## Matt Garza’s Crazy Start

When Matt Garza was acquired by the Chicago Cubs this off-season, he was expected to provide stability to a rotation with some question marks. Garza’s first start with the Cubs was anything but stable, as he likely posted one of the strangest pitching lines we will see this season. Over 7 innings, Garza managed to post 12 strikeouts while allowing 12 hits and no walks. His performance in the game actually led to Garza posting a FIP of -0.48. (Since we are dealing with really small samples here, it’s important to note that this isn’t terribly uncommon early in the season. Still, it’s kind of cool to have a negative FIP, right?) The line is so unique, however, because it’s tough to understand how one pitcher can be so hittable/un-hittable in the same game. Let’s take a closer look at how this could have happened?

As Albert Lyu reminded us in January, Matt Garza has become an extreme fly ball pitcher in recent years. While that type of approach will generally lead to fewer hits and a depressed BABIP, it also leaves pitchers susceptible to home runs. On Sunday, however, you would have sworn there was a different pitcher on the mound.

According to his FanGraphs Game Log; Garza gave up four fly balls, nine ground balls and seven line drives in his start. While this is typically an unusual stat line for Garza, it goes a long way in explaining his performance. The 12 hits Garza allowed can be directly attributed to the amount of grounders and line drives he gave up. Since we expect pitchers with higher GB% to post higher a higher BABIP, and we expect line drives to be hit really freakin hard, it’s easy to see why Garza would give up so many hits in this particular start. That doesn’t necessarily explain why Garza was so hittable, though.

Again, Garza’s Game Log helps shed some light on this situation. Of the 106 pitches Garza threw on Sunday, 80 of them were strikes. This approach affected Garza in two very different ways. Since he pounded the strike zone early and often, he was able to get ahead of hitters and rack up strikeouts at a high frequency. However, that same approach also led to more pitches in the strike zone for the Pirates (obviously), which meant more hittable pitches. Because Garza was pounding the zone, opposing hitters could be aggressive at the plate knowing he was going to throw a strike.

While it doesn’t fully explain the nuances behind Garza’s crazy pitching line, his Game Log at least provides some insight into how he arrived at the final result. We can’t come to conclusions about every aspect of his start (like how he was able to induce so many ground balls), but we spot a correlation between his approach and his stat line. Based on his career numbers, however, we shouldn’t expect Garza to post a similar line any time soon. This is one of those starts where you just sit back and marvel at the absurdity of the final line. You’re not going to see this one very often.

Print This Post

Can someone explain how FIP can be negative? I guess I misunderstand that stat – I would’ve thought 0.00 is the minimum. Does giving up a ton of hits somehow drive your FIP down?

Part of the equation is subtracting 3.25 or something like that. I’m too lazy to look it up. There’s a constant in there that you always subtract.

So then would any stat with a ton of Ks and no BBs have a negative FIP?

@ Ellis:

For Garza, his FIP would look like this:

(((13 x 0 home runs) + (3 x 0 walks) – (2 x 12 strikeouts)) / 7 innings) + ~3.0 = -0.48

As the season progresses, it will be in the positives for good.

People smarter than I should be answering your question, but let me take a stab at it.

First, I would refer you to Alex Remington’s excellent article on FIP.

http://sports.yahoo.com/mlb/blog/big_league_stew/post/Everything-you-always-wanted-to-know-about-FIP?urn=mlb-206286

He lays out the simple formula, which tells you how FIP is calculated. As you can see, hits aren’t actually part of the equation since FIP focuses on the things the pitcher has absolute control over.

Since Garza gave up 0 home runs, and walked 0 batters while racking up 12 Ks, you can see how that would throw off the equation. As Jack Weiland mentions, it probably has something to do with the constant that Alex mentions in his article.

You are right, a good fielding independent statistic would have a minimum of zero.

The fact that it can be negative is an argument for being wary of FIP. Simply put, it is a little hacked together in the way all advanced saber stats are a little hacked together.

Ks enter negatively, homers and walks enter positively. Intentional walks do not count, I guess lacking confidence doesn’t count against you.

