34 replies to this topic

[Foulkey Reese]

Dewan, who pioneered the plus/minus rating system (which measures how a player stacks up defensively in terms of numbers of plays made and runs allowed compared to an average player at his position), takes a detailed look at the Sox’ 2009 performance at third base, shortstop and in center field, and compares it to the performances of newcomers Adrian Beltre, Marco Scutaro and Mike Cameron at those positions. Those changes, coupled with the move of Jacoby Ellsbury from center (where he rated as below average) to left field (where he has been above average, and replaces Jason Bay, who rated as below average at the position), will have a significant effect on the Sox.

To wit, Dewan claims that the Sox will be roughly 80-90 runs better defensively in 2010 than they were in 2009. Defensively, the team was about 52 runs worse than the average team in 2009; in 2010, the team now projects to be well above average.

Seems a bit optimistic, but I admit that I know way less about defensive stats than I would like to.

WEEI

Dewan

[RingoOSU]

It's impossible for a team that allowed 736 runs to be 52 runs worse than average. If the sox were worse defensively, the pitching more than made up for it. These numbers are over calculating the effect of defense.

[Vermonter At Large]

It's impossible for a team that allowed 736 runs to be 52 runs worse than average. If the sox were worse defensively, the pitching more than made up for it. These numbers are over calculating the effect of defense.

Well ... if Dewan believes this, I'm guessing that at least some of the F.O. have also bought into this notion as well. I agree with RIngo that it the concept is absurd, and feel that this team is more likely to allow around the same number of runs, or perhaps 20-30 fewer at most, but score 80-90 fewer runs on offense, but we shall see.

If the F.O. has bought into Dewan's +/- and has purposefully constructed the team expecting 89-90 fewer runs allowed. If nothing else, I guess, the 2010 season could end up being a great experiment in defensive metrics ...

P.S. Have I ever mentioned how much I hate defensive metrics that are calculated in runs allowed?

[URI]

I think, unless I'm reading the context wrong, is that Dewan is using 'defense' interchangeably with 'fielding', meaning that he's talking about the fielders, not run prevention.

So the Sox were 52 runs worse than the average team at fielding, not at overall preventing runs.

It's imprecise, since defense has usually meant fielding+pitching (run prevention). Essentially, Dewan thinks that through catching the ball, the Sox are going to shave about 80 runs off their total, independent of pitching.

[pokey_reese]

Well ... if Dewan believes this, I'm guessing that at least some of the F.O. have also bought into this notion as well. I agree with RIngo that it the concept is absurd, and feel that this team is more likely to allow around the same number of runs, or perhaps 20-30 fewer at most, but score 80-90 fewer runs on offense, but we shall see.

If the F.O. has bought into Dewan's +/- and has purposefully constructed the team expecting 89-90 fewer runs allowed, that would be very interesting. If nothing else, I guess, the 2010 season could end up being a great experiment in defensive metrics ...

While the 80-90 number might be a little high (assume a slight decline for the 37 year old Cameron, and a return to normalcy for Scutaro who had a career year on offense and defense last year), 60+ runs of defense seems to make sense easily. Who watched the left side of this infield last year and didn't think that there was room for massive improvement? Metrics and eyeballs agree that Lowell alone was not just below average, but horrible; how many innings were extended when balls that were 6 feet to his left got by Lowell and he just looked at them sadly?

This team took 4 positions out of 7 (excluding pitcher and catcher, where fielding is the smallest component of their run prevention) and replaced below average defense with average to great defense, as well as upgraded the rotation with Lackey. While the exact number is up for debate, I am surprised that you think that the team is likely to allow roughly the same number of runs. I certainly understand taking defensive metrics with a grain of salt, but unless you think that the pitching is going to regress greatly from last year, why wouldn't this team allow significantly fewer runs even ignoring the metrics? Better defense and better pitching should equal fewer runs, right? Plus, at least in terms of the pitching, the fact that Sox pitchers had better FIPs than their actual earned run results suggests that they shouldn't regress in front of a better defense.

[Arock78]

If we take the 52 number at face-value, with an average fielding team, they'd have allowed 684 runs. The average in MLB was 747 in 2009, so it's clear he's not lumping pitching in with defense.

736-80=656, so the question is whether the team's fielding is 28-38 runs above average for 2010 (assuming no change in the league average).

edit: I'll add that run prevention is much greater in the NL than AL, so 28 runs above average in MLB would have been 115 in the AL.

Edited by Arock78, 13 January 2010 - 01:06 PM.

[philly sox fan]

P.S. Have I ever mentioned how much I hate defensive metrics that are calculated in runs allowed?

I'm not sure if you prefer "plays made" or something like that, but ironically when Dewan first came otu with his +/- system it was based on plays instead of runs and everybody in the sabre community who is used to talking in terms of runs threw a fit. Dewan's move towards Defensive Runs Saved (or whatever he's calling it) is only because of that feedback.

[smastroyin]

Without going into the details of +/-, just looking at simple defensive efficiency gives us a clue that there is a lot of room for improvement year over year.

Defensive efficiency is basically 1-BABIP.

2007 .714
2008 .708
2009 .687

You can argue about the difficulty of the balls hit in every year but the fact is that last year the Sox were horrendous, even compared to their recent selves, at converting batted balls into outs.

[phrenile]

Assists are a very crude metric, and one year's worth of numbers is probably worthless, but for poops and giggles: based on last year's numbers, upgrading shortstop and third base is worth +106 assists.

