There is one of our despised phrases, and one of the Twins’ favorites and it’s “pitching to contact.” The Twins use this to justify their affinity of pitchers that don’t walk anyone, but unfortunately also fail to strike out many hitters. We’re seeing that phrase quite a bit this offseason, as the Twins have just signed RHP Jason Marquis to fill out the rotation, and he’s essentially a Carl Pavano or Nick Blackburn clone. Those three pitchers are going to go the chuck-’n’-duck route, whereas the other two starters, Francisco Liriano and Scott Baker, will actually be able to strike out some hitters.

Last season, there was some controversy when the Twins expressed they wanted Francisco Liriano to “pitch to contact” more. Here we had a starter whose main weapon was getting hitters to swing-and-miss, and yet the Twins wanted him to stray from that strategy. It was a comment that most of us (including myself) didn’t take to very well. However, beneath the surface, the Twins were trying to say that they wanted Liriano to reduce his number of pitches in an outing by getting quicker outs. In fact, here’s a quote from Ron Gardenhire in a Phil Mackey column from last season:

“Use that two-seamer, and use that slider down and in every once in a while, and that changeup, but pitch to contact early,” Gardenhire said. “That’ll get him deep into games. Because his stuff is so good. There’s times when you need to go for the strikeout.

“That’s when you save your Mr. Nasty, as they say. You throw the nasty pitches then. But those other times you need to pitch to contact to get you deeper into games. When you want that big strikeout, maybe with a man on second, and you’ve got an open base, take your shot with your stuff.”

When I first started thinking about the Twins’ strategy when it comes to starting pitchers, I thought they believed that high-contact pitchers had better ERAs than the Lirianos of the world. Then I saw Gardy’s quote above and realized that they employ so many of these pitchers because they favor the quick outs. It appears as though the Twins have a philosophy that striking out hitters, as Liriano has done in the past, raises your pitch count in a hurry. Think about Scott Baker. He spends so much time trying to make the perfect pitch for a strikeout that a hitter ends up fouling off about 6 pitches in an at-bat, and next thing we know, he’s at 80 pitches through 4 innings. At least this is what it sounds like from Gardy and Bert Blyleven. Meanwhile, since Pavano, Blackburn, and Marquis don’t strike out many hitters, the Twins seem to believe that these guys do a better job of getting deeper into games.

This is what I’ve chosen to focus on. I’m going to be checking for correlations by pairing two statistics that seem to be at the root of pitching to contact – K/9 and BB/9 – with each of these following statistics that appear to be related with the benefits of pitching to contact: innings pitched, pitches per start, and innings per start. Just for fun, I also checked K/9 and BB/9 against ERA to see if this would yield any interesting results. Also, I limited my sample to all 2011 starting pitchers that pitched at least 150 innings last season.

Just as a refresher, correlations range from -1.0 to 1.0, where -1.0 implies a negative correlation (as one goes up, the other comes down), 1.0 is a positive correlation (one goes up or down, the other goes the same direction), and 0 is no correlation. A disclaimer I’ll make here is that I was only checking for linear correlations. It’s very possible that nonlinear correlations existed for each data set, but I chose not to look for them for simplicity’s sake.

First, the just-for-fun correlations, K/9 vs. ERA and BB/9 vs. ERA.

** ERA vs. K/9: **-0.3644** ERA vs. BB/9**: 0.3446

*Click to embiggen (true for all graphs).*

Well, I think most of us could have predicted that there would be some correlation between these two with ERA. More strikeouts means a smaller ERA, while more walks means a higher ERA. Now, I chose ERA over FIP for two reasons. First, because I feel the Twins look more at ERA than FIP, and second, because strikeouts and walks are part of FIP to begin with, and I didn’t feel the need to check the correlation of sugar and chocolate chips to chocolate chip cookies.

From these two correlations, we can see that the Twins are correct in wanting their pitchers to have low walk rates, but the low strikeout rates aren’t favorable in having a low ERA. It’s certainly not impossible (just look at Pavano’s 2010 season), but it’s more difficult.

