that much worse than any other kind of out. This data seems to blow that idea out of the water. So why do we still largely dismiss strikeouts as just another out?
Addendum: I meant “from the point of view of the batter” in my last comment.
Qualls faced 329 batters and Williams faced 321. So, no, they shouldn’t.
Lowe’s problem this year was the home run. He gave up 28 dingers on just 153 flyballs, meaning 18 percent of his flyballs left the yard, compared with a league average of 11 percent. This appears to be a fluke; his HR allowed have always been right around what we’d expect based on his FB rates.
And yea, the HBP being that much more valuable than a walk to me is strange. I don’t really understand that.
Also, just for fun, only three pitchers in MLB faced at least 300 batters and gave up line drives in less than 10 percent of those matchups: Tom Gordon (8 %), Mariano Rivera (9 %), and Felix Hernandez (9 %). The closest starters to Felix were Oliver Perez, Daniel Cabrera, and Johan Santana, all at 11 percent.
The LD % numbers are really interesting. Relievers give up far less line drives than starters, and the starters who hold LD % down all tend to be young, somewhat wild, and Hispanic.
I’m interested in what this means for the prevailing sabermetric thought that strikeouts are not that much worse than any other kind of out. This data seems to blow that idea out of the water. So why do we still largely dismiss strikeouts as just another out?
I wondered how long it would take for someone to bring this up.
Read your paragraph again. The sabermetric thought is that “strikeouts are not that much worse than any other kind of out“.
This data doesn’t deal in results, with the exceptions of walks and strikeouts. It deals in batted ball types. You can look at this data and clearly say that a strikeout is significantly worse than making contact. You can not look at this data and say that a strikeout is significantly worse than a groundout to second base.
A strikeout is the worst kind of out, but not by that much. But that’s not really what this table is talking about.
Walks, intentional walks and HBP passes to first base having differnt outcomes is fascinating, bewildering and is beginning to point towards a real impact on the mind set of both offense and defense. In pure mechanical terms, it should not matter how the base runner got on but the data shows otherwise.
Cool work. I just bought the book, looking forward to reading more about this and also your work there also. It will undoubtedly be a more entertaining use of $22 then then the 3 tickets I bought last week to a crappy movie.
The rest of you should buy a copy today.
RE: #6
Are LD% numbers considered repeatable, for either hitters or pitchers?
Fascinating.
re: HBP/walks
This make any sense… if unintentional walks are a sign of lack of control, then (and this presupposes that most HBPs are unintentional) HBP is a sign of an even greater lack of control… leading to a worse expected result for the floowing batters? (also perhaps historically rattles the pitcher?)
Okay, but then why does being “a sign of lack of control” make it more or less valuable?
My thought on this: we’ve seen the semi-intentional walk, where pitchers don’t want to give a guy anything good to hit and are willing to take the walk if that’s what it comes to. These are, obviously, less frequent in situations where it’s really bad to put a runner on.
No pitcher wants to bean a hitter, though (wellllll… okay, just grant me that for a second). When they do, it’s more likely to be a situation where doing so will hurt them a lot, compared to an intentional walk, or even a normal walk.
Still… this requires more thought.
I think #9 and #11 are on to something, with respect to run values of HBP vs. BB vs. IBB. I think the expected run values show variance because of the situations that those events accumulate in.
I don’t know exactly how the data are measured, but it sounds to me like they just got an average change in run production from various types of events. Walks and HBP tend to be issued in situations where the pitcher is struggling. When the surrounding events are bad, a walk or HBP would magnify the “badness” of the situation.
IBB, on the other hand, don’t come when there’s deteriorating control, and are issued at those points that are least likely to hurt you by having another runner on base. (Note, however, that a IBB still increases expected run value. Issuing an intentional walk is still bad.)
Sorry about the earlier dumb question about Williams and Qualls. I hastily grabbed their 2004 stats.
One more for you: Do you think this indicates that Jason Johnson would be a good FA pickup?
This is fascinating.
One question: Is there any info on the impact of reaching base on an error? It would seem that an error, like an HBP, might result in more runs than logic would otherwise dictate.
“That a HBP would be worth that much more than a straight walk… my mind is a-flurry with theories. ”
Is this results based (i.e. examining how often someone hit by a pitch comes into score)? If so, could it be just randomness?
The one theory I would float is that players who walk a lot frequently are the better hitters but HBP’s (with some exceptions) are more random, so a player who gets on by a HBP is more likely to have better players in the lineup behind him to drive him in. That’s just an initial thought though.
Are LD% numbers considered repeatable, for either hitters or pitchers?
We don’t really know. There hasn’t been a thorough study on LD% correlations yet. I’d imagine there is, but probably with more fluctuation than BB% or K%.
The interesting thing to me is that relievers have a significant and obvious advantage in preventing line drives. It’s clear that pitching in relief is just a totally different animal than starting. My just total guess is that velocity is the big factor here, and that a lot of relievers can throw in the mid-to-upper 90s for a huge majority of their outings, while it’s nearly impossible to pitch as a starter and maintain that velocity.