2009 Assists

Position	Player	Assists
SS	Nick Green	198
SS	Alex Gonzalez	107
SS	Jed Lowrie	52
	Total SS	357
SS	Marco Scutaro	421
	Difference	+64
3B	Mike Lowell	174
3B	Nick Green	4
3B	Jed Lowrie	4
	Total 3B	182
3B	Adrian Beltre	224
	Difference	+42

Edited by phrenile, 13 January 2010 - 01:40 PM.

[Eric Van]

I'm amazed that Dewan has the Sox as -52 last year (which apparently includes about -4 for VMart, maybe more), since UZR has them as -16.

My projections for next year, averaging FB and UZR:

Youk +10
Pedroia +8
Scutaro +7
Beltre +18
Ellsbury +4 (gains 12 moving from CF to LF; could be more)
Cameron +6
Drew +6

That's +59. If they were really -32 last year (again, averaging both metrics), then that is a 90 run upgrade.

[Eric Van]

My projections for next year, averaging FB and UZR:

I'm now interested enough to do this separately with each metric.

UZR
Youk +7
Pedroia +8
Scutaro +3
Beltre +14
Ellsbury +2
Cameron +9
Drew +11

That's +54, for a +70 upgrade.

FB
Youk +13
Pedroia +9
Scutaro +10
Beltre +21
Ellsbury +6
Cameron +3
Drew +1

That's +63, for a +111 upgrade (if they were -48 without VMart).

[Frisbetarian]

I'm amazed that Dewan has the Sox as -52 last year (which apparently includes about -4 for VMart, maybe more), since UZR has them as -16.

My projections for next year, averaging FB and UZR:

Youk +10
Pedroia +8
Scutaro +7
Beltre +18
Ellsbury +4 (gains 12 moving from CF to LF; could be more)
Cameron +6
Drew +6

That's +59. If they were really -32 last year (again, averaging both metrics), then that is a 90 run upgrade.

Plus/Minus includes an assessment of pitchers and catchers including stolen bases, while UZR does not. The Sox pitchers and catchers were -25 per Plus/Minus last year, which makes the metrics a lot closer.

[Eric Van]

Plus/Minus includes an assessment of pitchers and catchers including stolen bases, while UZR does not. The Sox pitchers and catchers were -25 per Plus/Minus last year, which makes the metrics a lot closer.

That does explain a lot. I looked up Varitek and he was +8 for Catcher's ERA (which they may or may not be including) and -8 for SB / CS.

Revising the previous numbers, you get +70 by UZR and +90 by FB. I'm comfortable saying the defense looks to be 80 runs better -- of which almost half (39) is at 3B.

[502 to Right]

Defense certainly will not be the "fatal flaw" of the 2010 team. Let's hope it brings the same success as the post-Nomar 2004 squad.

[Vermonter At Large]

Well ... let's take a look at run attribution from a different slant.

It is possible to get a qualitative look at defensive run attribution by analyzing hits allowed (1B, 3B, 3B and HR) against hit trajectory (GB, FB, LD). The general notion is that by looking at, for instance, the team's 1B/GB rate against the league average, we can get a rough estimate of the singles that could be attributed to the team defense.

First, though, we need to equalize the numbers a bit, since the Sox pitching staff gave up GB, FB and LD rates quite different from the league average. In fact, the Sox pitchers gave up fewer GB's per BIP than the league average (41.5% to 43.8% league average), more fly balls (38.8 vs 35.9) and roughly the same number of line drives (18.9 vs 19.2). In total the Sox pitchers gave up 44 fewer BIPs than the league average, so in general, the Sox staff did a pretty decent job overall last season.

To equalize the rates, then, we calculate the 1B, 2B, 3B and HR per BIP rates, then multiply them by the league BIP average to yield something resembling a BABIP-neutral assessment of hits allowed for each of the hit trajectory. Here are the results for the 2009 Red Sox compared to the league average:

CODE

     1B   2B   3B   HR
-----------------------
GB  +21   -6   -1   NA
FB  +24  +23   -2   -7
LD  +18   +1   -2   -4
---------------------
Tot +63  +18   -5  -11

Looking at ground balls first, I think that it is safe to assume that the Sox 2009 infield was a bit on the pourous side, allowing 21 more singles than the league average. There are no real mitigating factors other than karma to adjust this rate, and it is probably safe to assume that we can assign all of these extra singles to shoddy defense, or roughly 10 runs.

I think that the six fewer doubles on ground balls can be largely attributed to the dimensions of Fenway. Hot grounders hit to the right or left field corners would account for nearly any extra base hit on a ground ball, but the Monster probably prevents many of these hot grounders down the left field line from becoming doubles. In general, however, I would find it difficult to penalize a fielder too badly for a ball hit hard down the line.

Fly ball analysis is a bit trickier. The Sox allowed +24 more singles on FBs than the league average, and to get a better picture of the skew, the numbers were 101 against Boston vs 77 against the league average, so the skew would probably be statistically higher if the Fenway numbers were deducted from the overall set. There are essentially two types of FBs that land for singles - seeing eye pop flies and balls hit off the Monster. I suppose that a very small percentage of these could be attributed to a fielder, but most are clearly karmic.

Similarly, the +23 doubles on fly balls were probably primarily Monster shots.

