PDA

View Full Version : Jr. or Manny, who's better historically?



cincrazy
06-02-2008, 03:25 PM
http://proxy.espn.go.com/chat/chatESPN?event_id=20905&lpos=spotlight&lid=tab4pos2


THE CASE FOR RAMIREZ
We've all spent so much time focusing on what a knucklehead Manny can be that we?ve overlooked something slightly more important: He's one of the greatest right-handed hitters who ever lived.

I'll get to some of those numbers as we roll through this debate. But of more direct relevance to this particular topic, you can certainly argue that Ramirez was a more complete hitter than Griffey (.312 average to Griffey's .289, .408 on-base percentage to Griffey's .373, .591 slugging percentage to Griffey's .550). And it's impossible to dispute that Manny has had the steadier career--even if "steady" isn't a word that should ever be used in the same sentence as "Manny" without serious soul-searching.

THE CASE FOR GRIFFEY
Well, let's start with some basic mathematics. Last time I checked, 600 was a lot bigger number than 500. Manny was the 24th name in the ever-swelling 500-Homer Club. Griffey will be joining only Barry Bonds, Hank Aaron, Babe Ruth, Willie Mays and Sammy Sosa in the 600-Homer Club. Big difference.

But of course, Griffey also has a few other attributes. Like 10 consecutive Gold Gloves, a unanimous MVP award, three Players Choice awards, a spot on the All-Century Team and another spot on the all-time Rawlings Gold Glove team. Among other things.

THE VERDICT
With all apologies to the Manny Being Manny Fan Club, I deduct points for all those jogs to first base, week-long midseason vacations and inability to fit spring training into our man Manny's busy schedule. Plus, I guess I missed the news about those 10 Gold Glove trophies Manny secretly won. So I'm going with Junior Griffey, one of the greatest and most charismatic players of modern times. But if you disagree, hey, that's what we're here for.

Quite frankly, I don't even think this is up for debate. If you take away Griffey's injury plagued stint here in Cincy, it certainly isn't up for debate. One of the greatest defender's of our era in his prime, the sweetest swing you'll ever see, and a complete all around player. Manny is a slugger, but what else is he? Good defensively? No way. Good baserunner? HA! He's known for taking it easy, sitting out crucial games, etc. etc.

Must've been a slow news week for espn.com, I don't think this really merits a discussion. Anyone disagree?

Cyclone792
06-02-2008, 03:38 PM
I haven't looked at this in any detail so this is all off the top of my head, but ...

I'm sure I'll take some flogging for this, even though it's off the top of my head ... but I think Ramirez may actually have been a slightly greater hitter than Griffey. Griffey has more career value right now, but Manny might have him in peak value as far as hitting goes. And Manny is putting up a better offensive season this season than Griffey so Manny may age better (he's a bit younger too).

Overall player goes to Griffey though and it's not close. Once you account for defense and positional advantage, whatever hitting advantage Ramirez might have (again, not sure) disappears quickly.

cincrazy
06-02-2008, 03:43 PM
I haven't looked at this in any detail so this is all off the top of my head, but ...

I'm sure I'll take some flogging for this, even though it's off the top of my head ... but I think Ramirez may actually have been a slightly greater hitter than Griffey. Griffey has more career value right now, but Manny might have him in peak value as far as hitting goes. And Manny is putting up a better offensive season this season than Griffey so Manny may age better (he's a bit younger too).

Overall player goes to Griffey though and it's not close. Once you account for defense and positional advantage, whatever hitting advantage Ramirez might have (again, not sure) disappears quickly.

You're dead on with pretty much everything, especially the offensive comparison. Ramirez does rank slightly higher in most of the key categories outside of home run's. And clearly Ramirez has aged better. But I also think that can be directly related to the style of play of the two. Griffey has taken a pounding diving into the gaps on Astroturf, and crashing into the outfield walls giving it 100% max effort. I'm not sure anyone can say that about Manny when all is said and done.

If Griffey would've played his whole career along the storyline "Griffey being Griffey," perhaps he would have avoided the injuries, and racked up more career stats. That's the funny thing about the game of baseball. Sometimes the people that play it the right way get punished a hell of a lot more harshly than the one's that don't.

PuffyPig
06-02-2008, 03:45 PM
A GG CF is worth quite abit more than a poor fielding LF, even if that LF is a better overall offensive hitter.