This is the main reason why relievers have a significantly lower BABIP than starters do.
I understand the IBB thing though, since that’s a strategic move to either try to get a double play or get to a lesser batter. A lot of IBB’s are to the #8 hitters in the National League when there’s a runner in scoring position and two outs. How often would those runners score?
DMZ, Robbbb…
I think that’s all true.
To me, the key point is this: it’s not how THAT batter got on base that is better or worse, it’s that the at bats following a HBP (for more than one reason) statistically have worse results for the pitcher than at bats following a regular or intentional walk.
Are LD% numbers considered repeatable, for either hitters or pitchers?
Okay, I lied, there is a study done on year-to-year correlation of LD%. It’s in the Hardball Times 2006 Annual. I just hadn’t gotten around to reading it yet. Yet another reason to buy the book.
And, to answer the question, it apparenly is a resounding no. The correlation from 2002-2004 for pitchers who faced at least 350 batters in consecutive years was -0.03. On a scale where 0 equals “no predictive value” thats about as close to no correlation as you can get.
So, I’m surprised, but apparently the fact that Felix had a great LD% this year has no relevance on whether he’ll have one next year. This definitely requires more research.
but apparently the fact that Felix had a great LD% this year has no relevance on whether he’ll have one next year.
I don’t know quite where I’m headed here but perhaps the inability to repeat LD% is that the hitter determines this and not the pitcher. We know that some hitters are better LD hitters then others. The mix of hitters faced is not constant and with players moving from team to team and league to league, the “batters faced” are very random.
I don’t want to give away all the info in the book (seriously, if you’re reading this thread, you’ll love it, so buy one), but hitters have a very weak correlation on LD% as well (0.10).
Basically, this data tell us that we don’t know much about what drives line drive percentage yet. It may or may not be random. But it’s certainly not at the point where its predictable.
Also, the stuff on HR to flyball rate in the annual is really good. Great reading, especially for those who have been touting HR rates of potential acquisitions as a strength.
So, Dave, what you are really saying is that Felix got lucky and is not likely to be as good at “preventing” LDs next year. Maybe this is why his MLB ERA was better than his AAA ERA.
If the IBB numbers conflate NL & AL statistics, that probably explains a lot of the difference between regular BBs & IBBs. With an IBB you’re just about always pitching to a worse hitter than the one you walked–that’s true for either league. But if you’re walking the #8 hitter to get to the pitcher with 2 out in the National League, particularly early in the game when a pinch hitter is unlikely, that’s got to skew the results and hence the predictive value of the IBB for both leagues–low for the AL & high for the NL.
Dave,
Line drive percentage is not repeatable…
Yet, reading over you list of pitchers posting the top “effective percentage” is like reading over a list of “who’s who” in pitching. Is this simply a list of the pitchers that induce ground balls and control k/bb the best? Just by looking at the results, it appears that pitchers are affecting the fourth factor in your equation also.
So, Dave, what you are really saying is that Felix got lucky and is not likely to be as good at “preventing†LDs next year. Maybe this is why his MLB ERA was better than his AAA ERA.
He also cut his walk rate significantly from what it was in Tacoma. Those two factors were the keys, yes. Even if his LD% goes up a bit, as long as his GB%, K%, and BB% stay somewhere near where they were last year, he’s going to be dominant.
We know that some hitters are better LD hitters
hitters have a very weak correlation on LD% as well (0.10).
Ok, so we don’t know this…lol.
It is so funny to me that we (baseball fans) talk about batters being a certain type of hitter: good line drive and gap presence, sprays the ball around the field, use the gaps to drive the ball…
We are so full of it. I can’t wait for the book.
It makes sense that IBB are less harmful than UBB, if only because IBB are given specifically when they can do the least harm.
The HBP immediately spawns a few hypotheses.
1) Clearly the pitch numbers don’t matter in that small a range, even it it does take more pitches to BB than HBP, HBP is worse.
2) HBP requires (usually) a bigger miss. This could be indicative of significant fatigue.
3) Because a HBP is such a big miss, it could mess with the pitcher. You can walk a guy and still think you were making good (or at least fair) pitches. A HBP requires that you really suck for one pitch. That could cause a significant loss of confidence, if only for a few pitches.
4) The next batter following a HBP is more likely to be facing a new pitcher, one who may not have been warming up. This strikes me as a minor factor.
5) The next batter following a HBP may face a pitcher who’s afraid to come inside, and he’ll know it.
What’s an infield fly? That’s just an out, right? If any ball hits the infield dirt, whether it goes above a fielders head or not, is considered a ground ball, right? What’s a home run considered? A fly ball or a line drive or both/depends? Fascinating numbers, but they seem too wrapped up in context to be valuable for hitters – for example, HBPs’ results.
If these data include homeruns, then each is probably classified based on how it was hit, rather than where it landed.
Same goes for the infield flies. These aren’t scoring decisions – these are decriptions of where the ball went.
Well, I’ve never seen anything I’d consider an infield fly that wasn’t an out, or an error.
