Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Baseball Newsstand > Baseball Primer Newsblog > Discussion
Baseball Primer Newsblog
— The Best News Links from the Baseball Newsstand

Monday, April 15, 2013

The Bill James Mailbag

Bill with some Roy Halladay, Todd Helton HOF speak and this…

Why do analysts uses replacement level player as a point of comparison, as opposed to for instance major league average?

Well, it’s not that we use one and don’t use the other.  We figure both (comparison to average and comparison to replacement player); we publish both, we use both.  But that’s a quibble, in that you’re certainly correct that the replacement player comps get 90% of the attention.

The “average” is a nothing point, an imaginary line which has no real-life consequence.  Suppose that a player creates 100 runs, whereas an average player would have created 80 runs and a replacement player would have created 60 runs.  The 100 runs is analogous to “gross receipts” for a store, for example.  The +40 (100 runs over 60) is analogous to “profit”.  But what is the +20 analogous to (100 runs over 80)?

It’s not analogous to anything; there is no such concept.  It’s a non-existent line.  The real-life consequences center on whether or not you are good enough to play—which means, better than the replacement level.  If you drop below THAT line, there are real consequences. 

Running at this from a different angle. . .a comparison to AVERAGE does not indicate VALUE.  A below-average player has positive value.  An average player has real value.  An average player, compared to the average, is at zero, no value.  But an average player has real value.  An average player is paid $4 or $5 million a year—more than that if he’s a regular.  In fact, I would argue (and have argued) that MOST value in baseball consists in being average or less than average.  Only a small portion of value consists of being BETTER than average. 

Every year, almost, some loses a pennant race that they would have won or fails to qualify for a playoff position that they could have had, had they had an average player at some position.  You NEED average players, to win; you need lots of them.  You don’t measure from the average, because that is not a point of no value.

Repoz Posted: April 15, 2013 at 04:02 PM | 42 comment(s) Login to Bookmark
  Tags: history, hof, sabermetrics

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

   1. attaboy Posted: April 15, 2013 at 04:36 PM (#4414890)
Here is something I have never heard before, hidden in the depths of the mailbox, Walter Johnson apparently doing some play-by-play!

http://www.theatlantic.com/entertainment/archive/2013/04/when-baseball-was-a-whole-different-ballgame/274541/
   2. Walt Davis Posted: April 15, 2013 at 04:58 PM (#4414933)
James gets most of that answer completely backwards. Average exists and is precisely measurable. Replacement level is the theoretical construct.

Drawing your black/red profit line at 44 wins? 48 wins? 52 wins? Those are purely arbitrary decisions.

Average is 81 wins. No debate. Being above-average means you might win a division, make the playoffs, win a World Series ... being above-replacement means you might be the 2012 Cubs. Above replacement level is the 12-year-old Ford Focus, above average is the new Camry.

James has long fixated on this idea that it's "wrong" somehow to ever assign negative "value" to any player -- remember win shares originally allowed no negatives. That's just his hangup, not something real -- he's more Lake Wobegon than he'd like to admit. Assigning negatives to players below-average changes nothing about their actual value anymore than shifting replacement level by 4 wins does.

   3. Swedish Chef Posted: April 15, 2013 at 05:14 PM (#4414946)
The baseline means nothing. It's just as valid to put it at 110 wins when the measure is telling you how much the player is dragging down your assembly of an all-time great team. In fact, I think I'm going to sell that concept to the Yankees, Wins Under Greatness.
   4. smileyy Posted: April 15, 2013 at 05:21 PM (#4414955)
If we're talking about "average", should we be talking about average or median?

The "average" MLB position player is really really good. He's a starter, since there's what -- 11-12 position players per team, and 8 of those are starters?
   5. Greg K Posted: April 15, 2013 at 05:24 PM (#4414958)
I took his "has consequences" to mean if you're below this line you're out of a job. If a previously above-average player slips below average, he still plays major league baseball, he's just less valuable. If a previously above-replacement player slips below replacement level, he is (generally*) out of a job.

*Unless he's being paid millions of dollars. I'd use Vernon Wells as an example, but he is annoyingly/predictably having an excellent first couple weeks.
   6. Kiko Sakata Posted: April 15, 2013 at 05:25 PM (#4414961)
James has long fixated on this idea that it's "wrong" somehow to ever assign negative "value" to any player.... Assigning negatives to players below-average changes nothing about their actual value anymore than shifting replacement level by 4 wins does.


It screws up comparisons between players with different levels of playing time. Taken to its extreme, somebody who never played a game (e.g., me) ends up scoring higher (0) than a player who plays every inning of every game at a slightly below-average rate (e.g., Cal Ripken in 1997-98, WAA of -1.0, WAR of 3.7 for those two seasons combined). At the less extreme, it says that more playing time is a negative for below-average players.

   7. Swedish Chef Posted: April 15, 2013 at 05:32 PM (#4414969)
At the less extreme, it says that more playing time is a negative for below-average players.

