FG on Fox: Toronto’s Altered Offensive Approach at Home

by
January 4, 2016

Going into the 2015 season, we had a pretty good idea that the Toronto Blue Jays were going to hit a lot of home runs. After all, they hit the third-most home runs in baseball during 2014, and then added Josh Donaldson; the pieces were there for a huge offensive season from the entire team. But even with the talented personnel and a hitter-friendly home stadium, 2015 was the kind of season that was probably on the high-end of expectations: the Jays hit 232 home runs, the most by any team since the Yankees hit 245 in 2012.

As Matt Snyder pointed out in late September, the 2015 Blue Jays were only the 14th team in major league history to have three players with 35+ home runs each, and were the first team to have three since the 2006 White Sox. Those players, of course, were Josh Donaldson, Jose Bautista, and Edwin Encarnacion. Digging deeper into the stats, the offensive approach shown by those players at the Rogers Centre was a driving force behind the team’s power explosion.

By July, we had a sense that Donaldson was intentionally altering his plate approach at home to hit more homers: he was striking out more, walking less, and pulling the ball far more often when playing at the Rogers Centre than on the road. In short, he was being ultra-aggressive at the plate when at home, and it turned out to be a big part of what would become an MVP season for the third baseman. A quick look at the increase in his pull rate at home in 2015 when compared to 2013 & 2014 tells a big part of the story of his year:

Donaldson_Pull_Compare

Big power seasons often follow short-term increases in pull tendencies, and Donaldson was no different. And, looking further down the lineup, he wasn’t alone in changing his approach to get the most out of playing in Toronto’s hitter-friendly environment during 2015. Donaldson’s main partner in adopting these more aggressive changes was Bautista, who showed a few important tweaks to his Rogers Centre approach between 2014 and 2015. To begin with, he pulled the ball in Toronto more than he ever had before, owning the third-highest change in pull tendency out of all qualified hitters when at home.

Read the rest on Fox Sports.





 

Owen Watson writes for FanGraphs and The Hardball Times. Follow him on Twitter @ohwatson.

21 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
MGL
8 years ago

“….these large changes in approach seem to be intentional.”

For Donaldson, a 3.8% difference in pull percentage, while it looks big on a graph which is “blown up” (not a good way to present something, BTW), is only around 1 SD, which suggests that it could easily have occurred by chance alone.

13
Owen Watson
8 years ago
Reply to  MGL

Thanks for the comment on the SD! As for the graphs, truncated axes are widely accepted on line graphs. If these were bar graphs, I’d have started it at 0.

0
bluejaysstatsgeek
8 years ago
Reply to  Owen Watson

As someone that teaches university statistics courses, I disagree that truncated axes are widely accepted. Wide used, true. But if one wants to distort something to make a point, one of the easiest ways to do so is to truncate the axis. Of course the iconic reference on this is Darrell Huff’s “How to Lie with Statistics”

Mitchel’s point regarding the difference being overstated is still valid.

Now it is possible that if you can use hit vectors based on pitch location, you may find the significance that you want. Looking at the pull rate on all pitches means that any intent would get lost in the noise of (1) pitches that would not be pulled, such as the outer third of the plate, and (2) pitches that would almost always be pulled, on the inner third.

If you told me that the middle third pull rate went from say 18% to 22%, that might be more impressive, but with the smaller sample size, might still not be significant.

0
Owen Watson
8 years ago
Reply to  bluejaysstatsgeek

Thanks, statsgeek. I’m going to stick to my guns on the line graphs, as there seems to be a wide-ranging level of acceptability. Bar graphs, absolutely, the axis should start at zero. But we’re trying to show our data, and having axes that run from 0-100 does no one any good here — especially since qualified hitters ranged in pull % from 21.8% at the extreme low end to 60.2% at the extreme high end. Since the axes run from 25-65%, we’ve encompassed 99+% of possible hitters in that axis.

I also think that Mitchel’s point is entirely valid, and it was certainly not my intent to mislead with the graphs or disagree with him: my job is to try to find trend and narrative in a huge jumble of noise, and any specific sentences that concretely state something as fact (or almost fact) are most likely mistakes overlooked in the writing and editing process. Since these articles are written and pushed out quickly, they slip through the cracks sometimes, and I usually make a final pass to ensure that a healthy skepticism is included in most things I write about. But let me end with this: thanks for taking the time to comment, because it makes me better and keeps FanGraphs awesome. So thank you!

0
MGL
8 years ago
Reply to  bluejaysstatsgeek

I am not statistician so I will defer to them on when it is proper to truncate an axis. It seems to me that it is easy to represent the data in doing so (although I certainly don’t think the author was trying to do so in this case), so there has to be some standard that one should adhere to in order to avoid accidentally or intentionally doing so.

On the other hand, if the range is small and not near zero, it can be problematic and asthetically awkward to start an axis at zero. Again, I don’t know the correct answer or protocol, if there is any.

In this specific case, because we know that differences of 3 or 4% are small, as compared to the standard error for one full season, it is probably not a good idea to scale the axes such that it looks like a large difference.

I don’t think it’s a big deal and was just an aside in my original post.

All that being said, I think that there is lots of other evidence that Donaldson may have intentionally altered his swing to produce more pulled balls as pointed out in the article, but we should keep in mind that 3.8% is not much more than the expected noise (1 standard error).

The suggestion above about using pitch location to increase the power of the test is a good one!

0
MGL
8 years ago
Reply to  bluejaysstatsgeek

Misrepresent, not represent, the data I meant…

0
Owen Watson
8 years ago
Reply to  bluejaysstatsgeek

As one more small addition, there was an interesting point in respect to pitch location in the Donaldson article I referenced (and wrote) in July. At that point in the season, he was showing a large discrepancy between his pulled fly ball/line drive rate at home vs. his rate away from home (41.4% vs. 25.5%), which I postulated was the result of where he had been pitched up until that point. There are a couple heat maps in that article that show he saw more pitches inside at home leading up to July, whether intentional or not. With the whole season under our belt, that rate normalized (34.6% at home vs. 33% total), but it’s still an interesting point in this conversation! You can click here for that article.

0
rosen380
8 years ago
Reply to  MGL

For 2015, among the 196 players with 200+ PA each at home and away, it isn’t even 1SD.

The ABS difference was 3.7% with a SD of 2.8%, so 3.8% would be just about as close to typical as you can get.

The big outliers are:
DIFF HOME AWAY PLAYER
12% 30% 42% Adam Lind
12% 37% 49% Kevin Pillar
10% 30% 40% Ian Desmond
10% 42% 52% Cody Asche
10% 40% 49% Chase Headley
9% 42% 52% Stephen Drew
9% 40% 49% Carlos Beltran
9% 31% 40% Marcel Ozuna
9% 39% 31% Brett Gardner
9% 45% 37% Torii Hunter
9% 58% 49% Jose Bautista

0