Buster Olney strikes again

by dwintheiser on May 24, 2008 7:00 PM EDT

Joe Christiansen passes along a tidbit from Buster Olney's ESPN blog, where Olney wonders if Johan Santana, at age 29, is already past his prime.

Personally, I've never been much of a fan of Olney, despite his very high standing in the baseball writing community. Even so, I can't take much offense at his conclusion, as passed along by Christiansen:

This question remains: Will he be worth to the Mets what they will pay him over the duration of the contract? We’ll see.

Well, for starters, I can't really take much offense because there isn't much here to take offense to -- Olney doesn't even have the confidence in his analysis to make a definite statement that Santana either will or won't be worth his contract at the end of his seven year deal. The strange thing is, though, if Olney did say that Santana wouldn't be worth his contract at the end, I couldn't really argue with him: while we're able to project careers better than we have at any other time in baseball history, and we're in an era where ballplayers are playing effectively much longer into their careers than they normally have, it's impossible to say with any confidence precisely how Santana will be pitching six years from now, which is the main reason why the Twins were so reluctant to sign him to the long-term deal he wanted. It doesn't matter who'd make that argument, be it Olney, Aaron Gleeman, or Stephen A. Smith -- if you have to choose 'will he or won't he be worth the money at the end', the odds are he probably won't.

The problem I have with Olney isn't so much his writing, or even his conclusions -- it's that he sometimes tries to be a newer-model sportswriter, aware of and even able to use statistical and sabermetric analysis in his work, yet he almost always ends up botching the analysis because Olney tends to use statistics, to borrow a phrase, the way a drunk uses a lamppost -- for support, not illumination. (For example, consider Olney's season-and-a-half long quest to get Productive Outs accepted in the baseball and sabermetric communities, despite almost universal disdain for the stat in the latter community.)

Olney provides two pieces of information -- one traditional, one sabermetric -- to back up his assertion that Santana ain't what he used to be. The traditional information is that Santana's fastball is slipping in velocity; an observation that Olney notes has been confirmed by other major league scouts who have written him since he first published that Santana was losing zip on his fastball. I can't argue with that observation; I'm not a big-league scout and I don't have access to a radar gun (unless you count the one on Gameday webcasts that seems to be ridiculously out of calibration at times).

The other piece of information Olney supplies is a comparison of Santana's numbers from this season as compared to previous seasons, and that's something I can make an argument against. Olney shows the following numbers...

With that said, here are the primary indicators:

Opponents’ OPS over the past six seasons: .607, .642, .564, .594, .616, .678. This year: .723.

His ratio of strikeouts per nine innings over the past six seasons has been 11.38, 9.61, 10.46, 9.25, 9.44, 9.66. This year: 7.79.

His strikeout-to-walk ratio over the past six seasons: 2.80, 3.60, 4.91, 5.29, 5.21, 4.52. This year: 3.87

...and clearly wants you to believe that they are conclusive. They certainly look damning, to say the least. Of course, if you're a Twins fan who's watched Santana over the past few years, you may have a tickling little doubt at the back of your brain, as I did: hasn't Santana historically been less than himself in April and May, and wouldn't quoting this season's numbers be just quoting numbers from April and May and comparing them to full seasons' worth of numbers where we know Santana has pitched better?

Yes. Yes it would be. In Santana's career prior to 2008, he's been 25-19 with an ERA of just over 4.00 from the start of the season through May 31; from June 1 to the end of the season, he's 73-28 with an overall ERA of less than 3.00.

So how do Santana's numbers so far this year compare to his overall numbers in his career, looking just at April and May?

Opponent's OPS:
This season - .723
Overall April/May, not counting 2008 - .747

Santana's K/9I:
This season - 7.79
Overall April/May, not counting 2008 - 9.4

Santana's K/BB:
This season - 3.87
Overall April/May, not counting 2008 - 3.59

So, this season's OPS is actually better than his career average for this time of year, while his K/BB ratio is better than his career average. The K/9I ratio is clearly down, which might be related to the observational drop in Santana's velocity; on the other hand, the overall numbers include all of Santana's relief appearances, where a pitcher's K/9I ratio tends to be higher. (In the interest of fairness, it would take a lot of work that I'm just not up to right now to separate all of Santana's relief appearances from the overall numbers, and removing every season where Santana made no more than one start in a month actually causes his career K/9I ratio to go up, not down.)