Line drive analysis is a little trickier. We should logically expect that the Monster will increase singles to some extent, and probably reduce doubles (unlike fly balls) by the same general rate (thus converting balls that would be doubles in a bigger park into wall-ball singles). I think that there are potentially more subtle effects too. A left-fielder in Fenway generally plays much shallower than in other parks. If positioned correctly, the left fielder may be able to turn an occasional potential line drive single into an out, but I also think that playing shallower limits the outfielders range to his left or right, so may cause some potentially catchable balls to go to the wall. I also think that the larger-than-average zones in CF and RF may cause a few extra line drive singles, but may reduce some doubles and triples if the fielders are deep. Conversely, they could also go the other way if the fielders are cheating (either CF toward the Monster or RF toward the corner). Perhaps the most interesting scenario would be a LF shaded toward the Monster against a righty power hitter. Many of the great righty hitters have ample power to right-center, especially if pitched outside (as most are). This duality probably increases the difficulty of playing CF in Fenway.

Much of this is speculation, of course, and I'll try to take some time to look at past seasons to see if any solid trends emerge.

The bottom line, however, is that I don't think that either UZR or FB ratings accurately account for the idiosyncracies of fielding in the Fenway outfields. I believe that Dewan has made some adjustments on fly balls, but I don't believe he has done anything with CF or RF zones. I have no idea about UZR, but just a glance at the numbers that EV provided would seem to indicate that UZR doesn't correct as much as Dewan does.

I think it's a much fairer assessment of the 2009 team to suggest that the defense accounted for maybe 20 or 25 runs below the league average. If you give the defense full penalty for all of the +64 1Bs and +18 2Bs in the chart above, that still only accounts for 37 runs (using standard LWTS values). I also think that adding the new fielders will help the infield, perhaps to the tune of a 20-run swing, but will have much less effect on the outfield defense than predicted. Even the infield numbers may be smaller, given the fact that the Sox pitchers allowed 120 fewer groundballs than the league average last season, and adding Lackey is likely to increase that gap. My general rule of thumb with UZR and FB fielding runs is that if you cut them in half, you're probably in the ballpark, so we may see a 40-run swing, probably less.

[Papi's Mango Salsa]

What about the offense? If we are saving 90 runs on defense because we have great defensive guys, what happens if we don't score as many runs because they are a drop off on offense. Isn't there a netting that should also be looked at. If the 90 runs translates into 9 more games won, but we give back games because on lack of run creation, we haven't really changed much, have we?

[Mugsys Jock]

Assists are a very crude metric, and one year's worth of numbers is probably worthless, but for poops and giggles: based on last year's numbers, upgrading shortstop and third base is worth +106 assists.

2009 Assists

Position	Player	Assists
SS	Nick Green	198
SS	Alex Gonzalez	107
SS	Jed Lowrie	52
	Total SS	357
SS	Marco Scutaro	421
	Difference	+64
3B	Mike Lowell	174
3B	Nick Green	4
3B	Jed Lowrie	4
	Total 3B	182
3B	Adrian Beltre	224
	Difference	+42

Is this missing Youkilis' 3B assists last year?

[Bowlerman9]

Is this missing Youkilis' 3B assists last year?

Yup. And Lugo & Company for SS.

Last year's numbers:

Sox SS: 413

Sox 3B: 287

[PrometheusWakefield]

Out of curiosity, I ran through what a 90 run differential might look like in terms of impact on our pitchers ERAs. Assuming that all runs saved are equally distributed, and assuming that the 90 runs saved come out of earned rather than unearned runs (both highly questionable assumptions, of course), this is what I got:

Red Sox pitchers if we had 90 runs better Defense

	IP	ER	Adj. ER	ERA	Adj. ERA
Lester	203	77	64.28	3.41	2.85
Beckett	212.33	91	77.70	3.86	3.29
Wakefield	129	66	57.92	4.60	4.04
Penny	131.67	82	73.75	5.60	5.04
Matsuzaka	59.33	38	34.28	5.76	5.20
Papelbon	68	14	9.74	1.85	1.29
Okajima	61	23	19.18	3.39	2.83
Bard	49.33	20	16.91	3.65	3.09
Delcarmen	59.67	30	26.26	4.52	3.96

If the numbers end up anywhere close to that, it would make for a very exciting year...

EDIT: Heh. Looking over those numbers it became obvious that what you're seeing is a .56 decline in ERA regardless of innings pitched. Which, in retrospect, I could have calculated arithmatically.

Edited by PrometheusWakefield, 13 January 2010 - 09:57 PM.

[Rough Carrigan]

I think it's possible.

In 2007, Tampa Bay gave up 944 runs.

In 2008, Tampa Bay gave up 671 runs.

That's an amazing 273 run improvement, which an article at Baseball Prospectus attributed roughly half to defensive improvemnent (going from a 50+ years of MLB low defensive efficiency in 2007 to a top of the league defensive efficiency in 2008) and half to improved pitching.

So, you could say that Tampa improved about 135 runs on defense in 2008.

Of course, the Red Sox didn't have the worst team defensive efficiency since records are available post 1954 as the Rays did. But they were third worst in MLB last year (before park adjustments), only beating the juggernauts of Houston and KC (yes, the Dayton Moore "process").

And the Sox have made big defensive turnarounds before. They were near the bottom of the league in 2006 and then near the top in their championship year in 2007. 80 runs is a lot but seems possible given where they're coming from and where they might go.