But having below-average players play more is a negative for a team aiming to be above average.
   8. Voros McCracken of Pinkus Posted: April 15, 2013 at 05:42 PM (#4414980)
deleted
   9. Misirlou's been working for the drug squad Posted: April 15, 2013 at 06:28 PM (#4415019)
But having below-average players play more is a negative for a team aiming to be above average.


Not if the alternative is well below average.

(e.g., Cal Ripken in 1997-98, WAA of -1.0, WAR of 3.7 for those two seasons combined)


Dale Murphy. Using WAA, Murphy would have been far more valuable had his career ended after 1987 (WAA -3.8, WAR 4.8). Now in a HOF discussion context, I'm fine with the concept that 1988-1993 add nothing to his case, but he would have been more valuable to his teams if he didn't play 154 games at a 99 OPS+ in 1990, or 153 games at 103 OPS+ in 1991? Teams have guys like that all the time. Good teams, great teams. You think the Tigers could have used a guy like that last year? They got a 70 OPS+ out of their RF's last year. If a guy like the 1991 version of Dale Murphy were freely available, he would be quickly snatched up by a dozen or more teams.
   10. Jay Z Posted: April 15, 2013 at 08:49 PM (#4415186)
But having below-average players play more is a negative for a team aiming to be above average.


There are two lines. The higher line is the level YOUR average player needs to be at to attain the playoffs. This is probably somewhat above average, but not a lot. These sorts of players have no associated "playing time" cost. You want as many of them as you can get.

The second line is the replacement line. You don't want anyone under this line at all, because you can immediately do better. Those players have no value.

Everyone in between has value, but there is a "playing time" cost associated. You can't just begin accumulating players below the "playoff line" because your team won't make the playoffs. However, all of the players above the "playoff line" will probably be committed to a team before you've filled all of your roster spots. So you can't treat the players below the "playoff line" as if they have no value, can't throw them away, but you need to keep in mind that they are placeholders to a degree.
   11. Long Time Listener, First Time Caller Posted: April 15, 2013 at 09:26 PM (#4415235)
James gets most of that answer completely backwards. Average exists and is precisely measurable. Replacement level is the theoretical construct.


Well, we're talking about two different things. Average player is much more real from a statistical construct. It is a specific number for each category. It is easily calculated and absolute and there is no room for debate. However, from a conceptual concept, average player doesn't really exist. There maybe be some anomalous player in some given year who is within 5% of each average statistic but no one really knows who would be the prototypical average player. Fans of any given team could probably should out at least 5 names from their roster or AAA team who are clearly what a replacement level player is meant to represent.

Does this make James and others who calculate Win-metrics by replacement rather than average right? Wrong? I really don't feel qualified to say, but in terms of players I think we should be honest about what Bill, Sean, and Fangraphs are trying to do
   12. bobm Posted: April 15, 2013 at 09:30 PM (#4415241)
FTFA:

Everything that I know that indicates durability in a starting pitcher would have indicated that Johan Santana would be durable.    As I recall, Santana has never in his career--never once--thrown more than 130 pitches in a game.    His mechanics were good, his conditioning was good.  


                                                                                                                                               
Rk         Date  Tm Opp   Rslt  AppDec  IP H R ER BB SO HR Pit Str GSc IR IS BF AB 2B 3B IBB HBP SH SF GDP SB CS PO BK WP  ERA   WPA  RE24  aLI
1    2012-06-01 NYM STL W  8-0 SHO9  W 9.0 0 0  0  5  8  0 134  77  90       32 27  0  0   0   0  0  0   0  0  0  0  0  0 0.00 0.340 4.107 .772


   13. Harveys Wallbangers Posted: April 15, 2013 at 09:34 PM (#4415244)
bill was in good form in this mailbag. not that I have to agree with him to not think he's gone off the deep end but he wasn't taking some obviously extreme position just to be ornery

new meds maybe?
   14. cardsfanboy Posted: April 15, 2013 at 09:47 PM (#4415263)
James gets most of that answer completely backwards. Average exists and is precisely measurable. Replacement level is the theoretical construct.


I agree with him, for the most part. Maybe the way he worded it wasn't the best, but his point is that a below average player is a plus player. If you concentrate on just comparing to average player your going to grade plus players as a negative.

He is trying to say that when teams build, they are building from a base of replacement, not from a base of average, that is what he means by
The “average” is a nothing point, an imaginary line which has no real-life consequence.

....It’s not analogous to anything; there is no such concept. It’s a non-existent line. The real-life consequences center on whether or not you are good enough to play—which means, better than the replacement level. If you drop below THAT line, there are real consequences.


   15. valuearbitrageur Posted: April 15, 2013 at 09:51 PM (#4415267)
Helton's splits--.349 home BA to .289, 216 HR to 138, 816 RBIs to 530. They are WAY different. That made me think about Fred McGriff. McGriff's home/road splits are truly about equal. If you take these two hitters road stats, in 800 more ABs, McGriff has 680 runs as opposed to Helton's 514, 252 HRs as compared to Helton's 138, and 772 RBIs as compare do Helton's 530. Their road batting averages are .288 for McGriff and .289 for Helton


This questioner misses the point that McGriff has a huge edge over Helton when comparing road stats only.