BTW, it is specious to say that more grounders become hits. 25% of grounders and 25% of fly balls become hits. Its just that FG defines a homer as not a hit. Go figure.

it is specious to say that more grounders become hits.I don’t think Chris said that — he said that grounders have a higher BABIP, which is true. And of course, if Garza had given up a home run, that would’ve been factored into his FIP for the game.

If you like, you can think of the question as “How did Garza give up 12 singles while striking out 12 and walking none?”; ground balls are certainly more likely to produce singles.

haha, this guy is so bitter about everything.

“You are right, a good fielding independent statistic would have a minimum of zero.”

You really can’t infer anything with regard to FIP (or any metric) from a single start, similarly you can’t infer anything about the quality of any metric which gets its results from a single start. If Matt Garza ends up with a negative FIP at the end of the season, or heck even by the All Star Break, then you may have a valid point.

My thoughts exactly. I was listening to the game and didn’t even realize what was happening until I looked at the box later.

12 ks? 12 hits? NO walks? For Matt Garza? Whaaaaa?

hey he’s really good, he threw a no-hitter

Not noted here is that Garza allowed 12 SINGLES. Any other pitcher and we’d be emphasizing the luck factor here, wouldn’t we? Instead of talking about how “hittable” he was, which you have no way of knowing from this stat line (which is sort of why BABIP is favored by this particular community). Imagine a Halladay start with the same line. SOs? Check. Low BB? Check. No XBHs? Check. “Man, that was the most UNLUCKY start ever!”

I’m not sure we would if Halladay gave up 7 line drives. That’s not exactly good.

Last season Halladay had 5 starts where he gave up at least 7 line drives:

http://www.fangraphs.com/statsd.aspx?playerid=1303&position=P&season=2010

It is more accurate to say that it’s not terribly uncommon for a good pitcher to give up that many balls in play counted as line drives.

Nearly half of the hits were just perfectly good luck – you can’t see that from a box score, but while watching the game it was obvious Garza was dominant and the Pirates hitters kept getting very lucky to reach base with singles.

I was cursing at the tv for most of this game with all the dinky hits they got. The two most egregious ones I remember.

1. Diaz hit a little nubber in the right side hole. Pena was holding a runner at first and it got into the outfield.

2. Someone hit a ball right at shortshop immediately after Castro ran to cover the man running into second.

It was just a very weird/frustrating game to watch. Nobody every mashed a ball off of him, just a bunch of well placed ground balls and soft line drives.

“…it’s tough to understand how one pitcher can be so hittable/un-hittable in the same game. Let’s take a closer look at how this could have happened?”

one explanation would be that half the opposing lineup was really good and the other half sucked.

Nope, Pirates.

The Pirates have 4 truly excellent young hitters: Tabata, McCutchen, Walker, and Alvarez. 3 of those guys are over .340 through 6 games (and last year had wOBAs of .334, .363, .351, and .343). The other 4 position players are actually not-awful, but there’s certainly a dropoff in terms of talent.

At any rate, if you don’t know that the Pirates have 4 good young hitters, then you don’t know anything about them, and shouldn’t waste anyone’s time by commenting on them.

“3 of those guys are over .340 through 6 games”

SSS warning.

Lol @JRoth: “The Pirates have 4 truly excellent young hitters: Tabata, McCutchen, Walker, and Alvarez.”

Agreed on McCutchen, but we’ve got different ideas on the bar for “excellence.” Alvarez might get there, and the other two are probably above average as well, but I’d put them well short of excellent.

““…it’s tough to understand how one pitcher can be so hittable/un-hittable in the same game.”

No, its not tough to explain at all.

The simple answer is that he wasn’t that unhittable. The simple answer is that the only way he could get outs is via K. When your defense is terrible, it drives up K/9. Which is why K/9 is a terrible stat (K/PA is defense independant, K/IP is not).

“The simple answer is that he wasn’t that unhittable. The simple answer is that the only way he could get outs is via K.”

Actually, the exact opposite it true if we are to follow your logic.