So Olney's watered-down conclusion is impossible to argue with, but would be hard to argue with even if he took the obvious stance, and his observation that Santana's velocity is down does seem to be validated by his K/9I numbers compared to his career numbers during the same period of the season. On the other hand, Olney's statistical argument that this means Santana may be in decline at age 29? Much less credible. After all, Santana's ERA in 2008 is also lower than his career average in March through May (though ideally you'd adjust both sets of numbers for park effects and league effects and possibly even use a defense-independent ERA measure, it's still clear that Santana isn't pitching significantly worse by an ERA measure).

It seems much more credible that each of the following is true:

- As an East Coast writer, Olney never really paid much attention to Santana until it was clear that Santana was in the Cy Young race, and thus either never noticed his tendency to start slowly or never made much of it.

- Olney looked at the numbers, found a few that supported his argument,
and simply stopped looking.

Have I ever made an argument, looked for a few numbers to support my position, and stopped looking? Sure -- I'd guess we all have. Then again, most of us aren't being paid to write credible opinions, nor are we likely to be quoted by Joe Christiansen even if we do.

Then again, it is his blog, not an essay for the Journal of the Society for American Baseball Research, so maybe I should cut him some slack.

Tweet Comment 10 comments | 2 recs |

Do you like this story?

Comments

What about pitching in the NL?

I tend to agree that one needs to compare apples to apples in terms of Santana’s slow starts.

OTOH, they do have a documented drop in velocity. That’s something tangible. Also, he is pitching in the NL. Remember that, without the DH and pitchers hitting, everyone expected Santana’s K numbers and other peripherals to jump. Apparently, they haven’t.

This could just be more of Santana starting slow, but I have a hunch that Santana will be merely excellent instead of great this year. He tends to throw a lot of pitches. His fastball is hittable, and if you lay off the change, or can foul it off, you can make him sweat.

What happens down the road is anyone’s guess, but I think the Twins made a good trade, and I am content to struggle through the ups and downs of a young staff, trying to find consistent quality starts.

Meanwhile, I celebrate Santana’s victories, when they come, and quietly revel in his struggles, knowing that we have Gomez and a trio of hurlers laboring away somewhere not too far from the big leagues.

by Old Twins Cap on May 24, 2008 11:48 PM EDT reply actions

pitching

According to the numbers that dwintheiser posted, his peripherals HAVE jumped (with the exception of k/9). Santana has started pretty well, and has had a couple of rough (by his standards) starts lately but we all know to expect that from him this time of year.

That said, all the other stuff you said about him is true (which is why I also support the trade). But I do think that he will get on a roll this summer (as will the Mets) and he will be at the top of they Cy Young voting.

The really puzzling thing about Buster Olney’s story is that he is trying to argue that Santana might not be worth all that money by looking at current performance. Santana is a great pitcher right now, we all know that. A drop in fastball velocity is the only thing that he can point to that might actually suggest that Santana is already getting worse (and once again he needs to compare it with velocity in April/May, maybe they are the same, I don’t know). But as you said, his fastball is already hittable. As long as he can locate it does it matter? A drop in velocity won’t hurt his change up.

The reason the contract is bad is because he will be earning it when he is 36, not because of how he is pitching now (which as everyone has probably noticed is quite well).

http://noblingblings.blogspot.com/

by Aaron Fix on May 25, 2008 1:21 AM EDT up reply actions

Olney

He’s just trying to justify himself. He made a comment in his blog last year as well. He pointed out that it appeared Santana lost velocity last year after his 18 strikeout performance against Texas. His reasoning was that for the first time the Twins staff allowed Santana to pitch well over the 100 pitch count.
Anyways, I think it’s all ridiculous. Santana is a consummate professional and he knows his body and his stuff better than 99% of the pitchers out there. The guys a workout freak and probably has the best 5 day routine out there. That’s why the Twins weren’t really ever tempted to pitch him on short rest. Last year it was pretty clear that Santana shut it down a bit early because there was no need to push it going into a big contract situation year. Further, this year, as dwintheiser points out, we’re only two months in. They haven’t even started to see nasty yet in New York. Secondly, there are noticeable differences in National league style of play. I for one, fully expected the strikeouts to go down. There tends to be quite a few more “scrapiness” in the national league. In other words a lot more guys just trying to make contact as opposed to titan lineups like the White Sox where 1-9 is trying to take you deep on big swings.