Griffey has been the better player over his career for that reason alone.

Mario-Rijo
06-02-2008, 03:48 PM
I haven't looked at this in any detail so this is all off the top of my head, but ...

I'm sure I'll take some flogging for this, even though it's off the top of my head ... but I think Ramirez may actually have been a slightly greater hitter than Griffey. Griffey has more career value right now, but Manny might have him in peak value as far as hitting goes. And Manny is putting up a better offensive season this season than Griffey so Manny may age better (he's a bit younger too).

Overall player goes to Griffey though and it's not close. Once you account for defense and positional advantage, whatever hitting advantage Ramirez might have (again, not sure) disappears quickly.


But what effect does the Green Monster play into the discussion I would wonder? Right off the top of my head that's always been considered a built in advantage for RH hitters who play there.

Marc D
06-02-2008, 03:56 PM
I see they used BA......JR had a career BA over .300 untill he hit the decline. Lets see what Manny's final numbers are when he plays a few years after his prime (if he does) before making a comparison (not that there really is any).

typical ESPN bubble gum garbage imo.

RedsManRick
06-02-2008, 04:19 PM
Overall, I think Manny has been a more productive offensive player during the time he's played (and it's not terribly close), but Junior's defensive contributions easily make up the difference and more.

Of course, you have to consider that Junior has 1800 more plate appearances than Manny. He started his career 3 years younger than Manny and has made it to his age 38 season. Manny may not make that far, but even if he does, he'll not likely play long enough to offset Junior's playing time advantage. (imagine if Junior had been healthy!) Manny appears to be in just the 2nd year of offensive decline, so over the next few years, his counting stats will continue to accumulate, but his slash stats will regress a bit.

http://img175.imageshack.us/img175/7811/58273364xu1.jpg

A quick and dirty chart shows Manny with a pretty healthy advantage over most of their careers. Interestingly, if you use OPS+ instead of just OPS, you see Junior's earlier career is even more impressive, as it came before the long ball explosion of the mid 90's.

http://img385.imageshack.us/img385/8488/10939665bz9.jpg

If Griffey is something like 90% of the offensive player Manny is, his defense would probably add another 40%, easily eclipsing the overall production.

dabvu2498
06-02-2008, 04:21 PM
But what effect does the Green Monster play into the discussion I would wonder? Right off the top of my head that's always been considered a built in advantage for RH hitters who play there.

OPS+ is park-adjusted, I believe.

KGJ: 139
Manny: 154

I've had Red Sox fans tell me that the Monster has turned quite a few line drive home runs into doubles (or more likely with Manny, singles). Don't know how true that is.

SunDeck
06-02-2008, 04:28 PM
I can only imagine how these two would compare if Giriffey had played CF in NY or Boston. No, I take that back, there would be no comparison. Even with the injuries, Griffey is simply one of the best ballplayers who ever lived. Manny is a hitter. He's a great one, perhaps one of the best to ever play the game, but the old schooler in me can't past the belief that a ballplayer is more than a hitter.

OldRightHander
06-02-2008, 04:30 PM
I've had Red Sox fans tell me that the Monster has turned quite a few line drive home runs into doubles (or more likely with Manny, singles). Don't know how true that is.

No doubt that happens at times, but how many times is a routine fly ball a double off the wall there because the LF just ran out of room? Some of those line drives high off the wall might be homers at other parks, but that wall sure gives a lot of hits that would have been outs in those other parks.

In their prime, I would have taken Griffey. Take out the injuries, and the counting stats wouldn't even be close.

dabvu2498
06-02-2008, 04:39 PM
No doubt that happens at times, but how many times is a routine fly ball a double off the wall there because the LF just ran out of room? Some of those line drives high off the wall might be homers at other parks, but that wall sure gives a lot of hits that would have been outs in those other parks.



Can't deny that. Note that in my post I said "Red Sox fans" not "rational humans."

Another thing to take into consideration, and I know it's a team thing, is post-season.

Manny has played in 19 total post-season series and 4 World Series. Jr. played in 3 total post-season series.

Both of them have been very good in the playoffs, but let's face it, it's Manny that people have seen in the big games and big series in October.

I'm not saying Manny is a better player than KGJ, just saying that's a possibility as to why it's even a discussion.

blumj
06-02-2008, 04:49 PM
Can't deny that. Note that in my post I said "Red Sox fans" not "rational humans."