** K/9 vs. IP: **0.3599 **BB/9 vs. IP: **-0.2678

Well, it’s almost a repeat of the previous correlations. More strikeouts can mean more innings pitched, while fewer walks leads to more innings pitched. However, this doesn’t help pitching to contact much. Once again, limiting walks is beneficial, but there’s still no evidence that fewer strikeouts are helpful for a pitcher.

**Pitches/Start vs. K/9: **0.5071 **Pitches/Start vs. BB/9: **-0.0545

Ah, here’s where pitching to contact starts to gain some traction. Having a higher strikeout rate can lead to more pitches in a start, which means more mileage on a pitcher’s arm. Meanwhile, there’s virtually no correlation between the number of batters a pitcher walks and his number of pitches per start.

Something I should note is rather obvious once I mention it. We all know how (most) pitchers have a 100 pitch limit in starts. The range for pitches/start is not very much, as it appears to be between 85 and just under 120, so even if there is a moderate positive correlation between pitches/start and strikeout rate, it’s not adding on many pitches on average. Also, these two pairings fail to take into account the pitchers that are legitimately going deep into their starts versus those that are getting to the 100 pitch limit in 5 innings.

** IP/Start vs. K/9: **0.2399 **IP/Start vs. BB/9:** -0.5321

To remedy the issue with the previous correlations, I then looked at innings pitched per start versus strikeout and walk rates. With strikeouts, there is a weak positive correlation. While this doesn’t help the case of pitching to contact, it certainly doesn’t disprove it, either. As for walks, there’s a moderate negative correlation between that and getting deep into a start. I think that this can be easily explained. By limiting walks, a pitcher is limiting his number of baserunners allowed, and that means facing fewer hitters in a start.

After all this research, I felt that something was still missing. I’ve covered everything I said I would at the beginning, but it seemed like I needed to check strikeout and walk rates against one more thing. Then I remembered Gardy’s quote. The Twins preach getting quick outs, and none of the stats I’ve looked at seem to focus on quick outs. That’s when I determined what my last statistic should be.

**Pitches/IP vs. K/9: **0.2362 **Pitches/IP vs. BB/9: **0.7154

If a pitcher is successful at getting quick outs, then he should be able to limit his number of pitches per inning. The Twins seem to believe that a pitcher that strikes out a lot of hitters ends up throwing too many pitches in an inning, but the data here doesn’t fully support that claim. Yes, there is a positive correlation, but it’s weak. But if we look on the walk side of it, there’s the strongest correlation we’ve seen out of any of these. At 0.7154, it’s a moderately strong positive correlation between a pitcher’s walk rate and the number of pitches he throws in an inning. (In case you’re wondering, that one dot that is apart from the rest of the group on the walk graph is J.A. Happ.)

From all these correlations, it appears as though targeting pitch-to-contact pitchers is only a good idea if you’re focusing on those that limit their number of walks. However, there just isn’t a lot of evidence that suggests that sacrificing strikeouts for quicker outs is a good idea. For example, there’s a pitcher’s batting average on balls in play. It’s likely going to be around .300, so allowing hitters more chances to actually put the ball in play is going to mean more baserunners. This is partially supported as the correlation between K/9 and WHIP is -0.3846, so there is a weak-to-moderate negative correlation. If we go further, more baserunners means more hitters to pitch to, which leads to more pitches per inning.

Pitching to contact isn’t necessarily the wrong strategy, it’s just that from this data, I feel that its benefits are being overstated. Like I said in the last paragraph, if a pitcher wants to go deeper into a start, fewer walks are more important than fewer strikeouts. The Twins have definitely had it correct in preaching their pitchers to avoid the walks, but I don’t feel that avoiding the strikeout is as beneficial as the coaching staff believes.

December 23, 2011 at 4:23 pm |

First, great work. This is a neat study.