Gardenhire's major league career: Banjo hitting, futility infielder who couldn't lick it.
Rick Anderson's major league career: Strikethrower who never made it happen with his sub 90's fastball.
Really gives a new definition to living vicariously through other people, don't it?

by caseintheface on May 25, 2008 9:09 AM EDT reply actions

I saw olneys comments

and agree that judgment should be witheld until santana has put in a full year in the NL. Even though he should have better numbers in the NL, there is still the issue of getting to know the hitters.

I wonder why he doesnt throw his slider much anymore. MY guess is that that has as much to do with his ‘decline’ as anything. As dominant as his changeup is, if guys just sit fastball for a less than dominant fastball, most major leaguers can handle a fastball

by guinness junky on May 25, 2008 12:04 PM EDT reply actions

throwing fastballs and changeups

if he does that almost exclusively, he should be able to maintain his durability and extend his career. throwing lots of sliders just puts added pressure on the arm. since Santana has basically the same arm action with his fastball and changeup he’s basically throwing the same pitch 100 times a game. i think that’s one of the keys to his durability. maybe guys continue to figure him out once in a while. but seriously, if he has a 3.00-3.50 ERA and 200+innings over the life of the contract, isn’t that really getting your money’s worth?

by caseintheface on May 25, 2008 12:51 PM EDT reply actions

Productive outs

I’m done analyzing this deal. It’s clear to me that Santana is not worth to the Twins what the Mets are paying him right now. Hence, he won’t be worth to the Twins what the Mets pay him over six or seven years. And there’s no way the Twins could get two players of the caliber of Gomez and Guerra through the draft. Letting Santana walk and get draft choices was the only viable alternative. This way, the Twins get much better talent and an extra year to develop it.

But I do take issue with the comment on productive outs. For the record, I think the sabermetric community is wrong here. The assumption, based on the data, is that all outs are the same. But how can a strikeout be the same as a ground ball to the second baseman with a runner on second and no one out? How many times have you personally seen the runner on second get stranded because the batter did not move him up?

I haven’t analyzed the flaw in the science, but there must be a flaw. It’s like Descartes: Nobody has been able to refute his dream argument. Yet we know prima facia that our external reality is not merely a dream. No one has been able to refute the argument that all outs are the same, but we know prima facia that some types of outs are more valuable than others. The paradigm case is the sac fly that would not have happened if preceded by a strikeout rather than a weak grounder to second.

I wouldn’t chide Olney for failing to endorse the sabermetrics with the faith of a true convert. Sabermetrics is a work in progress. I would liken it to Newtonian Mechanics, which was a gross oversimplification of reality that worked most of the time. Sabermetrics works most of the time. But it is necessarily oversimplifies a complex game. It is arrogant to claim that if the science doesn’t account for the phenomena, the phenomena doesn’t exist.

"You're thinking too much. Just have fun." -- Bennie "The Jet" Rodriguez in Sandlot

by cmathewson on May 25, 2008 1:02 PM EDT reply actions

Productive Outs

But I do take issue with the comment on productive outs. For the record, I think the sabermetric community is wrong here. The assumption, based on the data, is that all outs are the same.

I don’t think that’s the assumption, because it’s obviously wrong. In any given situation, if you know you’re going to make an out, there’s clearly a better and worse way to make that out.

The problem with Productive Outs is that it goes from the premise that, if you’re going to make an out you’d better make it a good out. Ultimately, while there are better and worse outs, there’s really no such thing as a ‘good out’.

Let’s take a look at the two most common situations in which people want to talk about Productive Outs:

1) Runner on second, no outs.

The idea is, if you’re not going to drive the runner in, at least you want to move the runner to third so the next batter can get him in even with a fly out. And moving the runner over to third with an out does increase the odds of scoring a run in the inning, but not by as much as you’d think—right now I can only find a base-run index for the 1999 season, so if someone can find an index with a larger data sample I’d appreciate knowing if the ‘99 data is an unrealistic sample.

With a runner on second and no out, you have a 61% chance of scoring at least one run in the inning, and you’ll average 1.14 runs in every inning in which this situation occurs. Sacrificing the runner to third increases the chance of scoring at least one run to 65%—an improvement, but not that large of one. Meanwhile, your average runs decline to .99 runs per inning in which this situation occurs. Each seven such sacrifices costs your team a run, on average.

That’s not to say the move is always a bad move—in specific situations, increasing the chance to score by 4% can be the difference between a win and an extra-inning game. But doing this in general is hurting the ballclub, and since Productive Outs doesn’t distinguish between a #9 hitter sacrificing in the top of the 8th and a #2 hitter sacrificing in the bottom of the fourth, it’s not the tool that’s going to tell you who gets this job done most often.

