How many runs does poor defense add?

[Xeifrank]

If you took all errors, passed balls and wild pitches out of the game and assume they never happened that all errors/pb/wp were fielded cleanly how many less runs per game would that be worth?

Example:
If a team with an average defense allowed 4.5 runs per game, how many runs per game would it allow given the circumstances listed above?

vr, Xeifrank

[Tango Tiger]

A WP and PB adds .25 runs, an error adds .50 runs, more or less. Go to Retrosheet or b-r.com and find out how often each happens per game.

[weskelton]

Simply subtracting the linear weight values of the PB's and WP's works, as they simply become just another pitch. However, by assuming that an error is otherwise fielded cleanly, aren't we implying that it becomes an out (at least in most cases). In this case, it seems that not only do you need to subtract the value of the error, but also add the value of an out.

error ~= .5
out = ~= -.1

error -> out = -.1 - .5 = -.6

Does that make sense?

[Tango Tiger]

Yes, you are correct.

I was presuming 0.75 runs for the error on the batter events (turning out to hit), and 0.25 runs for the error on the runner events (getting an extra base). 0.50 was a nice round number.

[Xeifrank]

A WP and PB adds .25 runs, an error adds .50 runs, more or less. Go to Retrosheet or b-r.com and find out how often each happens per game.

Thanks, but I am not familiar with how to mine data from b-r.com to find such events and how often they happen per game. Anyone care to help explain how to do this?
vr, Xei

[Tango Tiger]

Look under "freq" to get the frequency of each event:

http://www.tangotiger.net/bsrexpl.html

[Xeifrank]

Look under "freq" to get the frequency of each event:
http://www.tangotiger.net/bsrexpl.html

Thanks, but i don't see anything called "freq" at the b-r.com website.
vr, Xei

[west coast orange and black]

on this (http://www.tangotiger.net/bsrexpl.html) particular link, "freq" is resides as the 5th column.

[Tango Tiger]

Right, "freq" is on the webpage I pointed you to. That tells you how often each event occurred.

[Xeifrank]

Right, "freq" is on the webpage I pointed you to. That tells you how often each event occurred.

Ok, it says an error occurred 31270 times. I guess the next step is to calculate how many games per team these 31270 errors occurred in. Which could get messy if there was any expansion in there.
vr, Xei

[SABR Matt]

don't need to break it down by tema...I think that table also shows the number of runs scored in that period and the total number of games...you can get the current RS/G and the take the runs and subtract out the bad defense elements and recalculate.

You're going to find that it's somewhere between 4 and 7% depending on the league...a 5 RPG league would reduce to something like a 4.75 RPG league.

[weskelton]

Since the table doesn't have these details...

1974-1990

G = 69,128
R = 295,349
R/G = 4.2725

So,

295349 Original Runs
- .25 * 4980 BK
- .276 * 5021 PB
- .279 * 18371 WP
- .481 * 31270 E
+ (-.098) * 31270 E -> outs
--------------------------------------
269487.36 Perfect Defense Runs

Perfect Defense
R/G total = 3.8984
R/G saved = .3741 (8.76%)

[Xeifrank]

Since the table doesn't have these details...

1974-1990

G = 69,128
R = 295,349
R/G = 4.2725

So,

295349 Original Runs
- .25 * 4980 BK
- .276 * 5021 PB
- .279 * 18371 WP
- .481 * 31270 E
+ (-.098) * 31270 E -> outs
--------------------------------------
269487.36 Perfect Defense Runs

Perfect Defense
R/G total = 3.8984
R/G saved = .3741 (8.76%)

Thanks.
vr, Xeifrank

[SABR Matt]

A tad higher than I thought.

But that's because it's 1972-1990...lower RS/G...same defense. In the 90s it's been a lower percentage because the defense hasn't gotten worse but the hitting has gotten better.

[StillFlash]

Mat, remember though that defense is not just about errors, and erros are not just about fielders.

A bad defense will give up more hits as well as more errors, and "bad" scorers will award hits on what should to be errors, making the fielding pcts better and ERAs worse than what they should be.

I recently started a study or reached on error rates from the RetroSheet play by play. There was a steady decline of errors as a pct of all runners reaching base safely from the late 70's to early 90's. About 1/3 of pre-1980 reached on errors are now called hits.