[Vermonter At Large]

To elaborate a little bit on my analysis of batted balls earlier, I compiled 5 years worth of data, adding the principle fielders on each team for texture:

CODE

2009 1B   2B   3B   HR
-----------------------
GB  +21   -6   -1   NA  Youkilis, Pedroia, Green, Lowell
FB  +24  +23   -2   -7  Bay, Ellsbury, Drew
LD  +18   +1   -2   -4
---------------------
Tot +63  +18   -5  -11


2008 1B   2B   3B   HR
-----------------------
GB   -7  -13   -1   NA  Youkilis, Pedroia, Lugo, Lowell
FB   +9  +27   -5   -3  Ramirez, Crisp, Drew
LD   +6  -13   -4   -1
---------------------
Tot  +8   +1  -10   -4

2007 1B   2B   3B   HR
-----------------------
GB  -24  -14   -2   NA  Youkilis, Pedroia, Lugo, Lowell
FB   +7  +14   -5  -10  Ramirez, Crisp, Drew
LD  -15   -1   -3   -2
---------------------
Tot -32   -1  -10  -12

2006 1B   2B   3B   HR
-----------------------
GB  +14   -5   -2   NA  Youkilis, Loretta, Gonzalez, Lowell
FB  +17  +20   -2   +3  Ramirez, Crisp, Nixon
LD  -16   -3   -0   -4
---------------------
Tot +15  +12   -4   -1

2005 1B   2B   3B   HR
-----------------------
GB  +20  +10   -2   NA  Millar, Bellhorn, Renteria, Mueller
FB  +16  +35   +9  -22  Ramirez, Damon, Nixon
LD   +3   +8    0   +1
---------------------
Tot +39  +53   +7  -11

There is a lot to look at here. A few observations:

1. 1B/GB numbers seem to be a major reflextion of infield range. Note that the 2009 team was fairly comparable to the 2005 team, which was a fairly ugly infield as well. We can also see how good the infield of Youkilis, Pedroia, Lugo and a healthy Lowell was, perhaps having forgotten that despite his lapses of concentration, Lugo was a very rangy SS.

2. FB rates appear to be almost exclusively park-related. There may be some specific instances of Manny or Damon screwing up a pop fly, but the Monster yields a high number of both singles and doubles on fly balls.

3. 2B/GB rates appear to be isolated largely to corner infielders in general, and to first basemen more specifically in Fenway. In every season since 2006, the Monster appears to have kept the rate in the negative numbers, with a fairly small variation. 2005's rate of +10 can probably be linked to some fairly awful play at first base, particularly Millar.

4. Line drive rates may not yield as much information as we might hope. Something like 73% of all line drives end up as base hits, so this is hardly surprising. Line drive range is very dependent on positioning and luck, especially in the infield. Still, the best 1B/LD rates in this five-year period were from the early Youkilis=Pedroia-Lowell teams, which is probably no accident. but as Lowell was restricted in 2008 and 2009, the numbers jumped up. Most line drives for singles are probably therefore infield-related, although left field probably also has some influence here. The jump to +18 in 2009 could therefore have been the difference between Manny and Bay in some part, although we would have to swallow the notion fhat Manny snagged five to ten more LD's than Bay, which is a pretty big leap. It is possible, however, since Manny played much shallower.

5. Line drive doubles rates show very little variation and probably vary no more than +/- five runs through normal karma. It can be noted that the three seasons of the five in which the Sox gave up 2B?LD rates below the league average were the three years in which Crisp was the primary CF. It is possible that Crisp was, therefore, perhaps five runs better than Ellsbury/Damon in reaction and tracking of line drives.

So again, I think the difference between a great infield and a poor one is in the general range of +/- 20 runs, based on GB range and to some extent, 1B/LD rates. I do not believe that there is an awful lot of variation among outfielders with regards to tracking fly balls. In Fenway, the principle influence on FB hit rates is the Monster. Weaknesses in fly ball defense are going to be manifested largely on the road, therefore. There is fairly good evidence that great fielders can affect line drive run rates, but again those are probably influenced more by the Monster and are largely karmic. I think the variation between great and poor defense of line drives could be plus or minus 20 runs as well, more likely about half of that.

[Bellhorn]

This team took 4 positions out of 7 (excluding pitcher and catcher, where fielding is the smallest component of their run prevention) and replaced below average defense with average to great defense, as well as upgraded the rotation with Lackey. While the exact number is up for debate, I am surprised that you think that the team is likely to allow roughly the same number of runs. I certainly understand taking defensive metrics with a grain of salt, but unless you think that the pitching is going to regress greatly from last year, why wouldn't this team allow significantly fewer runs even ignoring the metrics? Better defense and better pitching should equal fewer runs, right? Plus, at least in terms of the pitching, the fact that Sox pitchers had better FIPs than their actual earned run results suggests that they shouldn't regress in front of a better defense.

Pure speculation here, but I wonder if there might be a slightly disproportionate benefit from having an above average fielder at every position, kind of like the non-linear effect of adding an above-average hitter to a line-up that is already strong. Here I'm thinking it would be more of a function of pitcher confidence - they can think "I can go ahead and just make my pitch, no matter where it gets hit, if it's catchable my teammates will take care of it" as opposed to thinking, on some subconscious level, "I've mostly got good fielders behind me, but God I hope they don't hit another ground ball to the left side."

[Worst Trade Evah]

Well ... let's take a look at run attribution from a different slant.

It is possible to get a qualitative look at defensive run attribution by analyzing hits allowed (1B, 3B, 3B and HR) against hit trajectory (GB, FB, LD). The general notion is that by looking at, for instance, the team's 1B/GB rate against the league average, we can get a rough estimate of the singles that could be attributed to the team defense.