Another thing to take into consideration, and I know it's a team thing, is post-season.

Manny has played in 19 total post-season series and 4 World Series. Jr. played in 3 total post-season series.

Both of them have been very good in the playoffs, but let's face it, it's Manny that people have seen in the big games and big series in October.

I'm not saying Manny is a better player than KGJ, just saying that's a possibility as to why it's even a discussion.

It shouldn't really be a discussion, and Stark does a pretty good job dismissing it as one. Now, you can make the case that Manny might be the better hitter, and I do think, when it's all said and done, he probably will have been, but it's kind of a "so, what?" case.

cincinnati chili
06-02-2008, 04:49 PM
If you take each player's best 10 years, Manny was more dangerous in the batter's box without question. But Griffey more than made up for the difference in the outs he took away from the opposition.

In the modern era, it's a lot easier to find a corner outfielder to give you 80% of the production of Manny in his prime than it was to find a center fielder to give you 80% of Griffey in his prime. And I actually think Manny's a better defender (esp. in Fenway) than people realize.

Verdict: Griffey.

The other intangible is that for several years Griffey was the best player in the game. That's never been true for Manny in one single year of his career.

Cyclone792
06-02-2008, 05:17 PM
Here's an interesting comparison between the two with win shares ...

Career Win Shares
Griffey: 400 (I gave him ~10 win shares of credit for the 1994-95 strike)
Ramirez: 367 (I gave him ~6 win shares of credit for the 1994-95 strike)

Top 3 Season Win Shares
Griffey: 97 (Griffey's top season is 36)
Ramirez: 98 (Ramirez's top season is 35)

Top 5 Season Win Shares
Griffey: 155
Ramirez: 155

Top 7 Season Win Shares
Griffey: 212
Ramirez: 209

Top 10 Season Win Shares
Griffey: 285
Ramirez: 286

Top 12 Season Win Shares
Griffey: 325
Ramirez: 330

It's almost impossible to get much closer than that.

All that said, when you consider that Griffey did his work as a center fielder while Ramirez did his work as a left fielder, that pushes Griffey ahead of Ramirez with little doubt in regards to greater overall player.

cincinnati chili
06-02-2008, 05:30 PM
Cyclone,

That's good stuff, and I should look at the data closely sometime when I'm not studying for the bar exam.

Still, while Win Shares are a good quick/dirty tool, I'm skeptical of their precision because there are no such things as negative win shares. I'd be curious to see how they compare in systems such as VORP, with adjustments made for positive and negative fielding runs. My guess is that it wouldn't be so close.

StillFunkyB
06-02-2008, 05:55 PM
Griffey.

Not even close.

Manny is a great player, and one of the better hitters in history. He has always been lazy out in the OF, and running the bases.

If you are talking about just hitting, then it's a little closer but I would still give the edge to Griffey.

Cyclone792
06-02-2008, 06:02 PM
Cyclone,

That's good stuff, and I should look at the data closely sometime when I'm not studying for the bar exam.

Still, while Win Shares are a good quick/dirty tool, I'm skeptical of their precision because there are no such things as negative win shares. I'd be curious to see how they compare in systems such as VORP, with adjustments made for positive and negative fielding runs. My guess is that it wouldn't be so close.

I checked WARP and it leans toward Griffey a bit more, but it's still pretty close. Actually, their WARP per 162 games differed by only about a tenth in Griffey's favor. That may widen a bit as Ramirez enters his decline phase, though it's still hinging on how long each player hangs around.

I did give a very quick glance at positional rankings, and I think they both slot up fairly close historically at each player's respective position. Griffey is easily a top 10 center fielder all-time; I personally think he's 7th behind the Big Five and Charleston ... tough group to crack. I also think Manny will wind up in the top 10 among left fielders, perhaps as high as 7th or so, especially if he doesn't get injured this year since he seems to be having another decent year.

I do try to include Negro Leaguers in rankings too, and that's not much more than an educated guess. Part of me is fairly certain that Oscar Charleston was probably a greater player than Griffey, and another part of me has no idea. It's the same deal with Turkey Stearnes; they were both great, but was Stearnes a better player than Ramirez? Unfortunately, nobody will ever know.

Chip R
06-02-2008, 06:09 PM
I don't know how many times that Manny has been a DH but if it's been more than a few, he's had that advantage that Jr. hasn't had over the last 9 years. Unless it was an interleague game, there was no way a manager could put Jr. in as a DH after he came over here. I realize the Sox have Ortiz but I would think that they have taken the opportunity to play Manny at DH.

