Correlation, Causation, and Beat Writers
(Caution: Really Long Post Ahead. Get a cup of coffee and make sure you have some time.)
When I look back on the 2007 season, there’s going to be memories of positive developments and frustrations alike. We had Felix in Boston, and we’ve had Felix almost everywhere else. We had Kenji Johjima’s Pull Power Swing Of Doom, and we’ve had Turbo’s Double Play Special. But, interestingly enough, one of my favorite things about the 2007 Mariner season will not have occurred on the field or even involve any members of the roster.
Instead, one of the things that has made this season enjoyable (most days) to cover has been the ability to read Geoff Baker in the Seattle Times everyday. Now, we weren’t the biggest Bob Finnigan fans in the world, so pretty much anyone would have been an upgrade as the Times beat writer for the Mariners, but Baker has delivered above and beyond what we’ve come to expect from the local writers. The blog that he has maintained has been a source of terrific information and entertaining reading on a near daily basis, and he’s shown a passion for his job that, frankly, we’re not used to seeing, even posting multi-thousand word blog entries on his days off.
Baker has embraced the ability to interact with the fans and has shown what a traditional journalist can do with the power of today’s internet intertwined. It’s been fun to read, and honestly, quite refreshing to look forward to reading the Times sports page again.
So, it’s with respect for the work he’s done this year and his willingness to open himself up to new ideas and outside opinions that I feel compelled to respond to his latest blog entry, entitled “Mound of Trouble”. Geoff – you can do better than this.
The premise of the post is completely accurate – the M’s pitching, especially the rotation, has been their downfall. That’s obviously true. No one’s going to argue that point. But, well, the analysis that led to the conclusion leaves a lot to be desired. Let’s look at the specific points raised.
To further the discussion of this morning, I received a very interesting email today from a friend of this blog, Jack Lattemann, who has done an exhaustive study of whether teams with an earned run average of 4.50 or higher can even post winning records, let alone contend for a playoff spot. Jack has graciously allowed me to pass on his findings. They don’t look good on the Mariners, who have a 4.84 team ERA despite a rock solid bullpen. He found that no team before 1969 had qualified for the playoffs. Not surprising, given the two-league format. There were a few more, post-1969, that made it. During the two-division format (four teams making the playoffs) from 1969-1993, the only playoff team with a 4.50 ERA or higher was the 1987 World Series champion Minnesota Twins, who finished 85-77 with a 4.63 ERA.
Where to start with this paragraph – how about with the glaring, obvious problem, and one that I’ve been railing on for years here – Earned Run Average, by itself, is not any real indicator of pitching quality. It’s just not. I know it’s commonly accepted as the be-all, end-all pitching statistic, but the reliance on this inherently problematic stat has led to more bad analysis over the years than just about any other statistic out there. Using ERA to draw broad conclusions about pitching ability is a great way to be wrong on a large scale.
In reality, ERA kinda sorta measures the ability of the team’s run prevention skills when a specific pitcher is on the hill. ERA doesn’t attempt to separate responsibility for said run prevention between pitcher and defenders. It doesn’t attempt to take into account the context of the run scoring environment. And, just in case those weren’t big enough problems (they are), it introduces the biases of ballpark specific official scorers by excluding “unearned runs”, which are often classified as such due to arbitrary decisions on what constitutes an error.
Baker presents this research as evidence that teams with “bad pitching” don’t make the playoffs. Unfortunately, there’s no explanation for why we should accept a 4.50 ERA as a marker for “bad pitching” in any context. As Baker notes, not one team made the playoffs before 1969 with an ERA that high, while he does recognize that the fact that, before the league championship series creation, only two teams made the playoffs, making the point essentially irrelevant.
Unfortunately, he doesn’t also point out that the run scoring environment in, say, 1917, was massively different than the run scoring environment today. The Philadelphia Athletics had the worst team ERA in the American League 90 years ago. Their ERA? 3.27. The league average was a nifty 2.66. The A’s ERA+, adjusted for ballpark and era, was 84, meaning that their pitching staff was 16% worse than the league average that season. You know who has an ERA+ of 84 this year – the Texas Rangers, with their 5.48 ERA, have also been 16% below average, once you account for the fact that today’s game is massively different than 1917’s and the fact that the Rangers play half their games in a park that is highly condusive to offense.