Hitters posted a 96.8% Contact rate on pitches swung at in the zone, and plastered him for 12 hits. They had no problem with the Fastball, no problem with the Curve and no problem with the Changeup. Now, so why did the Pirates strike out so much if he was so hittable? Chasing Sliders out of the zone. We are talking about a fairly-below average hitting club who was having no problem at all putting wood on the ball on 3 of his 4 pitches otherwise.

The K’s Garza compiled are more a result of the Pirates hitters mistakes and shortcomings then his being on his game. If he were facing a team like the Yankees though… well, the outcome probably would have looked similar to his 9/14 or 9/20/2010 outings

So do the special Matt Garza only rules now include SOs so long as they were outside the strike zone? I thought SOs are the one thing pitchers can control, but I guess that’s only if you’re not Matt Garza. Anything written about Garza on this site from now on should be titled “Special Matt Garza Rules” followed by a subtitle like the title of this article.

“thought SOs are the one thing pitchers can control,”

Clearly not.

” Garza managed to post 12 strikeouts while allowing 12 hits and no walks. His performance in the game actually led to Garza posting a FIP of -0.48. ”

Which points out the biggest glaring flaw in FIP. FIP thinks its better for a pitcher to give up a hit than it is for a pitcher to get an out via any means other than a K.

…except it doesn’t. As explained earlier, hits don’t factor into the FIP equation at all.

But there’s also an opportunity cost to the other out. If you get an out without getting a K, you can’t get that out with a K.

Imagine two pitchers (or maybe it’s the same pitcher, with two different defenses) who alternate Ks and balls in play — but all of one pitcher’s balls in play turn into outs, and all the others go for hits. The one who gives up the hits will have a perfect FIP, since they’ll have 27 strikeouts/9 innings. The one who gets the outs will only have 13.5 strikeouts/9 innings. So in this case, FIP is punishing you for getting outs on balls in play.

Of course, if you get an out on a ball in play, it also gives you fewer opportunities to give up walks and HRs, so for below-average pitchers getting outs on balls in play will drive their FIP down. Still, I can see an argument that the denominator for FIP should be batters faced rather than innings pitched.

“all the others go for hits” s/b “all the other pitcher’s balls in play go for hits”

Matt, I do like the idea of using batters faced or perhaps go for percentage numbers instead K/9 or IP based numbers.

Well… to be precise, hits do factor into the FIP equation. The FIP equation (as explained in the glossary here) includes IP, and allowing hits (high BABIP) essentially depresses IP in proportion to the other FIP categories (HR, BB, K).

The formula in the glossary is (HR*13+TBB*3-K*2)/IP + X, where TBB is the actual number of free passes given (BB-IBB+HBP) and X is the cosmological constant.

So… if your (HR*13+TBB*3-K*2) is positive, depressing IP inflates FIP. If your (HR*13+TBB*3-K*2) is negative, depressing IP deflates FIP. There should be plenty of pitchers on each side of the divide every year if I have these numbers right in my head.

That’s sort of a weird result, ‘eh? Even if it’s a pretty minor effect overall, which I think it is. I think that, combined with FIP’s ability to go negative, shows that FIP isn’t a perfect formula, that it has problems with extreme inputs.

Of course, no formula is perfect, and I don’t think anyone ever claimed FIP was perfect. Only that it predicts future ERA better than ERA does for most pitchers. It’s still a very useful formula.

In FIP, a hit is a neutral event. It does not effect that stat at all. An out via anything other than a K increases your FIP because it lowers K/9. IE, an out is a negative event.

Hence, giving up hits is better than getting outs.

RC, you’re oversimplifying the effect of hits on FIP by ignoring the effect of hits on HR/9 and BB/9. In Garza’s case, giving up no walks and no homers, yes, hits help his FIP. I think on average more pitchers will be hurt than helped by hits over time.

Overall, I agree with several posters that FIP would be slightly better off using PA in the denominator instead of IP. It would be a pretty small improvement for extreme cases, and it would probably make it slightly harder to eyeball the formula.

The bigger issue for extreme cases like Garza’s is that FIP is a linear formula and the real relationship between the events encompassed and expected scoring is not linear. Most pitchers’ performances over the course of a season appear to fall in ranges where the linear approximation holds… so FIP should be a fine measure for a season of MLB starting pitching from typical MLB-caliber starters. But with extreme inputs it falls apart. Extreme rates can come from small sample sizes, as in Garza’s start. But I’d also perhaps be wary of FIP for pitchers whose rate stats are off the charts, like Carlos Marmol.