Todd Helton has hit in the toughest collection of road parks in the NL during his career, while McGriff, from his career stats appears to have spent his career hitting in road parks significantly more favorable than average.
   16. Walt Davis Posted: April 15, 2013 at 11:18 PM (#4415341)
It screws up comparisons between players with different levels of playing time.

Only if you do it dumbly.

Yes, if you just straight compare WAA, it doesn't adjust for playing time. WAR is a more easily usable way of doing that so by all means continue to use it. But it's measuring the exact same thing -- the value provided by the player doesn't change one bit based on where you define the arbitrary zero point. If James had simply responded with "it doesn't matter what we measure against but measuring against replacement is an easier way to combine quality and quantity" I'd have had no objection to what he said. I'm not suggesting use WAA not WAR, I'm saying average exists and replacement level is the artificial construct.

Sure, if you're below replacement level you lose your job. And if you fall below average you lose playing time. If you fall far enough below average, you lose your job. The real-life consequences are the same because the value is the same.

James's own example doesn't work. That "would have" is an assumption of equal playing time among those players so he's not really comparing 100 runs created to 60 runs created. The exact same logic applies. If one player creates 100 runs when an average player would have created 80 and another creates 60 then ... shazam, player 1 is 20 runs above-average and player 3 is 20 runs below average. To extend James's analogy, people often get fired and businesses often close if they are producing a below-average profit.

Negative relative to average does not mean negative value, it means below-average. Above replacement means "above an arbitrary cutoff that has been defined as the level where talent is 'free' even though baseball is one of the most restricted labor markets there is." The arbitrariness of the cutoff is highlighted by the (helfpul!) recent agreement for b-r and fangraphs to use the same replacement level -- with this new replacement level shockingly ending up being the average of the two.

He is trying to say that when teams build, they are building from a base of replacement, not from a base of average

No they don't. OK, maybe the 48 win Tigers were building from a base of replacement but even the freaking Astros made it to 55 wins. Historically speaking, the typical team is building from (wait for it) average.

But, no, that's not what James is trying to say. James is trying to say that it's somehow "wrong" to apply a negative number to players who are providing "value." He's saying that for some reason the criteria should be "can they keep a job" rather than "can they earn the average wage." Above-replacement surgeons are providing value too but I'd prefer the above-average one slice me open.

I'm saying value is value no matter where you put the zero point and that average actually exists (and can be estimated with a great deal of precision) while replacement level is an arbitrary zero point. There's nothing wrong with that but let's be very clear that it's an arbitrary point chosen for the convenience of calculation not something important.
   17. Never Give an Inge (Dave) Posted: April 16, 2013 at 12:16 AM (#4415373)

Average performance levels can be calculated precisely but that doesn't mean the concept has a lot of meaning. Plate appearances are not allocated equally to all players. If there are 150 outfielders on MLB rosters at any given time, what we think of as MLB average isn't the level the 75th best one performs at; it's probably more like the level that the 50-60th best performs at.

Yes, if you just straight compare WAA, it doesn't adjust for playing time. WAR is a more easily usable way of doing that so by all means continue to use it. But it's measuring the exact same thing -- the value provided by the player doesn't change one bit based on where you define the arbitrary zero point.

Obviously the player's performance is the player's performance; it's worth the same amount no matter what number you decide to put on it. The point of the number is to describe the value of the performance, not to determine the value.

Any value-based statistic like WAR or WAA takes datasets with more than one dimension (i.e. rate and quantity) - and condenses it into one dimension. You lose information in that process. Using WAA, we lose the ability to distinguish between two players who perform at an average rate based on how much they play. Using WAR, we have that ability, although we lose it for two players who perform at a replacement rate.
   18. bjhanke Posted: April 16, 2013 at 03:12 AM (#4415403)
In terms of the mathematics of baseball, Walt is certainly correct. When it comes to math, Walt is virtually always certainly correct. I, however, have a slightly different background from Walt's, and use a different terminology to distinguish between what average value does, what replacement rate does and what Win Shares does - "organic" stats. This is, essentially, me thinking like an applied (engineering) mathematician instead of a theoretical mathematician. This makes sense, because I am much much better at solving word problems than at proving theorems (my actual college degree in "math" is almost equally divided between applied math classes taught in the engineering school and theoretical math classes taught in the A&S school, and I did MUCH better in the engineering school classes, but Vanderbilt, at the time, did not have a major in applied math / computer science, and I had to major in something).

Anyway, credentials aside, my concept of "organic" stats is stats that derive from what baseball people actually do. Baseball people do not try to work from the average or from the replacement rate or any other mathematical rate, theoretical or applied. When confronting a player on their team, they work like this: "Do we have, or can we easily get, a better player than this guy we have at this position?" That is the justification for Replacement Rate - it's a consequence of how baseball people think about their teams. If your player is worse than replacement level, then the answer to the baseball person's question is likely to be yes; if he is better than replacement, the answer is likely to be no.