One other thing to consider that really doesn't have to do with stats is that Jr. pretty much saved baseball in Seattle. It's an intangible but a huge one, nonetheless.

Highlifeman21
06-02-2008, 09:51 PM
Pre-Cincinnati Griffey trumps Pre-Boston Manny, however Boston Manny trumps Cincinnati Griffey.

As many have said, when you factor in Pre-Cincinnati Griffey's defense compared to Manny's defense at any point in his career, that gives the edge to Griffey.

OldRightHander
06-02-2008, 10:04 PM
I do try to include Negro Leaguers in rankings too, and that's not much more than an educated guess. Part of me is fairly certain that Oscar Charleston was probably a greater player than Griffey, and another part of me has no idea. It's the same deal with Turkey Stearnes; they were both great, but was Stearnes a better player than Ramirez? Unfortunately, nobody will ever know.

I wonder where Josh Gibson would have ranked on the HR list had he been in the majors his whole career. A lot of speculation with those Negro League players.

M2
06-02-2008, 10:20 PM
, while Win Shares are a good quick/dirty tool, I'm skeptical of their precision because there are no such things as negative win shares.

Actually there are, though they're rare. Mathematically that's a good thing since I've yet to see a negative number on an MLB scoreboard. Beats the hell out of basing everything on a shifting and false norm, IMO. Beyond the internal problems with linear weights systems you've got the massively larger problem when you try to mix and match different systems (e.g. for hitting and fielding) in the pursuit of total player value - system A and system B usually give different values to the same events meaning that a "run" in one system ultimately does equal a "run" in the other.

As for the larger question, I expect Manny's next few years will make this a no-brainer, one way or the other. If he continues to terrorize opposing pitchers, then he'll eclipse Jr.'s career. If he doesn't, then Jr.'s early career will be enough to hold sway.

From a completely anecdotal perspective, being a Reds fan living in Boston, it's an easy call for me from an emotional perspective. I've watched Jr. play abysmal defense for most of this decade as his hitting skills diminished and injuries caused him miss huge chunks of time. All this for the worst run for the Reds in my lifetime. Meanwhile I've also watched Manny Ramirez beat the snot out of the ball during the best decade for the Red Sox since the 1910s. Jr. or Manny? To me that's not even a question (in terms of pure entertainment value), Manny going away.

Cyclone792
06-02-2008, 11:10 PM
I wonder where Josh Gibson would have ranked on the HR list had he been in the majors his whole career. A lot of speculation with those Negro League players.

Probably not as high as people think, mostly because he was a catcher and likely wouldn't have accumulated the total games in his career as catchers will break down quicker. Of course, it's also quite possible he'd have been moved from catcher at some point later in his major league career to prolong his stick.

John Holway has some Negro League data on Gibson and has accounted for 224 home runs in roughly 1,000 games. So if Gibson would have had a 2,000 game major league career, it's quite possible that he'd have hit 400+ home runs. Then he'd probably have mixed in a batting average over .320, an on-base percentage over .420, and a slugging percentage over .550. All that as a catcher, a good defensive catcher. That would have been sick.

I do think Gibson is the greatest catcher ever, and I'm pretty comfortable with that ... even as a Reds fan, even knowing all about Johnny Bench. I also think Gibson is probably one of the 15 or so greatest players of all-time, and I'm pretty comfortable with that too. Everything I've read suggests that Gibson and Oscar Charleston are two of baseball's all-time greatest players. I don't doubt that it's true.

It's just such a shame we couldn't ever see the world's best baseball players on the same fields until the second half of the 20th century.

Reds/Flyers Fan
06-03-2008, 02:52 AM
Don't fall into the media trap.

Manny plays for the Red Sox, which means every time he burps it's front page news on espn.com, and every game he plays leads off sportscenter. The Red Sox have become a media-inflated monstrosity. Espn got what it set out to create - a national team to rival the Yankees. Try finding any park in baseball besides Yankee Stadium where, when the Sox visit, they don't have at least half the fans. You'll see it next week in Cincinnati.

Griffey plays in Cincinnati, which probably wouldn't be the lead sportscenter story if volquez and cueto threw back-to-back no hitters ... unless it was against said Red Sox.