Now, this is an extreme example. However, the point remains – picking a random ERA number that reflects “bad pitching” and applying it to any context is going to result in a list that means absolutely nothing. If you want to use ERA to evaluate a pitching staff, you’d be forced to come to the conclusion that the Washington Nationals currently have a better pitching staff than the Chicago White Sox. After all, they have a lower ERA. Of course, everyone understands that there’s a huge difference between pitching in RFK stadium against National League hitters and not facing the DH and facing American League hitters in New Comiskey park. We wouldn’t expect Mike Bacsik to post a 4.59 ERA if he was traded to the White Sox. No one would.
So, unfortunately, the only conclusion that you can draw from the fact that the ’87 Twins were the only team to make the playoffs with a 4.50 ERA or higher before 1993 is that the game has changed dramatically as time has gone on. Using that selected data to make any inferences about the importance of pitching in terms of a team making the playoffs is simply a misuse of statistics. It’s a conclusion that can’t be supported by the data provided. It’s the kind of thing that people who hate statistics point to when they throw out their “lies, damn lies, and statistics” cliches. Essentially, this entire opening paragraph doesn’t tell us anything useful about the 2007 Seattle Mariners. Moving on…
Now, during the wild-card era, things have changed. So far, there have been 13 teams make it with an ERA of 4.50 or more. But don’t forget, that’s out of 96 playoff teams. So, basically, even with the diluted pitching and inflated ERA totals we’ve seen over the past decade or so, there’s still only about a one in 7.5 chance of getting there.
Now, here, Baker does outright state that “things have changed”, so he obviously understands everything I just wrote above. He just didn’t do a particularly good job of explaining the limitations of the data in his first paragraph. Here, though, we’re presented with a new data problem – 13 out of 96 teams that have made the playoffs since 1995 have had an ERA of greater than 4.50, so Geoff turns that into a “one in 7.5 chance” for a team with “bad pitching” to make the playoffs.
Unfortunately, the flaws of the data won’t let you make that claim, either. The 1996 Texas Rangers are one of the 13 teams that posted a 4.50 ERA or higher and still made the playoffs, finishing 90-72 and winning the AL West. They scored 928 runs and allowed 790, so the immediate reaction is that they slugged their way into the playoffs and overcame their poor pitching staff.
Except, that’s not true. The Rangers 928 runs ranked 5th in the A.L. that year, while their team ERA ranked 6th. Toss in the fact that The Ballpark in Arlington is a hitters haven, and it becomes clear that run prevention, not run scoring, was the strength of that team. Their ERA+ was 109, meaning that they allowed 9% less runs than an average team would, considering their home park and the run environment of the American League in 1996.
I could go through this process many more times, but I think you guys get the point. Even the 13 out of 96 number really doesn’t tell us anything about the respective merits of the pitching staffs of playoff teams or the importance of having a quality pitching staff in terms of winning your division. It just doesn’t. Again, moving on.
So, yes. It really does come down to the pitching. There have been 74 teams post winning records since 1921 with an ERA above 4.50. The Mariners did it several times in the late 1990s. One of the teams on Lattemann’s list is the 2003 Blue Jays, a squad I feel epitomizes the need to understand how good starting pitching will always be more valuable than hitting.
Argh. That last sentence was almost assuredly written with the intent of having me throw myself off a bridge. Okay, probably not, but if I read it too many times, I might do so anyways.
Geoff goes on to extrapolate about how the ’03 Blue Jays show that you can score a lot of runs, but if you don’t have enough quality starters, it won’t matter. He ends his Blue Jay rabbit-trail with this paragraph:
The point of this is, the Jays had the AL’s second-best offense. They had a guy who should have been an MVP in Delgado. They had the Cy Young winner. And they still won only 86 games.
Problem #1 – the Blue Jays didn’t have the AL’s second best offense. They did finish second in the league in runs scored, but the Skydome is a nifty place to hit. It’s not Texas, but it’s one of the higher run scoring environments in baseball. When you account for context, the Blue Jays had an OPS+ of 107, meaning that their offense was 7% better than league average. The ’03 Red Sox (118 OPS+) and Yankees (117 OPS+) both blow the Blue Jays out of the water, and while the Blue Jays did have the 3rd best offense in the AL that year, they were closer to 9th place than they were to 2nd place.
