Sunday, July 25, 2010

Out Rate: a Simple New Upgrade on OBP

On-base percentage is an extremely valuable stat because it shows the percentage of the time that a hitter avoids making an out at the plate.

However, as Nationals fans in particular have learned with Nyjer Morgan, there are more ways to make an out than to strike out, ground out, or fly out. You can get picked off, caught stealing, thrown out trying to advance on a flyball, and more.

And especially for lead-off hitters, the most important thing they need to achieve is to avoid creating outs so that there are base-runners for the heart of the order. A lead-off man who reaches base and then gets picked off or caught stealing hasn't really contributed any more than if they had made an out in the first place.

So I created a stat that I call out rate. Out rate is the percentage of all plate appearances that end in the hitter creating an out, either at the plate or on the bases. The stat also counts the extra outs created by GDPs. Here's the formula: Out rate = (H+BB+HBP-CS-GDP-POCS-OOB)/PA.

Some explanation on unfamiliar stats here. POCS stands for pick-off caught stealing. In the dataset that Baseball Reference uses, runners are charged with both a CS and a PO if he is judged to have been making a motion towards the next base before getting picked off. But BRef also tracks those "double counted" events as POCS, so by subtracting that number, we're able eliminate the overlap.

OOB is "outs on base"--other events when a runner is called out on the bases. Example plays are out advancing on a flyball, trying to "take the extra base" on a hit, or attempting to advance on a wild pitch.

The reason I think this represents an upgrade over OBP is because it represents a more complete view of the key thing OBP measures--how often a hitter does or doesn't create an out.

Some will object that I'm mixing base-running and hitting, two very different skill sets. Indeed I am. But when measuring the offensive output of a player (as opposed to their skills), it doesn't really matter how they are creating outs. Outs are outs. Indeed, OBP itself is a stat that is derived by looking at the outcomes of some very different skill sets--mainly strike zone discipline and contact ability. To really drill down to measure one skill at a time, we should be looking at more granular data still, like contact rate, walk rate, swing rates, etc.

I like out rate because it presents the data in the way it's more intuitively useful--i.e., how often a player creates an out, as opposed to measuring a negative, how often a player avoids making an out.

Here are the league-wide numbers, updated through the end of the weekend: http://bit.ly/bDBjhs.

And here are the rates for all the players in MLB who have started at least 30 games this season in the lead-off spot, ranked from best to worst, with OBP also listed for reference.

Nationals fans--take note. The only lead-off hitters worse than are guy aren't lead-off hitters anymore. A real good way to kill an offense is to create a league-leading number of outs from the lead-off spot.

Name Tm OBP Out %
Rickie Weeks MIL .377 .638
Rafael Furcal LAD .386 .645
Martin Prado ATL .363 .664
Kelly Johnson ARI .373 .673
Ichiro Suzuki SEA .369 .675
Austin Jackson DET .360 .676
Andres Torres SFG .367 .676
Angel Pagan NYM .362 .685
Felipe Lopez STL .338 .689
Marco Scutaro BOS .341 .696
David DeJesus KCR .331 .697
Chris Coghlan FLA .331 .697
Fred Lewis TOR .344 .698
Jimmy Rollins PHI .333 .700
Carlos Gonzalez COL .335 .700
Derek Jeter NYY .334 .703
Elvis Andrus TEX .357 .706
Shane Victorino PHI .314 .708
Denard Span MIN .341 .710
Scott Podsednik KCR .347 .710
Brandon Phillips CIN .342 .711
Erick Aybar LAA .339 .712
Jose Tabata PIT .335 .713
Corey Patterson BAL .328 .714
Jason Bartlett TBR .319 .715
Michael Bourn HOU .326 .718
Skip Schumaker STL .315 .723
Jose Reyes NYM .318 .724
Tony Gwynn SDP .330 .729
Ryan Theriot CHC .315 .730
Asdrubal Cabrera CLE .327 .735
Orlando Cabrera CIN .298 .738
Juan Pierre CHW .324 .739
Nyjer Morgan WSN .316 .743
Akinori Iwamura PIT .292 .751
Rajai Davis OAK .311 .752
Aaron Rowand SFG .288 .761

10 comments:

Eric said...

Ever thought about matching this with offensive contribution in bases?

Absolute Total Bases: TB+BB+IBB+HBP+SB.

ATB/OR=Bases per Out.

Eric said...
This comment has been removed by the author.
JB said...

I like the concept. How does this treat OOB that are just simple FCs? It's not a guy's fault if he's the forceout from a guy behind him.

Steven said...

not counted as outs.

JB said...

ahh... than the stat is perfect. Though I like the twist of giving a player credit for good things done on the bases suggested by Eric.

flippin said...

I agree with wieghting the type of "hit." I also think some sort of *net* value for base running would add more value to the stat. So, something like (sb-cs) would reflect the other reason you have a guy leading off: better chance of getting into scoring position. Also, if the player is charming, he should get points for that...

Will said...

I agree with flippin. There needs to be some way to weight stolen bases.

As you linked on the sidebar, the FanGraphs article about SB runs basically explains my problem. Rajai Davis, for example, is the third most valuable player in creating runs through SBs (2.75), but because he created 6 outs by getting caught stealing in 35 attempts (a fantastic 83% success rate), his Out Rate is negatively impacted, whereas a player with the exact same statistics, who never attempted to steal a base, would have a better Out Rate.

If you could integrate the FanGraphs SB runs into your stat, it would be greatly improved, or even simply adding in some sort of net stolen bases element (SB-CS).

Steven said...

This isn't intended to be a catch-all stat for offensive value. Look for wOBA to do that. This just pure and simple tells you how many outs per PA the player creates. That's a very useful metric, though not comprehensive.

Harper said...

I think this statement is a problem : "It doesn't really matter how they are creating outs. Outs are outs"

That's not right. An "OBP" out is an absolute. Either you get on and can score or you don't and can't. A CS out or an OOB out is an out trying to increase your ability to score. Because the goal of the game is to score runs not to avoid outs, this matters. By not factoring the successes along with the failure outrate will always favor those who don't get caught stealing (seems to be the main difference driver between OBP and this stat). Sometimes that'll be a great base stealer (Victorino) but usually it'll be those that just don't steal alot of bases (Lopez, DeJesus, Coghlan)

Of course that in itself could be the argument - that it's worth it to go with a slow leadoff hitter because they make less outs and so few SB guys can hit the break even point. But I'd hate to see an new era Vince Coleman get knocked down because he got CS so many times even if the percentage was good.

Section 222 said...

Fascinating. And who has the best OR among the Nats' hitters? Adam Dunn of course (.638), who has only 2 GDPs, compared to 10 for Zim and Hammer, and 18 for Pudge. Zim (.650) is next, followed by ... Michael Morse (.651). Willingham is a .698, same as Justin Maxwell.

Felipe Lopez has a 50 point advantage over Nyjer. Ouch.