“Hence, giving up hits is better than getting outs.”

Lets throw in some numbers just to compare the impact. Lets start with a pitcher with 200IP and 9.00 H9, 1.00 HR9, 3.5 [BB-IBB+HBP]9 and 7.0 SO9. Assuming a 3.20 run constant, I get a FIP of 4.24 for this guy.

Lets tack on 10 events-

If we add 10 HRA, his FIP jumps to 4.89, +0.65

If we add 10 non-intentional walks or HBP, his FIP goes to 4.39, +0.15

If we add 10 non-HR hits, his FIP stays the same, +0.00

If we add 10 non-SO outs, his FIP goes to 4.22, -0.02

If we add 10 SO, 4.12, -0.12

While, of course you’d rather record outs than give up hits and FIP says the opposite, the difference when you aren’t trading off dozens or hundreds of non-SO outs for non-HR hits is so minimal to FIP that it isn’t really worth worrying about a whole lot.

I confused myself there… the 10 extra non-SO outs, did lower FIP while the 10 extra non-HR hits kept it the same, so FIP *does* treat a non-SO out as being more valuable than a hit…

Except you’re completely ignoring the fact that none of these are independant variables. Adding non-HR hits typically will lower a pitcher’s IP, because of pitch counts and runs. Adding non-K outs will do the opposite. The results are a lot more drastic than you’re making them out to be.

If a pitcher faces 30 batters, and gives up the following line:

10 K

3 BB

17BIP – 5H, 11outs – (BABIP .294)

His FIP is 0.45

Take 3 of those BAP, and turn them into outs, and his FIP is 1.0. Except thats not actually how it works, because with less baserunners, the 2nd pitcher is going to be going after hitters more, and most likely have more balls put into play, which will reduce Ks, and increase the IP.

If you want a Fielding independant stat, these two pitchers should come out exactly the same. A fielding independant stat shouldn’t have a (negative) correlation with BABIP.

RC, you did the math wrong. Basic arithmetic. I don’t know what the cosmological FIP constant is for this hypothetical year, but supposing it’s X, FIP in the first case is X + (2*10 – 3*3)/(22/3) = X – 42/22 because 22 outs were recorded. In the second case the numerator is the same but the denominator is 25 because 25 outs were recorded. So it’s X – 42/25. The difference between these FIP numbers is about .3, the first being a little better. That’s for two hypothetical cases of an excellent pitching performance (FIP dependence on BABIP is greater when the absolute value of the numerator in the formula is greater, which occurs in excellent and poor performances) with a very large BABIP swing.

FIP is an estimator for ERA using a linear combination of three components and nice whole numbers so it’s easy to calculate. It correlates with reality pretty well within ranges of inputs that are typical for starting pitchers over long periods of time. It’s easy to come up with single-game results that make it look silly because that’s not what it’s for.

“Imagine two pitchers (or maybe it’s the same pitcher, with two different defenses) who alternate Ks and balls in play — but all of one pitcher’s balls in play turn into outs, and all the others go for hits. The one who gives up the hits will have a perfect FIP, since they’ll have 27 strikeouts/9 innings. The one who gets the outs will only have 13.5 strikeouts/9 innings. So in this case, FIP is punishing you for getting outs on balls in play.”

But can’t you perform such silly exercises on most stats?

Lets say a player can focus on contact and get groundball singles in one of three of his ABs [and GB outs otherise]. Lets say the same player, taking a different approach, can swing harder and go more for flyballs and instead bat .100, with 1 HR every 5 AB and all outs otherwise.

Batting average and OBP say he wass much worse, but I think we can agree that over 600AB, 200 singles and 400 outs is still less valuable than 120HR and 480 outs — no matter what AVG and OBP haveto say about it.