But this is a short-sighted thinking process. If your team's management is long-sighted enough to ask the question, "Is this player good enough that he can help us win a title?" they will often conclude that he is, even if he's a bit below average. Why? Because they will also have to ask the question, "How do we get a TEAM together that can win a title?" That may involve concepts like "This guy is a bit above average, but he's a straightforward singles hitter, and we lack power, and we know of an average power hitter who is available in trade for our average-+ singles hitter." Or the other way around, or "This guy is about average, but he doesn't get on base much and we need a leadoff man, and we've just been offered one in trade for our guy."

Mathematically, Walt says, "average exists and replacement level is the artificial construct. " This is absolutely correct, taken as math. Replacement level is not a real mathematical concept. It's an "organic" concept, derived from how baseball people think in ways that do not entirely rest on math. For example, I being from STL, let's think about the Cardinals. Right now, the Cards have, essentially, three power-hitting first basemen: Carlos Beltran, who can play the outfield well when he's healthy, but whose recent health history isn't very good, Allen Craig, who can sort of play left field, and Matt Adams, who apparently can't play the outfield at all. In short, the Cards can afford to trade a power-hitting 1B for a bit less than his value within all of MLB. What do they need? Well, their leadoff man, CF John Jay, isn't all that good, and their middle infield could collapse any time the league catches up to Pete Kozma. Could they use a good-glove shortstop or 2B who is a good leadoff man? Oh, yeah. Do they need one of those more than they need all three power hitters? Again, yes. Could they benefit from trading one of their three boppers for a SS / leadoff man who is not, mathematically rated, quite as good as their power hitters? Sure.

And that's where Walt, although absolutely right in what he says, isn't answering the whole question, probably because the article doesn't ask it (Walt knows the argument I've just made as well as I do). Replacement rate is an artificial concept because each team has a small sample size of players. The best move a team could make is not always to trade up in terms of mathematical quality. Sometimes - often - the best move they can make is to trade down in terms of math value, but trade up in terms of the overall quality of the team, by trading a grade B bopper who is sitting on the bench for a grade C guy who would replace the grade D guy who is actually starting at shortstop. Where the concept of average comes in is in evaluating the team as a whole. Your team, as a whole, needs to grade out to above average if you want to win (I'm not getting into interleague play or wild card slots). But it doesn't matter whether you have an above-average lineup because you have 6 grade C guys and two Grade A guys, or one Grade A, three grade Bs, three grade Cs and one grade D. What matters is that your TEAM is above average. And that's where Walt is absolutely right. Having the grade D guy might even be a HELP, because he's so easy to upgrade.

However, if you're trying to figure out who to vote into the Hall of Fame, it's a pain to find out that the last three years of a 22-year career were a bit below average, and so this candidate's career value is a smaller number than it would have been if he'd have retired 3 years ago. Pete Palmer's rankings, based on using the average as a zero point, run into that problem regularly. WAR systems seldom do. Win Shares never does, because Bill doesn't allow negative numbers. So, which mathematical zero-point you should use depends less on the mathematical quality of the method as much as it does on what question you're trying to answer - what the actual word problem is. This particular thread is asking the question to which the right answer is "the average." Walt has it right. It's just that there are other questions to ask, and they lead to different methods of answering them. - Brock Hanke
   19. Ben V-L Posted: April 16, 2013 at 04:28 AM (#4415405)
Look, there isn't a right or wrong method for establishing a baseline and then measuring your team with respect to it. So in that sense, James is wrong: the average can be used as a baseline. Or some definition of replacement level which would lead to, say, 50 wins for a team full of replacement players. Or some definition of "target excellence" level which would lead to, say, 100 wins. Define the baseline, and then measure your players relative to it.

The reason it all works is because real baseball is mostly linear, which WAR or any runs created type metric is more or less assuming (by virtue of not being conditional, e.g. we don't say 3.7 WAR on a .500 team, but only 2.4 WAR on a .600 team).

Having said all that, if I want to compare the overall productivity of the long, slightly productive career of, say, Harold Baines, to the short, highly productive career of, say, Chuck Knoblauch, then (1) I can't do it with a single number, but (2) as far as single numbers go, WAR is more useful than WAA.
   20. Rants Mulliniks Posted: April 16, 2013 at 08:45 AM (#4415453)
Todd Helton has hit in the toughest collection of road parks in the NL during his career


Thanks for bringing that up, I hadn't even thought about that. In the list of road parks he's played in, Dodger Stadium, SBC/AT&T/WTF Park and Petco are first, third and fourth ranked by PA.

Even so, he's only had a total of 1170 PA in the those parks, which is only 12.9% of his career PA and 26.5% of his road PA. I don't know if that outweighs the massive Coors advantage or not. I was shocked to see his career line is still .320/.419/.545, considering he hasn't been an elite hitter for seemingly forever.
   21. Crispix reaches boiling point with lackluster play Posted: April 16, 2013 at 09:06 AM (#4415468)
I was shocked to see his career line is still .320/.419/.545, considering he hasn't been an elite hitter for seemingly forever.