First, though, we need to equalize the numbers a bit, since the Sox pitching staff gave up GB, FB and LD rates quite different from the league average. In fact, the Sox pitchers gave up fewer GB's per BIP than the league average (41.5% to 43.8% league average), more fly balls (38.8 vs 35.9) and roughly the same number of line drives (18.9 vs 19.2). In total the Sox pitchers gave up 44 fewer BIPs than the league average, so in general, the Sox staff did a pretty decent job overall last season.

To equalize the rates, then, we calculate the 1B, 2B, 3B and HR per BIP rates, then multiply them by the league BIP average to yield something resembling a BABIP-neutral assessment of hits allowed for each of the hit trajectory. Here are the results for the 2009 Red Sox compared to the league average:

CODE

     1B   2B   3B   HR
-----------------------
GB  +21   -6   -1   NA
FB  +24  +23   -2   -7
LD  +18   +1   -2   -4
---------------------
Tot +63  +18   -5  -11

Looking at ground balls first, I think that it is safe to assume that the Sox 2009 infield was a bit on the pourous side, allowing 21 more singles than the league average. There are no real mitigating factors other than karma to adjust this rate, and it is probably safe to assume that we can assign all of these extra singles to shoddy defense, or roughly 10 runs.

I think that the six fewer doubles on ground balls can be largely attributed to the dimensions of Fenway. Hot grounders hit to the right or left field corners would account for nearly any extra base hit on a ground ball, but the Monster probably prevents many of these hot grounders down the left field line from becoming doubles. In general, however, I would find it difficult to penalize a fielder too badly for a ball hit hard down the line.

Fly ball analysis is a bit trickier. The Sox allowed +24 more singles on FBs than the league average, and to get a better picture of the skew, the numbers were 101 against Boston vs 77 against the league average, so the skew would probably be statistically higher if the Fenway numbers were deducted from the overall set. There are essentially two types of FBs that land for singles - seeing eye pop flies and balls hit off the Monster. I suppose that a very small percentage of these could be attributed to a fielder, but most are clearly karmic.

Similarly, the +23 doubles on fly balls were probably primarily Monster shots.

Line drive analysis is a little trickier. We should logically expect that the Monster will increase singles to some extent, and probably reduce doubles (unlike fly balls) by the same general rate (thus converting balls that would be doubles in a bigger park into wall-ball singles). I think that there are potentially more subtle effects too. A left-fielder in Fenway generally plays much shallower than in other parks. If positioned correctly, the left fielder may be able to turn an occasional potential line drive single into an out, but I also think that playing shallower limits the outfielders range to his left or right, so may cause some potentially catchable balls to go to the wall. I also think that the larger-than-average zones in CF and RF may cause a few extra line drive singles, but may reduce some doubles and triples if the fielders are deep. Conversely, they could also go the other way if the fielders are cheating (either CF toward the Monster or RF toward the corner). Perhaps the most interesting scenario would be a LF shaded toward the Monster against a righty power hitter. Many of the great righty hitters have ample power to right-center, especially if pitched outside (as most are). This duality probably increases the difficulty of playing CF in Fenway.

Much of this is speculation, of course, and I'll try to take some time to look at past seasons to see if any solid trends emerge.

The bottom line, however, is that I don't think that either UZR or FB ratings accurately account for the idiosyncracies of fielding in the Fenway outfields. I believe that Dewan has made some adjustments on fly balls, but I don't believe he has done anything with CF or RF zones. I have no idea about UZR, but just a glance at the numbers that EV provided would seem to indicate that UZR doesn't correct as much as Dewan does.

I think it's a much fairer assessment of the 2009 team to suggest that the defense accounted for maybe 20 or 25 runs below the league average. If you give the defense full penalty for all of the +64 1Bs and +18 2Bs in the chart above, that still only accounts for 37 runs (using standard LWTS values). I also think that adding the new fielders will help the infield, perhaps to the tune of a 20-run swing, but will have much less effect on the outfield defense than predicted. Even the infield numbers may be smaller, given the fact that the Sox pitchers allowed 120 fewer groundballs than the league average last season, and adding Lackey is likely to increase that gap. My general rule of thumb with UZR and FB fielding runs is that if you cut them in half, you're probably in the ballpark, so we may see a 40-run swing, probably less.

I must be missing something obvious. The Red Sox staff here is being compared to a league average staff, and some small differences noted, which may or may not be attributable to defense. But you seem to be using the league average staff as a comparable baseline to the Red Sox staff, when in fact it's very likely the Red Sox staff was significantly better than league average -- but made to look a bit more like league average because of the crappy defense. Aren't you significantly underestimating the hits on defense here? For example, you say the Sox allowed 21 more singles on defense -- but have you taken into account not just the extra singles compared to league average, but the fact that the Sox pitchers were not, on average, league average?

[pokey_reese]

Pure speculation here, but I wonder if there might be a slightly disproportionate benefit from having an above average fielder at every position, kind of like the non-linear effect of adding an above-average hitter to a line-up that is already strong. Here I'm thinking it would be more of a function of pitcher confidence - they can think "I can go ahead and just make my pitch, no matter where it gets hit, if it's catchable my teammates will take care of it" as opposed to thinking, on some subconscious level, "I've mostly got good fielders behind me, but God I hope they don't hit another ground ball to the left side."