Now the first quibble. I might suggest it’s a little misguided to suggest that coaching a pitcher to throw the ball in the strike zone necessarily relates to an overall study of pitchers K/9 percentages to their efficiency in a game. I think it’s fair to say that if a pitcher could strike everyone out that he face, he probably would. So would coaches, even Gardy.

So there is a balancing act here: coaches are more willing to ask a pitcher to concentrate on throwing the ball in the zone if the coaching staff has more confidence in it missing bats or creating weak contact than the pitcher himself has. But, of course, I don’t know how you filter those pitchers.

Ultimately, what this shows, is that pitchers who don’t walk a lot of hitters tend to last a lot longer. It also shows that pitchers who strikeout pitchers tend to go longer in games but throw more pitches in an inning. Both seem fairly straightforward conclusions. But it is still a neat study, because we tend to worship at the alter of K/9 and it’s interesting to see it’s impact on these other stats.

(BTW, what would happen if you did K/BB to these other metrics? It seems like that is the standard for a lot of the Twins signings. And a stat that the coaching staff reveres, as if it represents the confidence they’re trying to instill.)

One thing I struggle with with these studies is what the correlation coefficients mean. Is .23 that much less a correlation than .36? It might be if it was straight ratio, but I don’t think it is. Is it 1/3 less effective? 4/5? 1/5? So when you say that .23 is a weak ratio, but don’t blanch at a .36 ratio for K/9 to ERA, I wonder if that is accurate.

So I don’t know if the study did what you wanted it to do, but that’s the way studies are sometimes. Still, I think it’s great that you did it, and greater still that you shared it with us.

December 23, 2011 at 4:43 pm |

I’m glad you liked this. I completely understand the quibbles. I did try to spit this out a little faster than I had originally planned, so I ended up cutting some corners. I do understand how you might say that a .23 correlation shouldn’t be too much different from a .36 correlation. I’m willing to admit that I likely chose to interpret the data in the way that most suited my argument.

Here’s the info for K/BB ratio.

K/BB vs. IP: 0.4376

K/BB vs. Pitches/Start: 0.3791

K/BB vs. IP/Start: 0.5996

K/BB vs. Pitches/IP: -0.4281

So there is some correlation between K/BB and these four statistics, though none are strong. K/BB seems to have a moderate positive correlation with innings pitched, pitches per start, and innings pitched per start, which the strongest one being to innings pitched per start. Meanwhile, there is a moderate negative correlation with pitches per inning pitched, which is still good, as this means that a high K/BB ratio means a pitcher is more likely to throw fewer pitches in an inning. A good K/BB ratio doesn’t guarantee success, but it seems to give you a better chance, so I would say that the Twins would be correct if they acquired pitchers based on K/BB ratio.

However, Jason Marquis doesn’t have a great K/BB ratio, so the Twins would have acquired him for a different reason. Perhaps it has to do with what I was tackling with the original post, or maybe it’s just that they felt they would get the most bang for their buck with him.

December 23, 2011 at 4:50 pm |

Now that I think about it, I believe I remember why I said two different things for the .23 and .36 correlations. I believe I was operating under sorting each correlation into a range, and I think in my head I had separated those two values into different ranges. Therefore, I felt that .23 was a weak correlation, but not weak enough to say there was no correlation, while .36 was more of a moderate correlation.

December 23, 2011 at 4:26 pm |

Two other tenets of the “pitch to contact” philosophy, as I understand it, are getting your guys off the field faster and giving the opposing team less of a look at your “stuff” that day.

December 23, 2011 at 10:17 pm |

It may also go unsaid that “pitch to contact” may be more efficient from a payroll perspective.

December 26, 2011 at 1:53 pm |

“Also, I limited my sample to all 2011 starting pitchers that pitched at least 150 innings last season.”

You are limiting your “sample” to the 107 pitchers who were successful in getting the most outs last year. Even with a K/9 of 9, a pitcher gets two times as many outs on balls in play as they will by strikeout. For the average pitcher its more like three times as many.