If 11% of all fly balls go over the wall, then why is the expected value of a fly ball so low? Shouldn’t it be closer to .11, at least (assuming every one in the park is an out, and there is never anyone on base)?
Are these numbers from 2005 or from a larger sample?
If it’s just from 2005, I’m pretty sure that the HBP average run value is within at least 2 standard deviations of the UIBB average run value. I’d have to know the exact breakdown of how often a HBP increased the run value by N runs to get the exact standard deviation, but it’s not too hard to get close with an estimate.
The smallest the standard deviation could be is if the only two possibilities were for HBP to increase the average run value by 1 or to not change it at all. This gives a standard deviation of 0.47. There were 1797 HBP last season, so that would make the standard deviation of the sample mean 0.011 using the minimum uncertainty of 0.47. I’d say that the actual standard deviation in the average run value is probably closer to 0.02. This puts it pretty safely within 2 standard deviations of the UIBB average run value. (And assuming this is just 2005 data, that 2 in the third decimal place is surely not significant.)
Considering that UIBB is such a similar event to HBP, I’m pretty comfortable chalking that difference up to statistical uncertainty.
Brownie –
That explains why the value of an infield fly is so close to that of a K. It’s almost always an out, and it almost never advances a runner. It also prevents strike’em out-throw’em out double plays, too, pushing its value up very slightly.
Your flyball/HR analysis assumes the value of a HR is 1 (it’s actually better than that because it plates all the runs on base), and the the value of an out is 0 (when it’s actually negative because it uses an out, just like a K).
That’s really interesting data Dave.
I’ll go out and grab a copy just based on that table, but until I do…any chance you could throw percentages into your table for the 2005 M’s staff not named Felix?
Okay all we need now are stats of type of bat used when the ball is in play, and type of glove used to catch the ball when the ball is in play Anything else?
Why even consider the infield fly when there is the infield fly rule? I think that is one unneeded stat because umps will always call it an out regadless if it is caught or not. And I’m not talking about the exceptions (so few) of the dunderhead players who can’t decide who should catch the ball.
“Why even consider the infield fly when there is the infield fly rule?”
Because the infield fly rule isn’t called all the time?
Are there any bookstores in the Seattle area that will be carrying the HT Annual? I’m not totally against online purchasing, but it’s always a pain receiving packages at my place, so I prefer buying in person when it’s convenient. Thanks.
I’m fairly certain that the THT Annual will be carried at all major bookstores. Call your local Barnes and Noble and see if they have any.
Dave,
Wasn’t a supposed ability to limit line drives a possible cause for knuckleballers’ (and Jamie Moyer’s) ability to limit BABIP to a range lower than that for the league as a whole? Are we back to square one as to why Charlie Hough continually kept his BABIP down?
fascinating…
I think this data shows pretty clearly that inducing a groundball or a flyball is not the goal of a contact-oriented pitcher. The goal is to induce weak contact. The data shows a pretty clear difference between a weak popup and a towering scorcher that smacks against the very top of the center field wall, and I’m sure a pitcher would tell you that the execution of his pitch will have a strong influence on the outcome. I’d be interested to see a weak-contact / BIP average for pitchers. I’m sure it would help to separate those pitchers who achieved a “lucky” low BABIP and those who achieved a low BABIP through good pitch selection and execution.
Sure, a infield fly rule isn’t called all the time, but I bet you it is called more often then not. I don’t think the RUN Value stat takes that into consideration. I can’t imagine the infield fly rule not being called more then 20% (although it seems like 10%). BTW, does anyone have the actual ratio of infield fly: infield fly rule called? It would also be interesting if anyone knew why an ump wouldn’t call it? And the last question: how high does the ball have to travel to be considered a fly ball? I’m thinking of those times when the M’s tried to bunt only to have it be caught before it hit the ground, and those hits that just land over the head of the pitcher, behind the mound.
Sure, a infield fly rule isn’t called all the time, but I bet you it is called more often then not. I don’t think the RUN Value stat takes that into consideration. I can’t imagine the infield fly rule not being called more then 20% (although it seems like 10%).
Not being called more than 20. So you think it’s 30? 40? 50? 60? 70? 80? 90? 100?
Do you realize it requires at least one runner on to be called?
Do you realize that pitchers throw with the bases empty about 60% of the time?
Unless umpires are calling infield fly when there are no runners on, and doing so a lot, the upper limit of infield fly calls is 40% which means in 60% of the cases IT STILL WOULD NOT BE CALLED WHICH IS WHY IT IS WORTH KEEPING TRACK OF.
Getting stats that good took me about 20s. Maybe next time you could do it yourself, eh?
it requires two runners on (and less than two outs)…if the batter doesn’t run to first on an infield fly, that’s his problem and the rule won’t bail him out. It’s to prevent a double play since the two runners couldn’t advance if the ball is caught. Thus, the rule will be called for, guessing here, 5%-10% or so of IF fly balls. At most.
Ray, on the off chance you’re not a troll, go here:
http://ask.yahoo.com/20040716.html