I would never even attempt to quantify it, but I believe that a pitcher's confidence in his defense is at least somewhat of a factor. I'm not sure how much it changes the actual results on a pitch-by-pitch basis, but there were certainly times last year (I am thinking about a game or two against TB) when Lester and Beckett seemed really amazed/frustrated by balls that looked like sure outs that leaked through the left side of the infield. It may not actually cause a pitcher to try and strike more guys out or pitch to more contact, but I agree that the comfort and confidence that come from having a great defense behind them can't hurt a pitching staff. That said, I think that we have seen that the non-linear effect of adding one more good defender to an above-average group actually slopes the other way in terms of statistical run prevention, but there is no way to incorporate pitcher mentality into that.

[Vermonter At Large]

I must be missing something obvious. The Red Sox staff here is being compared to a league average staff, and some small differences noted, which may or may not be attributable to defense. But you seem to be using the league average staff as a comparable baseline to the Red Sox staff, when in fact it's very likely the Red Sox staff was significantly better than league average -- but made to look a bit more like league average because of the crappy defense. Aren't you significantly underestimating the hits on defense here? For example, you say the Sox allowed 21 more singles on defense -- but have you taken into account not just the extra singles compared to league average, but the fact that the Sox pitchers were not, on average, league average?

Well ... for the purpose of this exercise we are trying to separate the fielding from the pitching. The main tool we're using is hits per unit of ground balls, fly balls, and line drives, which is actually a percentage. We could have used those abstract rate percentages, but I thought that hits and runs were more germane to the discussion, so I multiplied those rate percentages times the league average number of events.

In fact, the Red Sox gave up 6 singles fewer on ground balls than the league average last season, but Sox pitchers allowed 120 ground balls fewer than the league average which worked out to a 1B/BIP percentage of .228, compared to the league average of .217. So to neutralize the pitching, I multiplied the .228 times the league average ground ball totals. In other words, although the Sox gave up 414 singles on ground balls last season, they would have given up 441 had they fielded the average number of ground balls that the league faced. The league average was 420 singles on ground balls, so the Sox defense was +21, even t hough the actual number was -6. We could also say that the pitching staff saved the Sox 27 runs by not allowing more ground balls to be hit to that crappy defense.

Make sense?

[Frisbetarian]

I think it's a much fairer assessment of the 2009 team to suggest that the defense accounted for maybe 20 or 25 runs below the league average. If you give the defense full penalty for all of the +64 1Bs and +18 2Bs in the chart above, that still only accounts for 37 runs (using standard LWTS values). I also think that adding the new fielders will help the infield, perhaps to the tune of a 20-run swing, but will have much less effect on the outfield defense than predicted. Even the infield numbers may be smaller, given the fact that the Sox pitchers allowed 120 fewer groundballs than the league average last season, and adding Lackey is likely to increase that gap. My general rule of thumb with UZR and FB fielding runs is that if you cut them in half, you're probably in the ballpark, so we may see a 40-run swing, probably less.

A major part of your difference with UZR and +/- is that you do not include the out value in your run calculation while they do. I'm not sure where you are getting 37 runs for 64 singles and 18 doubles, as the standard lwts value for a single is .48 and double is .77. This totals 44.6 runs in the above example, but UZR and +/- would also add an out lost value to those hits, bringing the total to just over 69 runs. I am, as you well know, of the opinion that on every play there has to be an outcome. On a ground ball hit to the infield, that outcome will usually be an out or a single. If it is a single, that is +0.48 runs for the offense, and if it is an out that is approximately -0.30 runs for the offense - there is a 0.78 run difference depending on the outcome of the ground ball. I realize by now that you will never be convinced this is correct, and absolutely have no desire to be drawn into this discussion with you again, but did feel it was relevant to point out where a large percentage of the discrepancy between your numbers and the defensive metrics is coming from.

Great work on the charts, btw. Really informative.

although the Sox gave up 414 singles on ground balls last season, they would have given up 441 had they fielded the average number of ground balls that the league faced. The league average was 420 singles on ground balls, so the Sox defense was +21, even t hough the actual number was -6. We could also say that the pitching staff saved the Sox 27 runs by not allowing more ground balls to be hit to that crappy defense.

Did you mean runs here or plays?

[jk333]

I think it's possible.

In 2007, Tampa Bay gave up 944 runs.

In 2008, Tampa Bay gave up 671 runs.

That's an amazing 273 run improvement, which an article at Baseball Prospectus attributed roughly half to defensive improvemnent (going from a 50+ years of MLB low defensive efficiency in 2007 to a top of the league defensive efficiency in 2008) and half to improved pitching.

So, you could say that Tampa improved about 135 runs on defense in 2008.

There's certainly credence to this point, both defense and pitching play important roles in determining how many runs a team will give up.

But, doesn't luck and competition (read offense) also play into this? Is it fair to assume that the 273 run improvement is 100% do to parameters that the Rays pitchers and defense controlled? If I set up a team of Josh Beckett's and had them backed with a team of Mike Cameron, Adrian Beltre's, and Adam Everett's but they were pitching to a team of 2009 NYY's in NYY stadium, would we not get elevated runs allowed? Would people be knocking the defense? the pitching?

I appreciate that defense and pitching are the two biggest factors... but I do think competition and luck are perhaps undervalued at times. I hadn't seen it brought up in this thread.

I also compiled a spreadsheet to look at the above a bit but don't have anything of note to report.