Griffey blows Manny Ramirez away. Not even close.

mth123
06-03-2008, 06:26 AM
A lot of people on here are giving Griffey the nod because of his defense and I generally agree. Here is a question though. At what point does spending a large portion of a career as one of the worst defenders in the game offset a decade of being one of the best at a premier position?

I'd say looking at Griffey's career as a whole, its hard to use defense as something that sets him apart. I know injury robbed him of his greatness, but this isn't a theoretical argument. If it was, we might be talking about Tony Conigliaro and Bobby Valentine as the standards of greatness. I think its a legitimate question.

blumj
06-03-2008, 07:46 AM
I'd say looking at Griffey's career as a whole, its hard to use defense as something that sets him apart. I know injury robbed him of his greatness, but this isn't a theoretical argument.
Sure, if you were comparing him to someone who spent more of his career making a positive defensive contribution than he did. In this particular case, though, you're comparing him to someone who spent less of his career making one.

Jpup
06-03-2008, 03:16 PM
Don't fall into the media trap.

Manny plays for the Red Sox, which means every time he burps it's front page news on espn.com, and every game he plays leads off sportscenter. The Red Sox have become a media-inflated monstrosity. Espn got what it set out to create - a national team to rival the Yankees. Try finding any park in baseball besides Yankee Stadium where, when the Sox visit, they don't have at least half the fans. You'll see it next week in Cincinnati.

Griffey plays in Cincinnati, which probably wouldn't be the lead sportscenter story if volquez and cueto threw back-to-back no hitters ... unless it was against said Red Sox.

Griffey blows Manny Ramirez away. Not even close.

Jay Bruce and Ken Griffey Jr. were the leads on SC on Sunday Night. They talked about Jay several times.

jojo
06-03-2008, 03:35 PM
Beyond the internal problems with linear weights systems you've got the massively larger problem when you try to mix and match different systems (e.g. for hitting and fielding) in the pursuit of total player value - system A and system B usually give different values to the same events meaning that a "run" in one system ultimately does equal a "run" in the other.

Using a linear weights-based run estimator for both offensive and defensive metrics eliminates the "problem".

M2
06-03-2008, 04:15 PM
Using a linear weights-based run estimator for both offensive and defensive metrics eliminates the "problem".

Which estimators? There are plenty to chose from. And do they value each event equally inside of each estimator (e.g. does a hit on offense = a hit on defense)? Because every attempt I've ever seen at that sort of splat math falls down on exactly that point. If none of your inputs (derived from the same events) are given the same value in each estimator then what you've got are two totally separate representations of "runs above average." In fact in every attempt I've ever seen you couldn't begin to normalize average between the offensive and defensive equations. That's a huge problem because incongruous averages could vastly overstate the importance of offense or defense in a given equation.

Given some of the wild vicissitudes in some defensive weight systems, it could lead a total valuation that treats defense as the primary skill of a position player. Now, I like defense plenty and value it highly, but any sensible representation of value needs to know better. And that is a problem born of not really having a full understanding of where the average lies or how to weight +/- in proportion to that average.

Mario-Rijo
06-03-2008, 04:28 PM
Which estimators? There are plenty to chose from. And do they value each event equally inside of each estimator (e.g. does a hit on offense = a hit on defense)? Because every attempt I've ever seen at that sort of splat math falls down on exactly that point. If none of your inputs (derived from the same events) are given the same value in each estimator then what you've got are two totally separate representations of "runs above average." In fact in every attempt I've ever seen you couldn't begin to normalize average between the offensive and defensive equations. That's a huge problem because incongruous averages could vastly overstate the importance of offense or defense in a given equation.

Given some of the wild vicissitudes in some defensive weight systems, it could lead a total valuation that treats defense as the primary skill of a position player. Now, I like defense plenty and value it highly, but any sensible representation of value needs to know better. And that is a problem born of not really having a full understanding of where the average lies or how to rank +/- in proportion to that average.

Now Breath.....;)

jojo
06-03-2008, 05:58 PM
Which estimators? There are plenty to chose from. And do they value each event equally inside of each estimator (e.g. does a hit on offense = a hit on defense)? Because every attempt I've ever seen at that sort of splat math falls down on exactly that point. If none of your inputs (derived from the same events) are given the same value in each estimator then what you've got are two totally separate representations of "runs above average." In fact in every attempt I've ever seen you couldn't begin to normalize average between the offensive and defensive equations. That's a huge problem because incongruous averages could vastly overstate the importance of offense or defense in a given equation.