The 2003 Blue Jays were good but not great at scoring runs and average at preventing them (ERA+ of 100, the epitome of league average). Being good at one aspect of the game and average at the other is rarely a recipe for a playoff team, regardless of whether your strength is offense or defense. Using an example from the same year, the 2003 Chicago White Sox posted an ERA+ of 108. Essentially, they were as good at preventing runs as the ’03 Blue Jays were at scoring them. Their ERA ranked 4th best in the league despite pitching in a good hitters park. They, also, only won 86 games and didn’t make the playoffs? Why? They didn’t score enough runs.
Geoff points to the ’03 Blue Jays as evidence that “good starting pitching will always be more valuable than good hitting”. However, he doesn’t mention the ’06 Los Angeles Angels, a team that ranked 3rd in the A.L. in ERA and had a tremendous rotation of Lackey-Escobar-Weaver-Santana, and failed to make the playoffs because they employed a bunch of terrible hitters in their line-up.
Now, Geoff’s not going out on a limb here. It’s established baseball doctrine that starting pitching is the most valuable commodity in the game, that good pitching beats good hitting, that pitching and defense win championships, etc… You’ve heard all this before. Unfortunately, it’s just. not. true.
There have been numerous studies done on the issue of whether teams should value run prevention over run scoring. Needless to say, these studies go just a bit further than sorting by ERA and counting the number of playoff squads under an arbitrarily selected number. The consensus is that, in general, a team that is strong in preventing runs will do slightly better than a team that is equally strong in scoring runs. The breakdown is accepted at about 52/48, meaning that stopping the opponent from scoring is approximately 2% more important than putting runs on the board yourself.
The problem, however, is that run prevention isn’t 100% pitching. I’ve been beating this drum for a couple of years, now, and I’ll keep beating it until people begin to realize that defense matters. An awful lot. The Detroit Tigers gloves carried them to the World Series last year. They had an average offense, an average pitching staff, and the best defense in baseball. And it got them the American League pennant.
Run prevention is probably about 52% of winning, but pitching is, at most, 80% of run prevention. The ability of the other eight guys on the field cannot be ignored, and credit cannot be simply handed to the pitcher for the strength of his teammates. In terms of value towards wins, the breakdown is close to something like this:
Offense: 48%
Pitching: 41%
Defense: 11%
No matter how you slice it, Geoff is simply wrong when he writes the following:
In other words, as you can see, it really is all about pitching. Especially starting pitching. Always has been, always will be. Through the dead-ball era, the Depression era, WWII, the onset of free-agency, expansion, advent of the wild-card, the Moneyball era and all the other so-called periods of change this game has gone through. One thing remains constant. Right now, the M’s don’t have that one thing. And they don’t have it in a division where two teams, the Angels and A’s, are loaded at starting pitching. Just hammering the point home.
It’s just demonstrably not true. It’s not. If you’re in the mood for some very mathy and hard to read evidence, here’s a table that essentially sums up the actual performances of historical teams based on their run scoring and run prevention, and adjusts for run scoring environments to give apples-to-apples comparisons. But, basically, as the chart shows, there have been winning teams that were great at hitting and not so good at pitching, just as there were winning teams that were great at pitching and not so good at hitting. In reality, pitching isn’t more important than offense. Never has been, and never will be.
Okay, that would be the end of this entirely too long post about a common misconception from a guy we like, but he just had to threw in one more statement that I can’t leave alone, just for good measure.
Do the Mariners have an MVP candidate? Uh, no.
Sorry, but they absolutely do. Ichiro Suzuki is hitting .356/.409/.466 while playing half his games in an extreme low scoring environment by today’s standards. He’s 19 for 21 in stolen bases. He’s 3rd in the major leagues in Value Over Replacement, which Keith Woolner (now working for the Cleveland Indians) created as a total measure of offensive value, adjusting for positional scarcity. In terms of VORP, Ichiro has been worth 36 runs to the Mariners offensively so far, ranking ahead of David Ortiz (35.2), Vladimir Guerrero (35.0), and Jorge Posada (32.3). The only guys in the majors who have contributed more offensively to their teams are Magglio Ordonez and Alex Rodriguez, who are both about 10 runs ahead of Ichiro based on their fantastic power displays to date.