[Vermonter At Large]

A major part of your difference with UZR and +/- is that you do not include the out value in your run calculation while they do. I'm not sure where you are getting 37 runs for 64 singles and 18 doubles, as the standard lwts value for a single is .48 and double is .77. This totals 44.6 runs in the above example, but UZR and +/- would also add an out lost value to those hits, bringing the total to just over 69 runs. I am, as you well know, of the opinion that on every play there has to be an outcome. On a ground ball hit to the infield, that outcome will usually be an out or a single. If it is a single, that is +0.48 runs for the offense, and if it is an out that is approximately -0.30 runs for the offense - there is a 0.78 run difference depending on the outcome of the ground ball. I realize by now that you will never be convinced this is correct, and absolutely have no desire to be drawn into this discussion with you again, but did feel it was relevant to point out where a large percentage of the discrepancy between your numbers and the defensive metrics is coming from.

Great work on the charts, btw. Really informative.



Did you mean runs here or plays?

Hah, I knew this would come up sooner or later ... I fully understand the logic of double-counting the run penalty for botching a play that got turned into a double when an out should have been made. I don't think it's right for many different reasons, but I really don't want to sidetrack this thread into that snake pit (even though mid-winter is the time to do it, lol). I do prefer working in terms of plays (and to answer your last question, I did mean plays) because I think the run conversion clouds the picture. Rather than looking qualitatively at the complex relationship between fielding and pitching, we just assign run values and leave it at that.

When I have some time, I'll try to do a break-down on the 2007-2008 Rays teams that were mentioned earlier in the thread to see if we can delve a little deeper into that relationship.

[SoxFanPJ]

Without getting specifically in depth into his methodology or the worthiness of the stats he is using. This article is a follow up of sorts to one that appeared in December describing dramatic improvements by Detroit and Seattle.

In 2009 the Tigers rediscovered defensive religion. They realized that what they tried in 2008 simply doesn't work in major league baseball. All offense and no defense doesn't cut it. A team needs balance. Inge went back to third base, and the Tigers brought in one of the best defensive catchers in baseball, Gerald Laird, to take his place behind the plate. At shortstop, the offensive-minded but mediocre defensively Edgar Renteria was sent packing. He was replaced by the best defensive shortstop of the 21strrd century so far, Adam Everett. (That is my assessment based on my defensive metrics outlined in The Fielding Bible and The Fielding Bible, Volume II.)

What happened? The Tigers improved by 29 runs defensively at third base. They improved by 17 runs at shortstop. They improved their win total by 12 games and won 86 games for the season. They just missed the playoffs by losing the 163rd game of the year to the Twins, but they spent more time in first place in 2009 than any other team in the American League.

The Seattle Mariners also were the most improved defensive team in baseball this past season. This is not a coincidence. They improved their Defensive Runs Saved as a team from 17 runs saved in 2008 to 109 runs saved in 2009. That's an increase of 92 runs in one year....

How did the Mariners do it? Outfield defense is the main answer. In 2008 they had only 11 runs saved defensively from their entire outfield. Defensively, all they had was Ichiro. He split time between right field and center field. In right field he saved 15 runs defensively, but in center field he was pretty much average (-1 run saved). In 2009 the team brought in two of the best defensive outfielders in the game to join Suzuki.

http://www.actasports.com/sows.php

The Mariners example may be closest to the Redsox current improvement

The Mariners went from

2008
C-Jojima (112G)
1B-Sexson (74G)
2B-Lopez (159G)
SS-Betancourt (153G)
3B-Beltre (143G)
LF-Ibanez (162G)
CF-Reed (97G)
RF-Ichiro (162G)

2009
C- Rob Johnson (80G)
1B-Branyan (116G)
2B-Lopez (153G)
SS-Betancourt (63G) Ronny Cedeno (59G)
3B-Beltre (111G)
LF-Balentien (56G) Endy Chavez (54G)
CF-Gutierez (153G)
RF-Ichiro (146G)

[Rudy Pemberton]

It's an interesting take; another one may simply be that the Mariners and Tigers were extremely lucky last year. They both allowed more runs than they scored. Now, did their improved defense a factor in winning more games "than they should have"? How much of the increased run prevention was better pitching (obviously, the two are related, but they did add Jackson and Porcello, much better year from Verlander, etc.) I guess we'll find out more this year but it wouldn't surprise me in the least if Detroit and Seattle lost more games next year than they did this year.

[paulftodd]

How much of fielding defense is positioning and pitchers hitting their locations (positioning in part is based on pitch location). I expect coaches have some influence on where fielders position themselves, so maybe in part it is a coaching thing. When you look at the wild year to year swings some players have, you have to believe something beyond the fielders talent is involved.

I am a bit of a skeptic about the Red Sox being 80-90 runs better due to fielding alone, although if the starting pitching is healthy and gets off to a good start (unlike last year), they could be 80-90 runs better from pitching alone. Not saying the fielding is not improved, it is, how much is the big question.

Projections unfortunately are subject to the whim of the following Law (not sure where I read this and I have amended it some) "Under the most rigorously observed conditions of skill, age, statistical rules and other variables, a ballplayer will perform as he damn well pleases, health permitting, and how well or poorly depending on if he is helped or hindered by the wind and/or the Gods of BABIP (this applies to fielding as well)"

As for the Tigers and Mariners great fielding, at least we know great fielding guarantees an 85-86 win season. And 5 of the 6 best defensive AL teams did not even make the playoffs. Pitching and hitting say Hi.

