December 29, 2006
Strikeout Wars
Steve Lombardi writes:
I have a question that I hope you can help me with - it's been mentioned recently that hitters, for the most part, control the ride on batted ball types over pitchers and their tendencies.
However, how does it work for contact hitters and strikeout pitchers? Who wins that battle - in the non-batted ball contest?
Anything you can share would be appreciated. Thanks and regards, Steve Lombardi
The Day by Day Database contains batter vs. pitcher matchups back to 2000. Using that data, I selected 20 low strikeout hitters and 20 high strikeout pitchers. Here are strikeout percentages (100*K/PA) for various combinations:
| Test | K Percentage |
| Overall MLB | 16.75% |
| Best Batters vs. All Pitchers | 7.67% |
| Best Pitchers vs. All Batters | 24.53% |
| Best vs. Best | 11.72% |
The best vs. best comparison resulted in 403 K in 3438 plate appearances. It certainly appears that the contact hitters "win" the battle against the strikeout pitchers. Maybe that should be a strategy. When you go up against Clemens, send up a lineup of batters that don't strike out often.
Update: TangoTiger writes in the comments:
Studes pointed out on my blog that the Odds Ratio method works pefectly here.
Following the step-by-step on my blog, I get a matchup expectation of 11.8%, which compares mighty favorably to the actual 11.7%.
That is, there is no greater advantage on either side. The resultant matchup is exactly predicted by the matchup method.
Studes was right.
Bill James invented a similar method in the mid 90's. I didn't know it was in general use, but it gives the same answer, an expected strikeout rate of 11.8%. What this basically means is that the low strikeout hitters are farther above average than high strikeout pitchers.
Posted by David Pinto at
10:21 AM
|
Statistics
|
TrackBack (0)
Studes pointed out on my blog that the Odds Ratio method works pefectly here.
Following the step-by-step on my blog, I get a matchup expectation of 11.8%, which compares mighty favorably to the actual 11.7%.
That is, there is no greater advantage on either side. The resultant matchup is exactly predicted by the matchup method.
Studes was right.
Tango,
Thanks for running that. I was going to run a similar analysis, but I was writing from my main computer with that software. Good to see these predictors work!
When Pedro was first with the Red Sox and dominating the Yankees, I thought Torre should tell the team to choke up on the bat. I'm still convinced that is a good way to approach the big strikeout pitchers.
Tom - what if you compared contact hitters as a group to other hitters as a group to see how large the gap is in their production.
And, then compare contact hitters and the other hitters against strikeout pitchers - to see if that gap remains the same, grows, or gets smaller.
If the gap between the contact and non-contact hitters gets very much smaller, when facing K-pitchers, would you agree that contact hitters have an advantage over strikeout pitchers?
I did a similar-type analysis in The Book (see Tables 29 through 36).
It's simply a matter for you to create whatever "profiles" you want, and seeing if the predicted matches reality. Of what I've studied, only handedness (obviously) and GB/FB tendencies showed an effect beyond predicted.
It will be interesting if you can run something similar with your parameters.
Couple of things:
I prefer to use contact rate ((PA-BB-HBP-K)/PA) as the dependent variable rather than strikeout rate, for both pitchers and hitters. I think that gives a better overall picture.
I also think that what you are interested in is not just strikeout rate, but overall production. It doesn't do contact hitters as much good to strike out less often if they are getting subtantially less production out of their balls in play. What I've seen in the limited data sets that I've looked at is that while contact hitters (who are typically lesser hitters than the population of hitters as a whole) lose *less* of their overall production against strikeout pitchers than do other classes of hitters, they still tend to be less productive overall.
In The Book, I looked at contact rate, and I looked at overall production. The resultant matchup was what was expected.
What this basically means is that the low strikeout hitters are farther above average than high strikeout pitchers.
On a percentage basis, yes, that's what it means.
I know this is counterintuitive to people, but the results don't mean that these particular batters have an edge against these particular pitchers. The results are just a natural outcome of probability and distribution. In the same vein, high K% batters will strike out much more often than average (theirs or the league's) against high K% pitchers.
What about Mike's point that contact hitters lose *less* of their overall production against strikeout pitchers than do other classes of hitters?
Tango didn't reach that conclusion in The Book. And, as Mike says, high-contact hitters are lower production batters anyway.
A bottom line number from the Book:
When high contact batters face low contact pitchers, their wOBA is .312. When low contact batters face low contact hitters, their wOBA is .335, .23 points better.
The finding is that high strikeout batters are more productive than low strikeout batters, even against strikeout pitchers.
Thanks studes - but what is the delta from when high and low batters face high contact pitchers?
Don't the high contact batters lose more off their game when they face low contact pitchers?
Just a tack on thing too...while I like Mike's approach, I think it's fishing with a really big net. When you grab all high contact hitters for the compare, you're going to get guys like Neifi Perez and Miggy Cairo. But, what are the odds of them getting 25 PAs in a post-season series?
It almost makes more sense to compare Johnny Damon types and an A-Rod types against low contact pitchers - since it's the good hitters, for the most part, that are on playoff teams and who get the PAs in the post-season.
It almost makes more sense to look at med-to-high OPS low contact hitters and med-to-high OPS high contact hitters against low contact pitchers to see the true difference come October. Does this make sense?