Fortunately he also hasn't played a full season in seemingly forever, so his career rates can't decline too fast.
   22. The Id of SugarBear Blanks Posted: April 16, 2013 at 09:23 AM (#4415476)
I'm saying value is value no matter where you put the zero point and that average actually exists (and can be estimated with a great deal of precision) while replacement level is an arbitrary zero point. There's nothing wrong with that but let's be very clear that it's an arbitrary point chosen for the convenience of calculation not something important.

Exactly. Making "replacement level" the zero point is no more demanded by logic than placing the zero point of temperature where water freezes (Celsius) or absolute zero (Kelvin).

Not to mention the fact that average considers the actual human beings comprising the era's baseball players, whereas replacement level does not.
   23. Never Give an Inge (Dave) Posted: April 16, 2013 at 12:20 PM (#4415630)

Exactly. Making "replacement level" the zero point is no more demanded by logic than placing the zero point of temperature where water freezes (Celsius) or absolute zero (Kelvin).

The right metric or baseline depends mostly on what question you're trying to answer.
   24. Fancy Pants Handles lap changes with class Posted: April 16, 2013 at 01:20 PM (#4415719)
The right metric or baseline depends mostly on what question you're trying to answer.

Exactly. Which is why I disagree with Walt et al about WAR and WAA being basically the same thing.
The 'wins above x' statistics are attempting to measure value. They aren't abstract statistics,
with large amounts of subjective interpretation. They have a defined purpose, to measure value.
And any statistic that 200 IP at 4.20 ERA is less valuable than 50 IP at 4.25 ERA, is flat out wrong.

Also bobn is a serial page breaker, and should have his posting rights taken away. And his children.
   25. valuearbitrageur Posted: April 16, 2013 at 01:55 PM (#4415771)
Even so, he's only had a total of 1170 PA in the those parks, which is only 12.9% of his career PA and 26.5% of his road PA. I don't know if that outweighs the massive Coors advantage or not.


It's simple math. If Coors Field has been the best hitters park in baseball by a mile during Helton's career, the road parks he will hit in will always be the toughest collection of road parks in baseball to hit in. Because Coors Field is so far above average, all of the other parks have to be significantly below average, for the average to be, well, average.

Imagine Coors FIeld is 120, or a 20% better than average hitters park in a 5 team league. That means we know mathematically that Petco Park, Dodgers Stadium, Coors Field, and PGE Park combined average a 95, in order for the "league" average to be 100.

Let's also assume an imaginary hitter, Todd H, who is normally a .320 hitter. If he hit in a 100 park at home and 100 parks on the road, he'd normally hit something like .328 at home and .314 on the road, because league-wide batters tend to hit 3-4% better at home than in road parks (last year .259 home, .250 road).

But since Todd H hits in Coors with that 20% favorable advantage going for him, instead he hits .392 (1.2*.328) at home.

So what should he hit on the road? Well his average road park is a 95, so he'll hit .320*.98*.95, or .298.

He's still a .320 hitter overall, but mathematics dictates ToddH to have extra large home/road splits because his home park is an extreme hitters park. Todd can't escape this math over a long sample size, and neither can any hitter (Walker) who spent long stretches of their careers in Coors. Obviously his home/road splits don't even out the simplistic way i've done the math, but the concept is correct, the math is just a bit harder than I have done here.

Now, let's talk about another imaginary hitter. We'll call him FredM and he's a .290 career in a neutral context. He spends his whole career in a collection of pitchers parks, the 95 parks in our example. So at home his real average is around .290*1.02*.95 = .281. The drag of hitting in a tough home park actually has him hitting worse at home than his neutral expectation.

So what does FredM hit on the road? Well his road parks are always 95,95,95,120, or an average of 101.5. FredM has far superior road parks to hit in because he occasionally gets to hit in Coors Field, while ToddH NEVER gets to hit on the road in Coors Field. Now we expect FredM's road batting average to be .290*.98*1.015 = .289, pretty much his expectation in a neutral environment.

So then some yahoo comes along and says, hey, everyone thinks ToddH is a HOFer, but if he is certainly FredM is also because their road stats are eerily similar despite Todd's career stats being so much better than FredM's. The yahoo correctly perceives that ToddH has been given an advantage in career stats because of home park effects, but then tries to compare the players in a bad context where FredM's stats are given a significant boost and ToddH's are substantially deflated, by road park effects.
   26. BDC Posted: April 16, 2013 at 01:55 PM (#4415773)
This is a terrific discussion.

That said, I will lower the standards by contributing :) One issue I have trouble thinking about (being bad at both math and word problems … actually not good at anything much except crossword puzzles) is the McCovey/Concepcion problem. We've talked about this numerous times before, but I wonder if the average/replacement distinction matters at all there.