Given some of the wild vicissitudes in some defensive weight systems, it could lead a total valuation that treats defense as the primary skill of a position player. Now, I like defense plenty and value it highly, but any sensible representation of value needs to know better. And that is a problem born of not really having a full understanding of where the average lies or how to weight +/- in proportion to that average.

The reality is that dealing in the differences between the commonly accepted linear weights-based run estimators is really dealing in minutia.

Another reality is that the wild "vicissitudes" in the gold standard defensive metrics really are overstated. I have yet to see a PBP-based metric that suggests a position player's primary skill is defense. Even if one system was more prone to overestimation based upon a flaw in the way it estimates run values, taking a survey of several metrics really should mitigate the effect (if all metrics give a similar run value then chances are whatever differences that might exist in their linear weights aren't big).

Defensive metrics often get debated in the context of Adam Dunn's defensive value as it's been suggested he's an outlier whose scores prove the wackiness of the run values spit out by such metrics. His -15 to -20 run defensive value still barely approaches 20% of the estimated offensive value of his worst offensive year ('06).

M2
06-04-2008, 09:35 AM
The reality is that dealing in the differences between the commonly accepted linear weights-based run estimators is really dealing in minutia.

I don't consider bad math inside a mathematical equation to be minutia. Your argument seems to be that bad math is convenient and you'd rather not deal with its implications, but those implications still exist regardless of you desire to ignore them.


Defensive metrics often get debated in the context of Adam Dunn's defensive value as it's been suggested he's an outlier whose scores prove the wackiness of the run values spit out by such metrics. His -15 to -20 run defensive value still barely approaches 20% of the estimated offensive value of his worst offensive year ('06).

That's because Dunn has a big bat. It also depends on which estimators you use. For instance, if you use VORP on the offensive side of the equation, then only 126 players in MLB last year garned a score of 20+. That's just a hair over four players per team. It leaves you with the very real situation where outliers on the defensive metric scales are having that more or less control their total player value.

In fact, if you use VORP, Dunn in 2006 scored a 23.5. Subtract the -20 that some defensive systems gave him (an exercise I believed you specifically engaged in after that season) and defense accounts for nearly half of his value. So was defense half of Dunn's story that season? I submit it was not and that what you're seeing there is what happens when you take two "runs" systems that don't agree on a what a "run" is.

jojo
06-04-2008, 10:41 AM
I don't consider bad math inside a mathematical equation to be minutia. Your argument seems to be that bad math is convenient and you'd rather not deal with its implications, but those implications still exist regardless of you desire to ignore them.

My argument is that the differences that exist are so small that in a practical sense they can be ignored. Or to put it another way as flaws go, they're easy to live with rather than "fatal".


That's because Dunn has a bad bat. It also depends on which estimators you use. For instance, if you use VORP on the offensive side of the equation, then only 126 players in MLB last year garned a score of 20+. That's just a hair over four players per team. It leaves you with the very situation where outliers on the defensive metric scales are having that more or less control their total player value.

In fact, if you use VORP, Dunn in 2006 scored a 23.5. Subtract the -20 that some defensive systems gave him (an exercise I believed you specifically engaged in after that season) and defense accounts for nearly half of his value. So was defense half of Dunn's story that season? I submit it was not and that what you're seeing there is what happens when you take two "runs" systems that don't agree on a what a "run" is.

I think in the above example it's just a case of tripping on baselines.

For instance, when comparing the magnitude of a player's defensive run value to that of his offensive value as defined by VORP, the defensive value should be compared to the run value equal to VORP+replacement level (i.e. it's easy to forget to add the run value of the baseline). I can't remember the exact details now (my spreadsheet where I reversed engineered VORP isn't on my laptop) but I think in '06 a replacement level left fielder created roughly 70 runs given Dunn's PA's (or to be more exact given Dunn's number of at bats and outs relative to those of the league).

So really the interpretation of Dunn's '06 season using VORP and defensive run value would be more like this: given his slump, his offensive production wasn't great enough in the end to make the difference between his bat and glove much more of an advantage than that which could have been provided by a replacement level left fielder (who mightve been expected to give the Reds roughly 70 RC while playing neutral defense).

Last season, Dunn was roughly 3 wins above replacement when considering his VORP and glove. There wasn't a long list of players that could boast that.

