Thursday, August 19, 2010

Another stat?

I just started a facebook page for this blog, check it out here, one of the people that "likes" the page had a question/comment that got me thinking. Greg Osmonson suggested: I think there should be a new stat in baseball - the double-play adjusted batting average. He then proceeded to work up an adjusted batting average for Joe Mauer and Michael Cuddyer. He used a formula: hits/(AB+GIDP).

For some reason, this touched a nerve with me. I'm the kind of guy that gets an idea in my head and obsesses about it until I am able to put it to some sort of practice. I started to ponder, is there a double play adjusted stat out there? If not, should I apply it to batting average, on-base percentage or slugging? I decided against Greg's original idea of using it to adjust batting average. This is an old time stat that has never really been changed. In fact, when this stat was invented, plate appearances that resulted in walks were not even considered at bats. Walks were looked down upon at that time and the powers that be in the game didn't feel a walk deserved to be counted as an at-bat. I also ruled out using double plays to adjust slugging. Slugging percentage is based on total bases in the hits made and is a measurement of power.

I did find that on-base percentage made the most sense to be adjusted for double plays. The reasoning is this: OBP measures how many times a batter doesn't get out per total plate appearance. A double play takes into account the out the batter creates for himself, but not the other out that was made. Should a consideration be made for the extra created out?

There are arguments against such a stat. Double plays can only happen with someone on base and with less than 2 outs. So is an adjustment relevant? A double play is somewhat based on luck of a situation. Hits, batting average, on-base percentage, and slugging are also somewhat based on luck, so I feel an adjusted stat is fine here. Aren't double plays accounted for in other offensive stats? WAR, VORP, and other stats that estimate total value of a player take into account double plays, but they aren't accounted for in an offensive counting stat. Will an adjusted stat really change or adjust OBP that much? That's what I will try to determine.

So, how to determine the stat. We'll use Joe Mauer as an example. In 2009 Mauer had a .444 OBP in 606 plate appearances. He had a total of 269 hits, walks and hit by pitch in 606 plate appearance. There are a couple of ways to adjust for double plays. Mauer grounded into 13 double plays in 2009. I could add the 13 DP to the plate appearances. Dividing the 269 times on base out 619 gives a Mauer a .435 adjusted OBP. Another way to adjust involves stripping the 13 extra outs from the 269 times on base and to divide 256 from 606. This would give Mauer a .422 adjusted OBP. Since double plays can be so crucial for pitchers and damaging to hitters and the fact that the amount of plate appearances stays the same, I think the latter is a better indicator. Penalizing the batter a time on base for every extra out he creates seems reasonable in determining how well a hitter is performing.

So Mauer's excellent .444 OBP in 2009 adjusts to a still excellent .422. His adjusted OPS, AVG/aOBP/SLG  goes from a 1.031 to a still very healthy 1.009. The inspiration for this post was Mauer's 2010 performance. Going into tonight's game, Mauer has an OBP of .407 with 17 GIDP in 460 PA. Adjusting his OBP with 187 hits/walks/HBP less 17 GIDP divided by 460 PA gives Mauer a .370 adjusted OBP. This is a much greater drop off and reflects Mauer's extra double plays in 2010. His adjusted OPS slips from .903 to .866.

Micheal Cuddyer has been killed even more than Mauer for his propensity to hit into double plays in 2010. Cuddyer has a .344 OBP and a .773 OPS with 20 GIDP in 503 PA. His  aOBP is .304 and aOPS is .733.

Mauer's drop off in 2010 vs. 2009 is 17 points due to more double plays and also fewer plate appearances, but is a stat like this significant? Would this affect all players in a similar way?

I will look at the regular and adjusted OBP of the OBP leaders in the major leagues.

Miguel Cabrera .433       adjusted .403
Joey Votto .423              adjusted .408
Kevin Youkilis .411        adjusted .402
Josh Hamilton .408         adjusted .392
Albert Pujols .408           adjusted .371
Joe Mauer .407               adjusted .370
Prince Fielder .403          adjusted .384
Adam Dunn .398             adjusted .388
Jayson Werth .396           adjusted .380
Adrian Gonzalez .391      adjusted .375

Miguel Cabrera leads the majors in OBP and is 2nd of the top 10 OBP leaders in aOBP. Albert Pujols is tied for 4th in OPS but slips to 12th in aOBP. I think this is a significant enough indicator that this statistic is useful as an extra measure to determine how well a batter is doing. Kevin Youkilis, before his season ended, created an out .3% less often than Pujols, but 3.1% less outs in total were created by Youkilis when double plays were factored in.