The problem is now classic: you have a first baseman who is the best hitter in the league, but only X runs above replacement for a 1B, and you have a shortstop who is a below-average hitter, but also X runs above replacement for a SS. (Leave defensive contribution aside for the moment; consider only position.)

Just on that basis, the first baseman seems to me a Hall of Famer and the shortstop not. The SS is the tallest of a bunch of little people in that given league, and the 1B is the tallest of a bunch of very tall. We've more or less settled that question; it's a matter of absolute scale, not local perspective.

But does that translate to value in the immediate environment? Defense and roster makeup being equal, would you trade the SS interchangeably for the 1B? Which would you choose first in an open draft?

A run is a run, but I can't help thinking (perhaps just swayed by verbal susceptibilities) that the 1B continues to be somewhat more valuable simply because he is the best hitter in the league. Is this a fallacy? Does it depend on run environment? Does it depend on how the players are distributed above replacement level (ie are most of them clustered around average, or are there several really good hitters and several below average but just above replacement?

These questions bewilder me and are why I read these threads :) Thanks, everyone.
   27. bobm Posted: April 16, 2013 at 02:48 PM (#4415872)
Also bobn is a serial page breaker, and should have his posting rights taken away. And his children.

Fancy Pants: please either suggest a fix or put me on ignore until the page flips, and then you can take me off. Or not.
   28. Moeball Posted: April 16, 2013 at 03:22 PM (#4415934)
Look, there isn't a right or wrong method for establishing a baseline and then measuring your team with respect to it. So in that sense, James is wrong: the average can be used as a baseline. Or some definition of replacement level which would lead to, say, 50 wins for a team full of replacement players. Or some definition of "target excellence" level which would lead to, say, 100 wins. Define the baseline, and then measure your players relative to it.

The reason it all works is because real baseball is mostly linear, which WAR or any runs created type metric is more or less assuming (by virtue of not being conditional, e.g. we don't say 3.7 WAR on a .500 team, but only 2.4 WAR on a .600 team).

Having said all that, if I want to compare the overall productivity of the long, slightly productive career of, say, Harold Baines, to the short, highly productive career of, say, Chuck Knoblauch, then (1) I can't do it with a single number, but (2) as far as single numbers go, WAR is more useful than WAA.


I get the whole idea of why WAR became popular in the first place – in most situations, if your star player goes down with an injury, the team doesn’t necessarily have a league-average player sitting on the bench to step in and fill the void – that “replacement” player is frequently well below average in quality, and the gap between that star player and the “replacement level” player may be quite substantial. But WAR in concept also tells you as much about a team as it does a specific player.

For example, if the star 3B for a big money team (say, NYY) goes down, there may be some drop-off in performance by his replacement, but his replacement may very likely perform at average or above-average level. There won't necessarily be a huge drop-off in performance by the "replacement level" player for that team, as the team may have a very deep and high-quality bench.

On the other hand, if the star 3B for another team (say, SD) goes down, his replacement player is...Jedd Gyorko.

Who?

Exactly - the star 3B for SD is much farther above "replacement level" for his situation than the star 3B for NYY.

It's all about context, which is why I prefer WAA to WAR.

With WAA, we can very easily determine what baseline each player is being compared to and it’s not varying all over the place like “replacement level” does. All the players at the same position are being compared to the same baseline.

Furthermore, you can do fun things with WAA that can really make you recognize truly historic performance levels.

For example: with WAA, each WAA represents one additional win above average and one fewer loss than average, or 2 games above .500. A player with a WAA of 9 would therefore essentially be 18 games over .500 for his team. Now, just imagine if your entire team could play at that level – 9 positions on the field, all playing at a level of 18 games over .500. With the current 162-game schedule, that would leave the team with a record of…162-0. That’s right, perfection. This is why I refer to a WAA of 9 as a “perfect” season (back in the days of the old 154-game schedule, it would take a WAA of 8.6 to reach “perfection”). Very few of these seasons have been recorded in baseball history, and it takes extreme peak value to reach this level.

Here’s a list of position players (post-1900) who have achieved this historic level of performance in at least one season, and it is a “who’s who” of some of the greatest to ever put on a uniform – Wagner, Cobb, Hornsby*, Ruth*, Gehrig, Williams*, Mantle*, Yastrzemski, Morgan, Ripken, Bonds*. In this case, the * represents players who’ve had multiple “perfect” seasons – I suppose it’s no surprise that Bonds would wind up with an asterisk!

An added bonus of using WAA is when looking at all-time great teams – I believe the 1927 Yankees with Ruth and Gehrig are the only team in history to have two players with “perfect” seasons the same year! This tells you just how stratospheric the level of performance was by those two!

Of course, if I’ve missed any players on the list, I’m sure someone will be sure to let me know!
   29. Fancy Pants Handles lap changes with class Posted: April 16, 2013 at 03:23 PM (#4415937)
A run is a run, but I can't help thinking (perhaps just swayed by verbal susceptibilities) that the 1B continues to be somewhat more valuable simply because he is the best hitter in the league. Is this a fallacy? Does it depend on run environment? Does it depend on how the players are distributed above replacement level (ie are most of them clustered around average, or are there several really good hitters and several below average but just above replacement?