2) Runner on first, no outs.

Here’s the other sacrifice situation, though in this case you’re tryiing to avoid the double play as much as you’re trying to increase your odds of scoring. Which turns out to be a good thing, because the sacrifice doesn’t actually increase your odds of scoring at all:

Runner on first, nobody out, a team will score at least one run 42% of the time, and will average .89 runs per inning. Runner on second, one out, a team will score at least one run 40% of the time, and will average .67 runs per inning. Five such sacrifices, and you’ve cost your team a run on average.

Well, you might say, at least you avoided the double play, which would have killed the inning. Except the averages with runner on first and nobody out already include the actual double plays which did occur in that situation, and you still end up scoring more runs overall if you don’t sacrifice.

The only other situation I can think of where an out could be described as productive is runner on third, less than two out, and he scores on a fly to the outfield. We already have a stat for that situation, though—the sacrifice fly.

Productive Outs, as formulated by Olney, aren’t a worthy stat because they don’t measure something that’s actually positive.

by dwintheiser on May 25, 2008 2:52 PM EDT up reply actions

You actually just proved that productive outs exist.

Using the stats you provided, moving a runner over from 2nd with no outs via groundball to the right side increases the chances of scoring, albeit minimally from 61% to 65%. However, as you cite in your next scenario (runner on first no outs)...”Runner on second, one out, a team will score at least one run 40% of the time, and will average .67 runs per inning.”

In other words, the inability to move the runner to third means your scoring chances suffer a 25% decline. And your average runs per inning shrinks from .99/inning to .67/inning.

by dontchange on May 25, 2008 6:38 PM EDT up reply actions

OK

That’s not how most of the sabermetric mavens have explained it to me. It’s always been an absolute thing: There’s no value in outs, productive or otherwise. But I don’t see baseball as a matter of absolutes. The value of an outcome rests on a continuum, from home run to triple play. All the scenarios in which the player does not make an out are on one side of the continuum, all the scenarios in which the player makes an out are on the other. Sacrificing and otherwise moving the runner into scoring position are closer to the center of the continuum than grounding into a double play, for example. And, as such, they have some value.

Anyway, that’s all a long-winded way of saying that to deny that productive outs exist (which is how it has been presented to me countless times) is just an oversimplification. Of course, you would prefer not to make an out. But hitting is not so easy. The pitcher is trying hard to get you out, and it is a zero sum game. If the pitcher does his job, he will get you to try to get a hit, and in so doing, you actually make a worse out, like grounding to the shortstop instead of the second baseman with nobody out and a runner at second. The odds of getting a hit are much lower than hitting a weak grounder to the second baseman. So teams play the odds, trying to make it easier on the hitters to at least make some positive contribution to scoring.

What I just wrote is obvious to anybody who’s ever played the game of baseball. For some reason, sabermatricians delight in concocting arguments that prove that obvious stuff doesn’t exist. In so doing, they are not engaging those who would rather not question the obvious. Certain of the sabermetric arguments have made their way into the mainstream, such as FIP and OPS. This one will never make it. And if I must believe every tenet of sabermetrics to be taken seriously by that community, I will forever be that drunk leaning on a lampost you talked about.

"You're thinking too much. Just have fun." -- Bennie "The Jet" Rodriguez in Sandlot

by cmathewson on May 26, 2008 12:03 AM EDT reply actions

the argument

As far as I can tell, the argument that SABR is generally pursuing is that sacrificing should be viewed in a very critical light, and not put in place as part of your general game plan. Most reasonable folk would not argue that sacrificing a runner to 3rd with your 9-hole hitter in the 8th inning of a tie game is a bad idea outright. That’s why we have managers that we pay the big bucks to make those decisions based on all the other factors in that particular game that are pretty well impossible to analyze using current models.

More than anything, I see the efforts of SABR types in this context (as in most) as a method to objectively analyze the game in such a way that conventional wisdom isn’t merely accepted because “that’s how Whitey Ford did it in ‘63” or whatever.

Also, I find base-run indexes interesting, but accepted as gospel by way too many people. It’s a neat shorthand tool, but I don’t see them as any more precise than using, for instance, OPS or WHIP by themselves to evaluate a player.

formerly known in these parts as adamb

by ravenfly on May 27, 2008 2:24 PM EDT up reply actions