[Bellhorn]

But, doesn't luck and competition (read offense) also play into this? Is it fair to assume that the 273 run improvement is 100% do to parameters that the Rays pitchers and defense controlled? If I set up a team of Josh Beckett's and had them backed with a team of Mike Cameron, Adrian Beltre's, and Adam Everett's but they were pitching to a team of 2009 NYY's in NYY stadium, would we not get elevated runs allowed? Would people be knocking the defense? the pitching?

I appreciate that defense and pitching are the two biggest factors... but I do think competition and luck are perhaps undervalued at times. I hadn't seen it brought up in this thread.

I think we're pretty well justified in assuming that luck plays a pretty minimal role over the course of a team-season. When talking about pitcher BABIP, we tend to pay most attention to extreme outliers in either direction, so it's easy to forget that the overall distribution of BABIP for starting pitchers over a season is not really all that wide - and what variation there is is largely explained by BABIP skill, park effects, and quality of both the pitcher's defense and the opposing batters faced. And that's over perhaps 200 innings, let alone the 1400+ played by a team in a given season. Still, if there ever were a case for luck playing an appreciable role, it would be one like this, where a team had historically poor results one year, and well above average results the next.

Opponent quality might be an important consideration if you're comparing between teams in different divisions, but for comparing the same team across different seasons, it seems like more trouble than it's worth, unless there is an obvious extreme change in extra-divisional schedule or personnel within the division.

[The Allented Mr Ripley]

Last night Bill James commented on the Sox' approach this winter with regards to defense:

"I believe it's accurate to say that it was our perception that that was where the value was in this year's market, in this year's set of conditions. It also had to do with the needs of last year's team. Last year's team needed some defense, we had to invest in some defense, and the market seemed pretty good for it. But to say that's the new thing and it will be that way from now on, I wouldn't do that.''

So, was this more of a temporary correction?

"Well, it's in the line of march,'' James said. "I think we understand we've had good defensive metrics now for five or six years. When I started with the Red Sox we didn't have them, we had kind of primitive ones. We've got pretty good ones now for several years. It has reached the point at which not only us but a lot of teams are confident about that now and are starting to let the money flow toward gloves, which is a good thing."

I'm curious to know what their metrics are, since there's still such debate over the incomplete picture most of them paint one way or another. I'm assuming it's a combination of multiple methods.

[paulftodd]

I'm curious to know what their metrics are, since there's still such debate over the incomplete picture most of them paint one way or another. I'm assuming it's a combination of multiple methods.

Whatever it is, I like the uncertainty expressed
here

I’ve been doing the [offensive] stuff all my life. I know what’s a normal gap between two seasons [offensively] and what isn’t. I don’t know the same [defensively].

snip

“What I’m most curious about in 2010 is how much better we’re going to be defensively,’’ he said. “I don’t think anyone questions that we’re going to have a better defensive team. But are we going to be as much better defensively as we want to believe we are, and is that going to have as much impact on [the pitching staff] as we hope it does?

[Vermonter At Large]

There's certainly credence to this point, both defense and pitching play important roles in determining how many runs a team will give up.

But, doesn't luck and competition (read offense) also play into this? Is it fair to assume that the 273 run improvement is 100% do to parameters that the Rays pitchers and defense controlled? If I set up a team of Josh Beckett's and had them backed with a team of Mike Cameron, Adrian Beltre's, and Adam Everett's but they were pitching to a team of 2009 NYY's in NYY stadium, would we not get elevated runs allowed? Would people be knocking the defense? the pitching?

I appreciate that defense and pitching are the two biggest factors... but I do think competition and luck are perhaps undervalued at times. I hadn't seen it brought up in this thread.

I also compiled a spreadsheet to look at the above a bit but don't have anything of note to report.

I don't have time to explain the metrics too deeply at this time, but I attribute that 273-run improvement from the 2007 to 2008 Rays as something like this:

92 Runs from Home Runs
51 Runs on balls in play attributable to pitching.
98 Runs on balls in play attributable to fielding.
10 Runs on non-contact events (walks, strikeouts, balks, wild pitches, etc attributable to pitchers)
25 Runs unaccounted for (variation, luck, etc.)

The basic methodology incorporates computing true defense-independent pitching runs from balls in play using league average baselines for the various BIP events and linear weights. This TDIPS/BIP value decreased from 428 to 377 runs. Pitcher's HR Runs also dropped from 354 to 262.

Fielding attibution is the difference between actual BIP runs and baseline pitching runs. These values dropped from +101 runs (above the league average) in 2007 to +3 runs. That improvement included a 69-run improvement in fielding ground balls, an actual 21-run decline in fielding fly balls, and a 45-run improvement in fielding line drives.

The 69-run improvement (plus 48 runs in 2007 to minus 21 in 2008) is difficult to assess from a quick glance, although Bartlett was a significant improvement over Brendan Harris, and Longoria also an improvement over Ty Wiggington, but that alone shouldn't have been that big of a difference. We can look more deeply at the numbers to try to attibute, but it could have been pitching-related as well (lots of hard ground balls).

The line drive improvement is also hard to assess. Some of it probably related to Longoria, but a lot of line drive fielding is random (i.e. positioning-related).

The fly ball numbers were pretty high in 2008, and I don't know exactly why. It could have been slanted by a couple of pitchers getting hammered in Fenway which has a huge park factor for fly balls (+25 runs for the Red Sox).

I'll try to isolate some of the specific causes over the weekend.