These questions bewilder me and are why I read these threads :) Thanks, everyone.

There is really no magical extra value in being the best hitter in baseball, beyond simply the runs it provides.
Put it this way, Mike Trout had a phenomenal season last year, and that is not diminished by Cabrera
being a slightly better hitter. Similarly Cabrera's season isn't elevated above Trouts just because
of that. And if say Kemp had put up a 1.100 OPS, it wouldn't have diminished Cabrera's season at all.

Which player you should choose, really all depends on who you can bring in to replace your guy that you
are trading away, and who the new guy replaces. So let's look at it slightly differently, instead of a
straight up swap of best hitter for best SS, let's instead look at best hitter and 10th best SS,
for best SS and 10th best hitter.

Last year that get's you .999 OPS and .689, for .845 and .921.

   30. Rusty Priske Posted: April 16, 2013 at 03:30 PM (#4415950)
Replacement level as a concept is one of the most important things when managing a team. It is pretty simple, "Are you giving playing time to someone when you could do better simply by replacing him?"

The problem is that it ISN'T used as a concept. It is used as a baseline and that implies that every team has access to this amorphous pool of replacement players, and that isn't true.

So since it isn't truly "replacement" level and is just an imaginary line that tells you how well the player is doing on a scale that has no purpose other than comparing players to each other, it really doesn't matter WHAT line you use. The first year I ever played roto baseball (around 1988) a guy in our league was using Jose Canseco as the baseline. It didn't matter what it was, as long as you had a common frame of reference.

(Which, of course, does not seem substantially different from what Brock said... just simplified somewhat, imo)
   31. Never Give an Inge (Dave) Posted: April 16, 2013 at 03:30 PM (#4415951)

If you want to compare a player to his actual replacement, just subtract one guy's WAR from the other's (and adjust for playing time).
   32. Fancy Pants Handles lap changes with class Posted: April 16, 2013 at 03:44 PM (#4415962)
For example, if the star 3B for a big money team (say, NYY) goes down, there may be some drop-off in performance by his replacement, but his replacement may very likely perform at average or above-average level. There won't necessarily be a huge drop-off in performance by the "replacement level" player for that team, as the team may have a very deep and high-quality bench.

On the other hand, if the star 3B for another team (say, SD) goes down, his replacement player is...Jedd Gyorko.

Who?

Exactly - the star 3B for SD is much farther above "replacement level" for his situation than the star 3B for NYY.


Jedd Gyorko seems to be bumbling about nicely at replacement level so far this season. And if the Padres
think he couldn't do that (and cared about it - it's the Padres, I am not sure they do), they could trade
a bad of baseballs for a random replacement level journeyman who could. That's the thing about replacement
level players. Teams don't value them, and have little compunction in dumping them.
   33. Fancy Pants Handles lap changes with class Posted: April 16, 2013 at 03:52 PM (#4415969)
Replacement level as a concept is one of the most important things when managing a team. It is pretty simple, "Are you giving playing time to someone when you could do better simply by replacing him?"

The problem is that it ISN'T used as a concept. It is used as a baseline and that implies that every team has access to this amorphous pool of replacement players, and that isn't true.

Nope. The concept of replacement level players relies upon the fact that since they have no value,
they are essentially freely available, as you can always find another team willing to give you one for
some different organizational filler. So even teams that may not have a replacement level guy at a
certain position in their system, they still have access to them via trade.

Now the exact placement of the replacement level line may not be accurate withing a microrun, but
that doesn't change the fact that the concept is sound.
   34. Petunia inquires about ponies Posted: April 16, 2013 at 07:06 PM (#4416141)
Also bobn is a serial page breaker, and should have his posting rights taken away. And his children.

You bite your tongue. Bobm is a ####### national treasure. Get a real monitor for Pete's sake.
   35. Bug Selig Posted: April 16, 2013 at 07:22 PM (#4416152)
The huge problem I have with average as a baseline is that, in the sense of roster-building for a DMB league (but it applies in any setting) is this:

Using average seems to assume that you have the option of playing nobody - and you don't. You can't simply punt a position and have that be the same as having an average player. As Bill says in the excerpt, much of the value on a team lies in simply filling all the spots in a non-disastrous manner. For instance, Nick Punto is a below-average player. But, having him on your roster doesn't take you further from having a contending team, he likely brings you closer. If you have Punto, you have eliminated the possibility of being forced to run Cale Iorg or Mark DeRosa out there for 140 games. He doesn't help you as much as a star, but WAR and Win Shares don't claim that it does. WAA, on the other hand, would claim that he is worth less than zero.
   36. BDC Posted: April 16, 2013 at 09:16 PM (#4416301)
#s 29, 30, and 35 would seem to suggest that replacement decisions in *real* (as opposed to fantasy) baseball are highly situation-dependent, which makes sense. Option A for team B may not be nearly as optimal as Option C for team D. (Option J always trumping everything, naturally.)