But, when you factor in defense, Ichiro closes that gap in a hurry.
You can call me a biased fanboy if you want, but the data is on my side – through June 18th, Ichiro has been every bit as valuable as any other player in baseball. Even if you don’t think Ichiro’s glove makes up for the ten run advantage that Ordonez and Rodriguez have created with their bats to date, you simply can’t argue that there’s anyone else in the American League who should rank #3 on your AL MVP ballot right now. And, well, if the #3 guy in the league isn’t an MVP candidate, then your standards are pretty freaking high.
You know why the Mariners are currently 35-31 despite having a disaster of a starting rotation? Their center fielder is absolutely awesome.
Okay, that’s enough writing for today. Geoff, we still like you, and we’re still thrilled you’re in Seattle, and I look forward to being entertained and informed by your writing in the many days ahead. On these issues, though, I couldn’t disagree more.
Comments
134 Responses to “Correlation, Causation, and Beat Writers”
Leave a Reply
You must be logged in to post a comment.
What about Sizemore and his lack of D ability I really don’t trust those D numbers.
100: That ’97 team was a blast. Every game was high anxiety – one of the best offenses ever, along with one of the worst bullpens ever. I wonder how many times we scored 10+ runs. At least the rotation (from what I recall) was pretty good. I believe that was Griffey’s one and only MVP season.
In other words, as you can see, it really is all about pitching.
And all this time I’ve been under the impression that winning baseball was about scoring more runs than your opponent does.
I guess this means we’ll have to really break the bank for some free agent starters this winter!
A big part of the 4.79 ERA was the park, but yeah that bullpen was god-awful, and yeah that offense was incredible.
#103, yeah, us and every other team (breaking the bank for starters). I think it’s the stupidest thing. You’re right scoring more runs than your opponents wins games. I think it’s stupid what teams are doing because pitching is WAY more volitile than hitting, you have a fair idea of how a hitter is going to do in a park, age, etc, with pitchers, people lose their stars quickly, and the unexpected Joes take their places way too often to justify making FA starting pitching a priority. I haven’t looked at the numbers, but I bet teams get burned worse on starting pitching than on offense and defense because of the inherent volatility.
Well, the volatility and the overpayment starters seem to get. That doesn’t mean I wouldn’t root for the M’s going after a star pitcher.
It’s too bad the M’s didn’t take a chance on that other dreadful Cardinals starter from last year (Marquis). Funny how his season is about the polar opposite of Weaver’s. I am glad Zito opted for San Fran instead of Seattle, as his best days seem behind him – and at the tender young age of 29. Good lord does Zito walk a lot of guys (40 BB in 83 2/3 IP).
For now I’d be happy if the M’s sign Tomo Ohko for the remainder of the season. As badly as he pitched for ths Jays this year he couldn’t do any worse than our band of merry tossers. Oh yeah, and release HoRam.
*Tomo Ohka, that is.
Geoff has weighed back in on the subject
#101 — Given Sizemore’s youth and power, he is probably more valuable over the long term, if not now, than Ichiro.
#107 and 108 — Okha was signed by the Cardinals. I think I read this on mlbtraderumors.com.
I like Geoff, but he missed the point again. Oh well. He’s not going anywhere, so we’ll turn him around eventually.
#97
I would imagine Ichiro’s home and road stats are so close is due to the fact that most of his hits are either on the ground or soft flies that land in front of the center fielder. Neither of those events are influenced as much by home park as say if he were trying to crank every pitch over the fence. I think that same nature of his hits is also reflected in his doubles rate.
True. But I think Dave’s point is that this kind of hitter is more valuable to the Mariners (though, perhaps not as much as a left power hitter) than other teams because it isn’t depressed by the confines of SAFECO. A more subtle point, but I think is quite important.
I’ve toyed with the numbers and while Ichiro is the same on the road as at home, it’s like netting 3-5 runs compared to what another team would gain because of the lack of a home/road split. He may not be 3-5 runs better because of Safeco, but that’s the value to the M’s. He’s simply worth more, and I mean more as in wins, to the M’s than any other team.
You know, if Geoff would just use ERA+, he could make his point within the constraints of his argument.
Sigh. Well, better than the alternate universe where Finnigan would be all Griffey all the time this week.