But, more to the point, the baseball databank database has HR, BB, SO, HBP and IBB going back to 1955 and from then until now, 23790 players have recorded atleast one out in a MLB season and NONE have gone below a 1.20 FIP [using 3.20 as the constant]. With atleast FIVE innings, none below 2.00. With atleast 110 innings, only Pedro Martinez has even been below 2.80 [1999 and 2001]. That said, I think using a crazy result after one start as a reason to not use FIP is just plain ridiculous…

“That said, I think using a crazy result after one start as a reason to not use FIP is just plain ridiculous…”

When the components of an aggregate aren’t descriptive, is the aggregat?

Why not? In one particular inning a batter might hit two grand slams [Tatis]. Is the fact that he got almost 2% of his career HRs and RBIs in about 0.03% of the innings he batted in really that relevant compared to his aggregates?

That was nothing more than a small sample fluke. Hell, go bigger sample, he had 1HR per 15.8 AB in 1999 and one per 31.8 AB for the rest of his career. Does this somehow invalidate the value of HR rates?

If the stat said Tatis had struck out 4 times while coming to the plate twice, would you believe it? Because thats closer to what we’re talking about here.

FIP is giving a non-sensical answer.

“Based on his career numbers, however, we shouldn’t expect Garza to post a similar line any time soon. This is one of those starts where you just sit back and marvel at the absurdity of the final line. You’re not going to see this one very often.”

I don’t think you should expect anyone to post a similar line like that. According to BR play index, there’s 109 games where a pitcher gave up 10 or more hits, had 10 or more K’s and 1 or less walks from 1919 to 2011, so a little more then one per year.

And for those that watched, at least 3-4 of those singles weren’t just BABIP luck, they were close your eyes and throw a dart luck and happen to hit the bullseye luck. Diaz got jammed so bad on one fastball, but it magically dribbled between first and second, there was a check swing hit and run that went through the vacated SS spot and Ramirez fell down ranging to his left and couldn’t make a throw that they called a single (off the top of my head).

fwiw, if you add XBH’s at 1 or less to the criteria you only get 18 games ever

the last being Randy Johnson on 8/9/00

if you go H > 9, K > 9, BB=0, XBH=0, you get 3 games ever

Garza’s, one on 6/25/83 and on 9/5/81

I don’t pay for their service so I can’t get the full results.

so yes it’s unique for Garza and don’t expect it to happen again. Just don’t expect it to happen for any pitcher either.

To have a negative FIP with no walks or HRs allowed, you would need more than 14.4 SO/9… who is this pitcher who is going to do that? For every 1.0 BB/9 you add, you’d have to also add 1.5 SO/9 on top of the 14.4.

For every 0.1 HR/9 you add in, you need to also tack on another 0.6 SO/9. So, a dominant pitcher with 0.5 HR/9 and 2.5 BB/9 rates would need to manage over 21 SO/9 to get down to a negative FIP! I’ll hold my nreath while we wait for this guy to show up :)

It was a weird start. As I said to a friend, Garza got BABIPed to death. A lot of the hits were groundballs that could have been snagged, they just found a hole. It happens in baseball. The Pirates are a MLB team filled mostly with players that would fit somewhere on almost all 29 other MLB teams. Ditto for the Cubs. Anything that this start says about the Cubs defense or the Pirates offense is pretty much meaningless so far.

It was a fascinating game to watch with a strange and interesting pitching line for Garza. It COULD portend a different pitching style for Garza. Chris Cwik’s points were spot-on. Some of the commenters…not as much.

====================================

“one explanation would be that half the opposing lineup was really good and the other half sucked.”

“The Pirates have 4 truly excellent young hitters: Tabata, McCutchen, Walker, and Alvarez.”

“Agreed on McCutchen, but we’ve got different ideas on the bar for “excellence.”

===================================

guys, McCutchen didnt even play in the game in question. not a single AB.

http://www.fangraphs.com/boxscore.aspx?date=2011-04-03&team=Cubs&dh=0

so even mentioning him in a discussion regarding how good/bad the Pirates lineup is, as being a reason for the high # of K’s is really pointless.

So the reason for Garza’s line — getting a lot of K’s and also giving up hits — was because he was throwing a lot of pitches in the strike zone? That’s earth shattering science right there.