Comments on: Smoltz = Weaver (The Good One)

By: Andrew

Andrew — Tue, 25 Aug 2009 08:27:46 +0000

http://www.baseball-reference.com/players/split.cgi?n1=martipe02&year=2008&t=p#innng – Yes I understand it’s a small sample size, but doesn’t the fact that the average batter that Pedro faced in the 5th turned into Barry Bonds tell you something? I can’t think of any other good pitchers to examine the splits of bur Pedro is in kind of a similar boat as Smoltz so I feel he’s the best comparison.

It is way too tedious to look at pitch f/x, but I will say as your arm fatigues you lose command and serve up fat pitches. That’s what makes someone like Halladay such a horse; he doesn’t seem to get fatigued. I agree pitchers can’t control their luck but they can definitely affect their opponents OPS, whether by not giving out walks or LD doubles sent down the middle of the plate,

yway I wouldn’t mind continuing this discussion over email so we don’t clog the thread up too much more. My email address is under the website link.

By: Nick

Nick — Tue, 25 Aug 2009 07:40:14 +0000

I shouldn’t say not predictive, but predictive to varying degrees. So a 140 inning sample, in which maybe 1/3 or those reveals the split in question, will only be, say, 20% predictive, meaning you would need to regress heavily to the mean. Over 5-6 years? You probably need very little regression.

At any rate, I took a look at Pedro’s velocity by inning in last year:

http://www.hardballtimes.com/images/uploads/speed_plott_thumb.jpg

And he doesn’t appear to have lost much velocity. He had a slight rise in the 4th, and a slight drop in the 6th, but nothing so severe as to explain his splits. I would take a look at movement, but it’s a lot more tedioius.

By: Nick

Nick — Tue, 25 Aug 2009 07:04:56 +0000

Andrew, trust me on this one. Just because the player is old, doesn’t mean you need less of a sample size.

If you have other qualitative data, such as pitch f/x, that shows that Smoltz or Martinez loses velocity or movement in later innings, then yes, you could say it’s repeatable. Otherwise, no.

And splits really, really, reallly aren’t predictive unless they are back up by other data or are in a HUGE sample size.

By: Andrew

Andrew — Tue, 25 Aug 2009 05:17:25 +0000

As I said, those numbers were from two seasons of Pedro pitching for the Mets. It’s not a small sample size. It’s a recognizable pattern: old guy with a ton of injuries gets hammered as his pitch count rises and he goes later in the game. Smoltz’s start today was great, but they were smart in keeping him at 75 pitches.

I mean, this is a pattern with elite pitchers and relievers too. If you leave them in too long opponents’ OPS rises with their pitch/inning count. Guys who are 40+ and have had a million surgeries obviously have less stamina so it comes into play earlier.

I’m not disputing that he has the talent to be a 5th starter and I would put money on him pitching better than Pedro has the rest of the season. However, no SP is like the pitching robot I linked to earlier. Splits are definitely real. Look at Dan Haren’s career pre- and post- ASB numbers: 651 IP 3.08 ERA before and 520 IP 4.19 ERA after. He has been predictably bad 5 seasons in a row after the ASB. Is he just unlucky?

By: Nick

Nick — Mon, 24 Aug 2009 22:47:27 +0000

It might, I’m not a statistician. It’s better to look at pitch f/x data for this kind of stuff, so that you can find a reason and not be a slave to the numbers. Smoltz’s velo hasn’t declined much and neither has his movement in the later innings. I would say that it’s probably just luck.

By: Andrew

Andrew — Mon, 24 Aug 2009 17:00:30 +0000

Would it mean anything that those numbers are from two seasons of Pedro Martinez? So like 140 innings.

By: Nick

Nick — Mon, 24 Aug 2009 08:29:56 +0000

Smoltz has not had a pronounced decrease in velocity and movement in the later innings.

By: Nick

Nick — Mon, 24 Aug 2009 08:29:00 +0000

Andrew (I’m a different Nick), those inning by inning splits are likely meaningless to the fact that they come in a small sample size, and are polluted by his high BABIP and low strand rate.

I am sure that pitchers do have differing abilities to control timing, however, it takes a lot more than 40 innings to determine that.

By: chris

chris — Sat, 22 Aug 2009 19:10:14 +0000

thanks debbie downer, you’re annoying

By: Joe R

Joe R — Fri, 21 Aug 2009 15:00:37 +0000

To be fair, Playoff race and Halladay haven’t exactly gone hand in hand over the years.

And like I said before (despite Nick’s insistence that we all think Smoltz sucks despite his xFIP and how we’re all idiots blah blah blah), Smoltz clearly can still get outs. He can still be a 7-8 K/9 guy. But seeing him hit the wall over and over again after a few innings tells me that he’s more suited for a middle relief/late inning role than a starting role at this stage of his career. Is that really that convoversial?