Geoff’s point is almost valid. The M’s offense is good, but it’s not good enough to carry such an awful rotation. His conclusion and analysis lacks though because he only looks at ERA, but the fact that the M’s need better pitching peformance to get into the playoffs is valid. This is the point where we could really use Jarod Washburn pretending to be Jamie Moyer at least until the break (when we can try and pick up a pitcher). I know that’s not likely, but one could only wish.
Yea, that’s the frustrating thing – Geoff’s mostly right, but he’s just using bad analysis to prove it.
If he just posted that “teams with bad pitching generally don’t win”, no one would disagree with him.
I hate Win Shares. No stat that gives players more or less credit based on the performance of their teammates should ever carry any weight in a serious discussion.
Since Win Shares are a measure of contribution to actual team wins, if the team doesn’t win any games, no player on the team is assigned any win shares for that period, regardless of how well he played.
Geoff probably feels bad for this Lattemann fellow, since he used his name and analysis for his post – which was really torn apart here. Seems like he’s stuck in a spot where he has to defend some false conclusions in order to save face for his friend.
We’ll get Geoff turned on to WPA in lieu of Win Shares. It’s exactly what he wants to measure (hits that matter, high leverage performance, etc…), and it’s 100 times better than Win Shares.
What other stats should we use in analyzing pitchers? I do think era is overrated.
I like BAA (batting average against). Am I also a moron?
http://ussmariner.com/2006/08/29/evaluating-pitcher-talent/
I thought the chief park effects for Safeco are that it suppresses both home runs AND singles, but provides a boost to doubles and triples.
Ichiro’s numbers don’t support that, but he’s a bloody freak.
I don’t like leverage, either. Not until someone demonstrates that performance in high-leverage situations is a skill. Otherwise you’re just rewarding luck.
In the context of an MVP discussion, though, it carries some weight. Vlad hit a walkoff home run the other day off Chris Ray that was worth .65 wins. With a single swing of the bat, he turned a likely loss into a win.
That happened, and it had a tangible effect on the Angels record. It held a lot of value to the team.
He might not be able to keep hitting walkoff home runs every time he bats in the bottom of the 9th, but it doesn’t diminish the real contribution that he made with that swing. And, in terms of an MVP discussion, I can see the perspective that you should reward players for events that actually occurred.
I’m not sure I would vote that way, but I think there’s merit in the reasoning of those who would.
It’s when people try to use WPA or Win Shares or similar things to evaluate talent that it becomes a problem.
I think someday in the future WPA will be useful for HoF voting, for example if someone outperforms their expected WPA consistently, that might serve as a shot in the arm for borderline cases. I think in ~ 15 years we should have a much better idea of which (current not past) players deserve HoF since we can look at their career via things like WPA and measure their career defense with things like UZR.
For hitters Win Shares are nifty, as long as you ignore Fielding Win Shares, that’s where the value of winshares figuratively comes off it’s wheels while driving down the road.
Okay, I get what you’re saying. Since you’re looking for value, you’re measuring anything that helped the team, regardless of whether is was the result of skill or luck.
I can see why people would want to count that in their MVP debate. I just think those people are wrong. It’s like giving credit to drunk drivers who don’t happen to hurt anyone.
My one issue with some of these newer stats is that no one is attempting to run them on older seasons to see if the quantitative data is reflecting the qualitative perception.
I’m reminded of how Bill James gave Craig Biggio insane value in the Historical Abstract, but then you look at Fielding Win Shares and see how he overvalued middle infield defense and wonder if Biggio was benefiting too much from that bump.
Which stats are you referring to, dw? Win Probability is a mathmatical formula. It’s just win expectancy put into practice based on current run scoring environments.
no one is attempting to run them on older seasons to see if the quantitative data is reflecting the qualitative perception.
No one?
I would propose that bad pitching causes high ERA, therefore making it a reliable stat for judging a team’s season pitching.*not individuals – sample size*
ERA+ would be better because it adjusts for park effects, but Baker is writing this article for the common fan, justifying the use of a rough stat.
4.50 seems arbitrary, but it’s a run every other inning, making it easy and understandable to use as a marker.
RA is better than ERA, especially at the team level, but also at the individual level.
re Dave’s comment 111: seems like you’re starting to succeed, see Baker’s latest blog entry.