Defensive Replacement Level Defined
Okay, this is a “look what I found” post, so I’m not drawing conclusions, but looking at data and saying, “What does this mean?” in hopes that the peer-review world of sabermetrics can take this type of information and run with it. One of keys to more advancement is open-source, which enables others, like Joe Arthur, Sean Smith, Colin Wyers, who are better at database work than I am, to see where they can take the information.
At any rate, a few years ago, when I was doing the 20-years worth of analysis (thanks again SG!!), Sean and Mike and I were discussing the limits of what a team would put on the field. I said at the time that could be valuable information, but I never really did anything about it. Then the other night in a discussion with Colin, I popped in the spreadsheet and did a few pivot tables, and something jumped out at me - defensive replacement level is similar to offensive replacement level. Yes, that is the conclusion, and it came first in this piece, so you don’t have to read all the drivel.
The old discussion essentially said “Is there a floor where a player isn’t allowed to play defense any longer, and where is that floor?” Now, I will point out there is a very small bias because if you don’t catch anything, you never make it to any reasonable number of innings played. Some player that were so awful in very few inning could have lowered the numbers, but the fact that they weren’t allowed to play a minimal number of innings supports the argument that MLB teams won’t allow a player that fields below the named level to play there for an extended time.
So what I did was take the 1987-2008 Defensive Runs Saved (DRS) database and look at the 20-year Zone Rating (ZR) data by position. It included several looks. Do I draw the cutoff at 500 IP (innings played)? Is it 1000 IP? Somewhere in between? For this argument, I am using 500 IP. That is more than a third of a season, and gives a player about 60 games to show that he stinks. this is where teams will say “That’s enough.” Some teams do it earlier, but 200 IP just isn’t enough to draw any conclusions…I think. In this piece, we are gong to claim that a team had to give a player 500 IP to show he could or could not handle a defensive position. I did a cursory glance at 200 IP, and the ZR drop markedly, but very few number of players only got 200 IP and not 500 IP, so I am going to make the initial claims based on 500 IP. That’s assumption #1.
So here’s what we are looking at with 500 IP (and I did round the Replacement Level (RL) Zone Rating) compared to average:
[b]Pos Avg ZR RL ZR RL %[/b]
1b 0.845 0.720 85.2%
2b 0.824 0.730 88.6%
3b 0.754 0.630 83.5%
cf 0.884 0.740 83.7%
lf 0.867 0.730 84.2%
rf 0.872 0.740 84.9%
ss 0.842 0.730 86.7%
Roughly, offensive RL is 80% of average. Defensive RL is approximately 85% of average, but slightly higher for Middle Infield.
What happens when you start expecting a player to play a full season? Well, he’s a poor fielder, so you sub out for him in the 7th inning, so he only get 1000 IP (minimum). What does that do to the RL?
Pos Avg ZR RL ZR RL % Replacement Player
1b 0.845 0.720 85.2% 1997 Mo Vaughn
2b 0.824 0.735 89.2% 1998 Todd Walker
3b 0.754 0.650 86.2% 1989 Howard Johnson
cf 0.884 0.805 91.0% 2005 Ken Griffey
lf 0.867 0.730 84.2% 1988 Kevin McReynolds
rf 0.872 0.740 84.9% 1988 Darryl Strawberry
ss 0.842 0.760 90.2% 2002 Tony Womack
Boy, Davey Johnson really didn’t care much for defensive prowess. Note that several positions, 1B, LF and RF didn’t change. And the middle positions really moved up. Teams have a strong aversion to putting poor fielder at the positions that get the most chances. It also has to make Yankee fans smile to see that there’s a defensive measure that doesn’t name Derek Jeter as the worst (he was second worst).
Another assumption: I eliminated Manny Ramirez in Boston from this. If you want to know how poorly Manny played LF in Boston, over these 22 years of data, no other Boston LF showed up in this analysis. No Troy O’Leary, No Mike Greenwell. Sure, if I scrolled up far enough, they would be on the list somewhere, but Manny has five seasons of poor defense in Boston that are unrivaled by any other Boston LF. That’s a real individual effect.
So there you have it - defensive replacement level, as defined in the same parameters of offensive replacement level - where you tend to be replaced if your performance reaches.
As a product of this analysis, we can get a good idea of how many runs a replacement level player will cost you defensively, permitted to play out the season:
Pos Worst Fielder RS
You can see from this that the worst player (that an MLB player will run out there for an entire season) will only cost you about 40 runs. More importantly, while a SS/2B/CF gets the most BIP, teams do not risk as much at those positions, so they aren’t really giving up as many runs as the corner outfielders nor third base. It’s also interesting to see that first base really is the lowest risk position.
Now when we say that while defensive RL is higher than offensive RL, we know by how much.
Posted: June 03, 2009 at 03:02 AM | 40 comment(s)
Login to Bookmark