M2
06-04-2008, 11:35 AM
My argument is that the differences that exist are so small that in a practical sense they can be ignored. Or to put it another way as flaws go, they're easy to live with rather than "fatal".

I think that's an argument of convenience and that attempts at this sort of polymorphic player valuation willfully ignore the large and often fatal flaws in this methodology.

jojo
06-04-2008, 11:59 AM
I think that's an argument of convenience and that attempts at this sort of polymorphic player valuation willfully ignore the large and often fatal flaws in this methodology.

Where exactly are the large and often fatal flaws? A concrete example would really help the discussion move forward. In any event, this is really a tangent to the true intent of the thread. Maybe we can have the ongoing debate in another thread.....

M2
06-04-2008, 12:29 PM
Where exactly are the large and often fatal flaws? A concrete example would really help the discussion move forward. In any event, this is really a tangent to the true intent of the thread. Maybe we can have the ongoing debate in another thread.....

Probably best, but the short answer is the methodologies can result in incongruous norms (large flaw) and aberrant relations to those norms (fatal flaw). The beauty of Win Shares is its got a clean and consistent baseline and the relationship of various inputs to each other is well-defined. While it may not be the tool to use for exact defensive measurement, it ballparks defense well enough and, IMO, is the gold standard for equating its value with offense in terms of total player value.

What I'd really like to see from PBP data is a system that starts at 0 and works its way up from there so that it could take advantage of the WS weighting methodology.

M2
06-04-2008, 02:09 PM
I think in the above example it's just a case of tripping on baselines.

For instance, when comparing the magnitude of a player's defensive run value to that of his offensive value as defined by VORP, the defensive value should be compared to the run value equal to VORP+replacement level (i.e. it's easy to forget to add the run value of the baseline). I can't remember the exact details now (my spreadsheet where I reversed engineered VORP isn't on my laptop) but I think in '06 a replacement level left fielder created roughly 70 runs given Dunn's PA's (or to be more exact given Dunn's number of at bats and outs relative to those of the league).

So really the interpretation of Dunn's '06 season using VORP and defensive run value would be more like this: given his slump, his offensive production wasn't great enough in the end to make the difference between his bat and glove much more of an advantage than that which could have been provided by a replacement level left fielder (who mightve been expected to give the Reds roughly 70 RC while playing neutral defense).

Let me address this real quick now that I've got a minute (you were good enough to supply an example and it deserves a response).

What you've got is:

VORP + replacement level offense (RLO) + defensive run value = total player value

The problem is that many/most of the defensive run value systems you could use express value in relation to the norm (e.g. good defenders are worth +10 runs, bad defenders are worth -10 runs). They are systems based on a supposed average. VORP is an offensive estimator also based on a supposed average (replacement level). Technically, if the two properly defined a true average and expressed proportional run values compared to that norm, there would be no need to add the third leg of the equation. Yet there is the need to make some kind of adjustment because the results of VORP + defensive run value fail the sniff test.

Before we even get to RLO, there's a problem in using the other methods in conjunction with each other. Replacement level, or average, offense and replacement level, or average, defense might very well be orthagonal concepts, the sort of thing that you can't neatly add together. In fact, the results of VORP + defensive run value indicate that very thing. The relationship of average to average begs for investigation rather than assumption.

Onto RLO, if you add that to VORP, you're now working with a base 0 system on offense (at least the number of sub-0 outputs should be minimal). Meanwhile you're still working with a linear weights system on defense, which is seeking out relationships to an average. What you've done with RLO is change the offensive side of the equation to a completely different expression of value from the sort you're using on defense (or likely using on defense). It's an algebraic no-no. Technically that should mean you're also adding in replacement level defense, just to keep everything congruous. Yet more importantly, you're combining expressions of offense and defense that no don't even purport to be doing something similar.

I realize that the results look better when you do it this way, but that's not a justification for using that system.

jojo
06-04-2008, 04:56 PM
Let me address this real quick now that I've got a minute (you were good enough to supply an example and it deserves a response).

What you've got is:

VORP + replacement level offense (RLO) + defensive run value = total player value

The problem is that many/most of the defensive run value systems you could use express value in relation to the norm (e.g. good defenders are worth +10 runs, bad defenders are worth -10 runs). They are systems based on a supposed average. VORP is an offensive estimator also based on a supposed average (replacement level). Technically, if the two properly defined a true average and expressed proportional run values compared to that norm, there would be no need to add the third leg of the equation. Yet there is the need to make some kind of adjustment because the results of VORP + defensive run value fail the sniff test.

