Or "Fooling around with numbers to get nowhere"
Fangraphs did a nice little article yesterday on how Gio Gonzalez absolutely destroys the opposing team's pitchers when they come up to bat. This is something we noticed last year (and now I'm kicking myself for not following up) and can by itself explain why Gio took a step up to the top tier last season. As the author says, this is simply demonstrative, and it isn't meant to be dismissive of Gio. He pitches in the National League. You have to pitch to pitchers. To do that better than anyone else is a very good skill to have.
It did raise a question for me though. What if Gio pitched better to bad hitters and worse to good ones?
Now I hear you. OF COURSE that is going to be the case. Better hitters are better. Worse hitters are worse. But what I'm saying is what if he dominates the bad hitters more than the usual pitcher and gets hit harder by the good hitters than the usual pitcher as compared to their own baseline pitching. In that case Gio would bode poorly as a playoff pitcher (in a very broad general sense and against his own high level - it's not like you'd rather have Wang out there) because you usually face better hitters in the playoffs and while every pitcher would get hit a little harder, he'd tend to get hit even a little harder than that. You follow?
So I compared the baseline OPS given up by an NL pitcher last year to the OPS given up by batting order position and the did the same for Gio. Did I find anything to be worried about? Maybe...but probably not.
Overall, unsuprisingly the best hitters are 3rd, 4th, 5th and 6th. (although this is bad planning because your 6th hitter will get up less than your 1st or 2nd - it's not really killing anything. As we've gone over, lineup construction has a surprisingly small effect on scoring). Gio has about the same ratio of "BOP OPS" (batting order position OPS) to overall OPS as the NL as a whole
3rd NL : 1.14 Gio : 1.10
5th NL : 1.05 Gio : 1.02
6th NL : 1.04 Gio : 0.98
Clean-up hitters do hit Gio harder than his baseline in comparison to the rest of the NL (NL : 1.13, Gio : 1.27) but given what we see above it doesn't seem indicative of anything. (A Note here - Gio's baseline of .582 is so much better than the NL's of .721 that even though #4 hitters hit him relatively harder, they still hit him a lot less than they do the rest of the NL. OPS NL : .818 OPS Gio : .739)
What about the rest of the lineup? Well here's where things are a little kooky. Gio dominates pitchers like none other but he also is a LOT better at facing #2 and #7 hitters.
2nd NL : 1.00 Gio : 0.74
7th NL : 1.01 Gio : 0.70
9th NL : 0.64 Gio : 0.39
But he's a LOT worse at getting out leadoff hitters and #8 guys.
1st NL : 0.98 Gio : 1.46
8th NL : 0.94 Gio : 1.20
What the hell does this all mean? Well I'm not sure. I could say that he's bearing down on the pitcher spot so much that he's actually hurting himself in the 8th and 1st spots. But then what's up with #2 and #7?
Really I don't want to say anything yet. There is some noise we have to get rid of here. Are there more lefties in the 2nd and 7th spots than in your average spots? More righties in leadoff and 8th spots? Are these spots specifically bad/good in the NL East? That could explain some of it. Or it could be just yearly fluctuation. Each position is roughly 1/9th of the batters Gio faces. That's not even a month worth of pitching. You wouldn't (well... you shouldn't) get overly excited by a month of good/bad pitching, so there's no reason to get excited by what we see here. If you asked me right now, that's what I would assume right now. It's nothing. A fluke of the limited data.
This is something to look at over the course of the year. It'll probably go away, or at least be really diminished by the time the season finishes up. If not, well then we can debate what it actually means and if whatever we decide really means anything in a practical sense because Gio's baseline is much better than the NL's baseline.
Subscribe to:
Post Comments (Atom)
11 comments:
How many of those #8 hitters were faced with 2 out already? Maybe, knowing he can get the pitcher, Gio pitches too carefully to #8's. I mean, if were delving into potentially meaningless stats, lets delve all the way!
Chaos
Chaos - yep there's another one. And maybe he doesn't do well against #1 because he gets the pitchers out so much he's facing them in a lot of 1 out or 2 out and no one on situations. (of course where did all those #8s go?) Lots of noise here. It's always useful to question though even if you get no answer.