In fantasy, heck, there's almost always somebody evident around you can just draft. The problem for a real manager is trying to decide whether Mitch Moreland is really a good move to replace Chris Davis when you haven't a clue, honestly.
   37. Good cripple hitter Posted: April 16, 2013 at 09:27 PM (#4416328)
Also bobn is a serial page breaker, and should have his posting rights taken away. And his children.

You bite your tongue. Bobm is a ####### national treasure. Get a real monitor for Pete's sake.


I put him on ignore a few days ago for the page breaking issue. I don't really have anything against him, but he does break the page on a lot of threads and I just got tired of it.

This page is broken on my laptop's monitor because instead of writing "Santana threw 130+ pitches on DD/MM/YY" and maybe a link, we got a long string of stats from the game that don't really add anything to the post.
   38. McCoy Posted: April 16, 2013 at 09:58 PM (#4416376)
WAA, on the other hand, would claim that he is worth less than zero.

No. It would claim that he is less than average.
   39. Jack Carter, calling Beleaguered Castle Posted: April 16, 2013 at 11:27 PM (#4416500)
I was shocked to see his career line is still .320/.419/.545, considering he hasn't been an elite hitter for seemingly forever.


Fortunately he also hasn't played a full season in seemingly forever, so his career rates can't decline too fast.


Well, he's still had a long decline. Around 2300 pas in the last five years, with an OPS+ of 107. Only two useful seasons out of five, and his D doesn't seem to bad. That's a lot of weak PAs for a RFer.
   40. beer on a stick Posted: April 16, 2013 at 11:52 PM (#4416520)
OK, so I'm listening to Vin do the Dodgers and the Padres on the satellite, and I clicked on the link with Walter Johnson doing PBP in 1939. So I have two games going at once, both being called by legends, about seven-and-a-half decades apart.

I can't decide if this is really cool or really weird.
   41. Greg K Posted: April 17, 2013 at 04:40 AM (#4416551)
It's kind of weird if they sync up and it turns out Scully and Johnson are actually having a conversation with one another.
   42. Greg K Posted: April 17, 2013 at 04:47 AM (#4416552)
Exactly. Making "replacement level" the zero point is no more demanded by logic than placing the zero point of temperature where water freezes (Celsius) or absolute zero (Kelvin).

Agree with #23, it depends on what question you're asking. The only thing we know for sure is that placing zero point of temperature at the freezing point of brine is ####### stupid.

What's the baseball equivalent, "Wins Above Rob Deer's 1988 Season"?

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
TedBerg
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Hot Topics

NewsblogFan Returns Home Run Ball to Ishikawa; Receives World Series tickets
(35 - 8:04am, Oct 21)
Last: Never Give an Inge (Dave)

NewsblogOT: Politics, October 2014: Sunshine, Baseball, and Etch A Sketch: How Politicians Use Analogies
(2776 - 7:59am, Oct 21)
Last: Bitter Mouse

NewsblogCould the Yankees ever be Royals? Young and athletic K.C. is everything that Bombers are not - NY Daily News
(30 - 7:51am, Oct 21)
Last: BDC

NewsblogBrisbee: The 5 worst commercials of the MLB postseason
(134 - 7:47am, Oct 21)
Last: toratoratora

NewsblogOT: Monthly NBA Thread - October 2014
(273 - 6:08am, Oct 21)
Last: HMS Moses Taylor

NewsblogDealing or dueling – what’s a manager to do? | MGL on Baseball
(16 - 5:43am, Oct 21)
Last: vivaelpujols

NewsblogOT: NFL/NHL thread
(8367 - 3:29am, Oct 21)
Last: Randomly Fluctuating Defensive Metric

NewsblogHitting coaches blamed for lack of offense - Sports - The Boston Globe
(16 - 3:26am, Oct 21)
Last: ptodd

NewsblogSielski: A friend fights for ex-Phillie Dick Allen's Hall of Fame induction
(71 - 1:20am, Oct 21)
Last: Jacob

NewsblogCalcaterra: So, if you’re not a fan of the Royals or Giants, who ya got?
(93 - 12:26am, Oct 21)
Last: Howie Menckel

NewsblogPitch from Zito helped sell Hudson on Giants | MLB.com
(6 - 9:15pm, Oct 20)
Last: the Hugh Jorgan returns

NewsblogWhy Royals great Frank White no longer associates with the team whose stadium he built - Yahoo Sports
(19 - 9:06pm, Oct 20)
Last: A New Leaf (Black Hawk Reign of Terror)

NewsblogAngell: Gigantic
(38 - 8:22pm, Oct 20)
Last: Jolly Old St. Nick Is A Jolly Old St. Crip

NewsblogMorosi: Could Cain’s story make baseball king of sports world again?
(97 - 6:24pm, Oct 20)
Last: BDC

NewsblogESPN: Brian Roberts retires
(22 - 6:19pm, Oct 20)
Last: Captain Supporter

Page rendered in 0.5392 seconds
52 querie(s) executed