I will also look at the regular and adjusted OPS of the OPS leaders in the major leagues.

Cabrera 1.077              adjusted 1.047
Votto 1.016                  adjusted 1.001
Pujols .998                    adjusted .961
Youkilis .975                 adjusted .966
Jose Bautista .959         adjusted .945
Paul Konerko .954       adjusted .937
Robinson Cano .942     adjusted .919
Ryan Zimmerman .938  adjusted .912
Adrian Beltre .934         adjusted .889

Adrian Beltre has hit into 22 double plays. Factoring in this high amount of double plays into his OBP and OPS shows a truer value of his offensive prowess in the batters box.

So, What now? I submit this blog-post to my readers. I am looking for feedback on this idea. Is this a worthwhile statistic? Does a statistic already exist that I am missing? Is there a better way to calculate this? I will consider all feedback before going forward with using the statistic in future posts. An example of where it would come in handy: "Mauer is coming around and should be considered an MVP candidate. He has a .337/.407/.496 with 17 double plays in 460 plate appearances" or  "Mauer is coming around and should be considered an MVP candidate. He has a .337/.370/.496 in 460 plate appearances"

Please give me any and all feedback! Such as below:

Bill from "The Platoon Advantage" suggests so it would be great to factor in DP "rate"--% of GIDP per PA with a runner on 1st and < 2 outs--to OBP.

Here is a well written response that I found at the Victoria Times. I am still looking for more input on this subject and plan a follow-up:

Double Play Stats...

This post goes out to my Baby Brother, Jacob. For without him, I probably wouldn't have found the link to the blog post this blog post is about. That statement really don't make much sense, but really, it does.

Shawn wrote an article on his blog about a new stat he was working on coming up with. I think his stat has a legitimate point, but I think the execution of calculating it still needs work.

Shawn pointed out that guys who hit into double plays hurts his team. Thus, he calculated a stat that took the on-base percentage and subtracted an out for each double play the hitter hit into.

Here's where the problem comes in for me. Let's say Tom, Dick, and Harry bat in order during a game. Tom and Dick have excellent on-base percentages but very little power. This means Harry is likely to often come up to bat with Tom or Dick on first base. Poor Harry has a lot of opportunities to hit into double plays, and therefore, likely has hit into a lot of them.

On the other hand, Huey, Dewey, and Louie bat in order. Huey and Dewey have pretty good power, but their on-base percentages are pretty poor. Therefore, when Louie comes up to the plate, it's not very likely that there's someone on first base, so he doesn't really hit into that many double plays.

Here's the thing: both Harry and Louie could hit the ball exactly the same in every at-bat. But Harry's on-base percentage would be much lower than Louie's based on the stat Shawn presented us with. To me, it strikes me as unfair. It's simply a matter of where Harry and Louie are in the line-up. If you'd switch them, they'd likely be hitting exactly the same. (In fact, Dick is probably fairly vulnerable to the double-play, too, while Tom is probably not at all, when in reality all three are similar hitters in my hypothetical line-up.)

I told my cousin who was begging for the Twins to trade Joe Mauer to his team: "Why, do you need more guys to hit into double plays?" The reality is that Mauer is often like Harry in our hypothetical example: Span and Hudson typically get on base often, and Mauer is somewhat of a ground-ball hitter, making him vulnerable to double plays, putting him on the leader board for grounding into double plays. But people who know stats also point out that based on opportunities, his double play percentage isn't on the leader board.

I think there should be some way to account for double plays in stats. Right now, the only one I'm in favor of is the double play percentage, which calculates the percentage of double plays based on the opportunities to hit into them (runner at first, less than two outs). I wouldn't mind seeing the two stats (OBP and GDP%) combined, actually. I haven't worked out the science, but I'd like to see the math that Shawn figured out, but not straight-out subtracting a hit for every double play. With that, players get penalized when they have more opportunities, when really it's the manager's fault for putting him in that place in the line-up.


  1. It's interesting, to be sure. I'm a little hesitant because you're penalizing a player for coming up to bat with a player on first and less than an two outs. And yet I can see your point of the questionable practice of ignoring it altogether. I do like when we're giving the GIDP% stat--the number of times grounding into double plays based on opportunity to ground into double plays.

  2. You could have at least fixed my mis-spelling - it should be in not im. Thanks for the credit. I like where you took the concept.