Harper -- I think you could get further here by doing two things:
1) for each batter Gio has faced (regardless of spot in order), compare wRC+ (or whatever method you choose) vs Gio against the same method against other pitchers (say that year or whatever time period seems appropriate). Obviously you'd need to split the metric into bins (otherwise you'd just find that Gio is better than the average pitcher). You could probably play around a little bit with the bin choice to make sure that things aren't arbitrary.
2) By getting around the batting order spot metric, you then can, say, include data from the past few years when Gion was in the AL and get a larger sample size. You can also include league and year fixed effects to control somewhat for the variation.
If you did this, I think you could get a much tighter estimate which you could be more confident about.
Matt - Oh so you actually want to answer the question I posed to start I see. What are you? Some sort of smart guy?
I'm not sure - the admittedly very cursory attempt showed that for the 3rd-6th batting positions, Gio didn't suffer any worse than any other pitcher. That makes me hesitant to put the effort into looking further.
Rather than bins though - I think I'd chart gio's performance and test his correlation vs the league's (which should be -1 I believe if I'm thinking of this right) Well actually we should chart first and see if there are any obvious inflection points. (Is it called an inflection point? We're not changing concavity sign - we'd just be moving from zero...)
gah I'm getting distracted. Anyway I felt that more than the "Gio hurt more by better hitters" what came out of the cursory analysis was "Gio might have some funny Batting order tendencies" so if anything I'd look at that - but then again that doesn't hold up to the AL given the different status of the 9-hole.
You might have to check more pitchers to see if similar funny tendencies occur. It's possible that the #7 spot is more often filled by a free swinger hoping to drive in runs before the lineup descends to the #8 and #9 guys. Espinosa pops to mind as the prototype. The #8 guy tends to be more of a cautious, obp type to give the pitcher an opportunity to bunt.
In any case, it might be interesting to see if someone like Gio, who has a good FB and devastating curve differs from someone like Detwiler who pounds the same pitch. Maybe Gio approaches the #1, #2, #7 and #8 guys differently based on his arsenal. I will agree with Chaos, though in that if I were Gio and knew I could strike out the pitcher at will, I might be a bit more cautious to the #8 guy.
I was told that there would be no math on this exam....
We need the regular season to start so that you have something to write about.
He'd probably be facing the #1 guy the most, that 4th at bat might be an issue the more often he does it.
Baseball logic implies there would be more lefties hitting second to take advantage of the hole opened up with the leadoff guy on first.
I agree with the #8 rationale, he's already a guy who walks some folks.
Harper --
1) makes sense to me.
2) I went with bins because of your batting order approach. It also makes it easier to put in fixed effects which help control for changes in setting. The obvious bins are really above/below median or perhaps top quartile vs bottom quartile. I wouldn't go further than that to avoid data mining.
3) This makes sense. The problem is, as you say, that you can't incorporate AL stats into this so easily. I guess the alternative is wait a few years when Gio has more NL stats, then go from there.
Ah, I see now how Gio's baseline split metric bins could have more qualitative inflection points inspite of the inverse relationship of his AL sample verses a clearly verifiable concavity sign influence, which would CLEARLY affect his cobalt axiom of the warble theory. But only to #8 hitting lefties in a two out situation with runners at first.
Although I understand none of what you guys are talking about I do comprendo with "There is some noise we have to get rid of here."
...geh...
Seriously, interesting post Harper even though it made my head hurt. Also good timing as Gio rocked it last night. Also, how about Detweiler's line on Saturday as well?
Donald - he's compared against all NL pitchers here so I think that covers "more pitchers" but there could be something to say about checking all the Nats or a few lefties to see if there a pattern.
the cautiousness to the 8 guy is interesting.
Sec 314 - odd because it was a math exam. Well, if I told you that then I told you that. Class dismissed! ("School's Out for the Summer" starts playing)
blovy8 - another interesting point, but then why does that #2 guy get dominated? Lefty like you say? maybe....noise noise noise. I'm like the Grinch here.
Matt -
2) Agreed. I got nothing against bins, but since this isn't serious work I'm looking for the easy way out. I'm an idea guy. I see the macro level. Other cliches for people that don't want to do actual work.
Froggy - You make some great points. Where do you teach advanced stats?
Greetings from Carolina! I'm bored at work so I decided to check out your website on my iphone during lunch break. I love the info you provide here and can't wait to take a look when I get home.
I'm surprised at how fast your blog loaded on my mobile .. I'm not even using WIFI, just 3G .
. Anyways, wonderful blog!
My web-site: roulette online
Post a Comment