Since you are evaluating ONLY pitchers who are successful at getting people out, they will likely almost all be pitchers who are successful when hitters make contact. K/9 is the percentage of outs made by strikeout, it goes up either with more strikeouts or fewer other outs. But here, you have a base line of pitchers who are getting hitters out. So its not surprising that pitchers who also get strikeouts are going to do slightly better than those that don’t.

Of course, the key issue here is what happens when a batter makes contact. That is where most outs come from. And I don’t see how anything you posted deals with that.

To put that in numbers:

Pitcher A:

24 batters faced

4 strikeouts

%BIP .250

Total outs 19

K/9 5.8

Pitcher B:

24 batters faced

6 strikeouts

%BIP .333

Total outs 18

K/9 9

Pitcher C:

24 batters faced

6 strikeouts

%BIP .300

Total outs 18.6

K/9 8.71

Pitcher D:

24 batters faced

4 strikeouts

%BIP .300

Total outs 18.0

K/9 6.0

This is obviously simplified. The point is that K/9 goes up as a pitcher is less successful getting batters out on balls in play. And a pitcher with limited strikeouts who is very good at getting batters out on balls in play will be better than a pitcher who strikes batters out, but isn’t very good at getting hitters out on balls in play.

December 26, 2011 at 9:34 pm |

Basically I could have used K% (percentage of batters faced that struck out) instead of K/9. This corrects for pitchers that faced a different number of batters in a season, which means that a 4.5 K/9 isn’t necessarily the same for two pitchers. I’ll admit that the reason I didn’t use K% is just because I’m not comfortable with it yet.

Here’s what I mean by suggesting K% would have been better.

Pitcher A18 outs (6 innings)

6 walks + hits allowed

24 batters faced

3 strikeouts

4.5 K/9

12.5 K%

Pitcher B18 outs

12 walks + hits allowed

30 batters faced

3 strikeouts

4.5 K/9

10 K%

These pitchers have the same K/9, but by using K% we see that Pitcher A was technically more adept at getting strikeouts, because he struck out a greater percentage of hitters he faced.

Finally, why can’t I use a cutoff of 150 innings? I purposely avoided using the minimum number of innings needed to qualify for the pitching leaders lists (162 innings). I’ll admit that 150 innings isn’t much less than 162, but I had to do that because I felt it would make my results more reliable. Sample size was an issue; making it smaller would have skewed the results.

December 26, 2011 at 10:55 pm |

“Finally, why can’t I use a cutoff of 150 innings?”

Because Innings Pitched measures a pitcher’s success at getting people out. So you are assuming a certain level of success, 450 or more outs to be exact. Since most outs are a result of balls in play, that eliminates from your “sample” any pitcher who strikes out a lot of batters but doesn’t get a lot of batters out on balls in play. In essence, you have started with a group of pitchers who almost have to be effective pitching to contact in order to qualify.

“which means that a 4.5 K/9 isn’t necessarily the same for two pitchers.”

I think you missed the point. The two pitchers k/9 IS the same, since they both get the same percentage of outs from strikeouts.

What it means is that K/9 is not a measure of how often a pitcher strikes batters out. Strikeouts are just one component of K/9. The other component is how many outs a pitcher gets in other ways. The fewer outs a pitcher gets, the higher their k/9.

The formula is k9=(9*K)/(TotalOuts/3) so obviously a strikeout has a bigger impact on the final number than an out, but both effect the number.

December 29, 2011 at 3:32 am |

[...] a recent blog post, Andrew Bryz-Gornia attempts to understand the concept of Pitch to Contact. I have no idea what to make of pitch charts and how they make sense to anyone, but Bryz-Gornia [...]

December 30, 2011 at 8:47 am |

[...] Off the Mark makes an attempt to understand the concept of “pitch to contact.” [...]

January 2, 2012 at 12:36 pm |

[...] Off The Mark « An Attempt At Understanding “Pitch To Contact” [...]

March 28, 2012 at 7:21 am |

[...] to try to get quick outs and keep pitch counts down. There’s a problem with this philosophy: it’s wrong. There’s little to no evidence that sacrificing strikeouts in favor of quick ground-outs is [...]