What I was trying to argue was that the sniff test being applied was inappropriate because it ignores the offensive value of the baselines. :cool:

Just so everyone reading is on the same page with the assumptions we’re arguing, replacement level defense is defined as average (or in this case its value is more properly described as neutral or 0) and this is basically dictated by the character of the population of replacement level players. So I'd argue that defensive run values are appropriate for comparing to offensive metrics that are expressed against either replacement level or league average. The “answers” will have different magnitudes but they’re baseline-dependent and really the interpretation is the same.

Really what the sum of a metric like VORP with a defensive run value indicates is a player's value relative to a replacement level player. The point I was trying to make in a previous post is that it's important to remember that even replacement level production has a value and this needs to be considered when arguing defensive run values overestimate the value of defense. While using VORP in conjunction with defensive run values might make it appear defensive run values are exorbitant, it's really an artifact of VORP "hiding" a big chuck of offensive value in its baseline.


Before we even get to RLO, there's a problem in using the other methods in conjunction with each other. Replacement level, or average, offense and replacement level, or average, defense might very well be orthagonal concepts, the sort of thing that you can't neatly add together. In fact, the results of VORP + defensive run value indicate that very thing. The relationship of average to average begs for investigation rather than assumption.

To me it's perfectly intuitive that the mean would vary for different skills even though they are being measured within the same population of individuals (i.e. defensive value might be higher than offensive value for a given population).

Really what using a marginal value approach is reflecting is how much easier it is to find a defensive replacement than an offensive one (i.e. defensive skill is greater in marginal players than offensive skill).


Onto RLO, if you add that to VORP, you're now working with a base 0 system on offense (at least the number of sub-0 outputs should be minimal). Meanwhile you're still working with a linear weights system on defense, which is seeking out relationships to an average. What you've done with RLO is change the offensive side of the equation to a completely different expression of value from the sort you're using on defense (or likely using on defense). It's an algebraic no-no. Technically that should mean you're also adding in replacement level defense, just to keep everything congruous. Yet more importantly, you're combining expressions of offense and defense that no don't even purport to be doing something similar.

I realize that the results look better when you do it this way, but that's not a justification for using that system.

It's not really about trying to justify an approach by making it look better. As discussed above, though the addition of a metric like VORP and defensive run values can make it appear like defense is overvalued, it's mostly because a significant amount of offense becomes transparent. Ignoring additive issues for a moment, its obvious that defensive metrics aren't spitting out numbers that dwarf offensive metrics. Generally defensive metrics suggest Dunn is a -15 to -20 run defender while he's generally created around 100 to 110 runs (using THT's formula).

That said, I agree it really becomes a "rough justice" estimate when adding total RC and defensive values for instance and generally I try to avoid that kind of comparison but even then it's not going to lead to an overestimate of the value of defense per se (actually it probably serves to make Dunn's defensive run value appear to have less impact because of all of the RC he spits out). Given how poorly winshares handles defense, I'd still argue that even this approach allows better valuation of a player's overall worth (though I recognize that you disagree with my views on winshares).

In any event, we've discussed this many times and even though I think we're both open minded individuals, I doubt either of us are going to convince the other to move dramatically closer on these issues. Hopefully, this at least explains better why I don't share similar concerns (i.e. why I don't see adding replacement level/average as an algebraic no no or why I don't think PBP-based metrics overvalue defense).

RedsManRick
06-04-2008, 05:23 PM
Really what using a marginal value approach is reflecting is how much easier it is to find a defensive replacement than an offensive one (i.e. defensive skill is greater in marginal players than offensive skill).

I think this is perhaps the key to the whole confusion, and the point M2 was also get at with his point about the systems not quite meshing.

Of the pool of replacement (RL) offensive talent, RL defensive talent is quite common.

But of the pool of RL defensive talent, RL offensive talent is less common.

The composition of the talent pool makes replacement level offense more valuable than replacement level defense. And as you point out, that gets hidden in these measurement systems that start at replacement level and set it as the zero point for both.

I appreciate that you two were willing to have this exchange out in the open and in a civil manner. I'm sure many others besides myself have had trouble sorting through the various concerns and this has been quite helpful.