I look forward to this discussion.

So, its some "weighted win shares over time" argument to define who the best player on any given year, so as to not have the year-to-year fluctuations associated with doing each year individually.

I mean, Trout wasn't going to stay #1 after missing large swathes of the last couple years, so, this is fair, even if Trout is likely to take it back if he can get off the Buxton-train. I wouldn't have thought of Freeman automatically as the "2nd best player" over that time. But, he's not a ridiculous choice by any means and has only missed 10 games over the last 5 years, so, while not Ripken exactly, he's barely missing time.

I get the general gist of what James is going for, and it's hard to disagree that Trout's injuries hurt his claim on best player in baseball.

I just don't think Freeman has a particularly strong case as an alternative. He has two big years in the past three, but one of those was 2020, and in the other three of the past five years he's been, well, very good, but not the type of guy you think of as the best in the game. I think you could make a fair case for Ohtani or Mookie Betts, but I actually think the strongest case might be Jose Ramirez, who has put up five MVP level seasons in six years.

I actually think the strongest case might be Jose Ramirez

Yeah, Betts would be ahead of both Ramirez and Freeman minus playing time considerations. James is looking for raw counted win shares, so, missing time has a big impact. I imagine Jose Ramirez must be really close to Freeman, or James' system is broken.

My instinct would be that any 1B who's not producing at near-Pujols levels can't possibly (well, you know what I mean) be the best player in the game. Obviously such a thing does depend on how many years you look across ... if we are only considering players who have played each of the last X years and we're emphasizing playing time then the pool of eligible players will be small.

With Betts, it matters a lot whether you include 2018 or not. It's such an outlier for him, especially offensively. But anyway, even ignoring 2018, by bWAR, Betts averages 5.3 per year and 6.4 WAR/650 while Freeman is at 4.4/4.9. Ramirez is a little better by year and substantially better by 650. Correa is a little short by year but ahead by 650. He had a stinker 2020 but Marcus Semien has an argument here with 5.2/5.4. Speaking of outlier seasons, how can we even consider Freeman when Judge is around -- 5.8 WAR/year; 7.9 per 650. Soto at 4.6/5.7 is within 400 PA of Freeman over the last 5 years. Arenado also had a stinker 2020 but 5.4/6.3 and 2800 PA; even Goldschmidt is at 4.8/5.3 with 2900 PA.

And of course Ohtani. I don't think there's a serious argument right now about who the best player in baseball is.

FWIW, Steamer projection for 2023 have the following top 12 WAR projections. Freeman comes in at number 21 with a 4.8 WAR projection. His durability an reliability factor obviously bump him up, possibly into top 10.


Ohtani. 7.3 (Batting pitching combined)
Soto 7.1
Judge 6.8
Rodriguez 6.0
Betts 5.9
Tatis Jr. 5.8
Rutschman 5.8
Franco 5.6
Trout 5.6 (This is based on 150 games played, 677 PA !!)
Acuna 5.5
DeGrom 5.5 (Only pitcher above 5)
Guerrero 5.4

There are (must be) strong arguments for Ohtani, Soto, or Judge. If I could get one for my team, I’d want it to be Soto.

Pretty incredible that Steamer is projecting very new players Rodriguez, Tatis, and Rutschman near the very top, higher than Trout.

Whatever the methodology, once the eyeball test reveals that there are flaws at almost every transition point, shouldn’t the methodology be adjusted?

Instead, Bill just changes the names to better match reality, manually overriding his model.

(Weighting the current year more heavily could help, or maybe diminishing the weight of past (and future(!)) years…)

Also, given the future year calculations in this model, if I understand this correctly, Freeman won’t appear on future versions of this toy, possibly at all. 2022 will be Rutschman or Ohtani or Soto or someone else who will play better over the next ten years than 33-42 year-old first baseman, Freddie Freeman.

I really don't like this concept. I think the multi-year look is okay, but only going backwards. Someone is great for what they are doing, or have done, not because of what they will do in the future. Because you have to assess greatness in the moment, and the future is not known. You could (and IMO should) include projected future value, but that's still a function of current/past performance.

Also, given the future year calculations in this model, if I understand this correctly, Freeman won’t appear on future versions of this toy, possibly at all. 2022 will be Rutschman or Ohtani or Soto or someone else who will play better over the next ten years than 33-42 year-old first baseman, Freddie Freeman.

Yeah, I was going to say that I suspect the window of time in which Freddie Freeman is the best player in baseball is probably about as long as the time it takes to read this sentence. That doesn't mean he isn't - or wasn't!

Yikes, what's with the Trout pessimism? He just posted a WAR higher than that in fewer PAs. Steamer is projecting the lowest OPS of his career (excluding his cup of coffee). Looking at B-R, Marcel is too. Is 31 really that old?

Manny Machado should be in this discussion.

FTA:

The Troutvocates will say that Trout had higher WAR than Freddie Freeman did in 2022. Well, yeah, but Freeman had more Win Shares, 34 to 22, and I think that’s accurate. Regardless of what WAR says, Freddie Freeman was more valuable in 2022 than Trout was.

There's an arrogance here that "old school" types love to crap all over.

Win Shares isn't inherently better than any other system.

Well, I suppose, sort of like if Eddie Murray snuck into the list in '83 or something. But it should be short-lived, would think Judge or Ohtani takes over eventually.

The Marcel projection for Trout is a bit hard to fathom. It's projecting to just 464 PA which is fine given the methodology. But this is a guy who, over the last 3 seasons, has a 330 ISO and he's projected to 276 ... I didn't think that was even possible with Marcel methodology.

But yes, at some point age-based Marcel starts to ding a batter something like 5% a year. Looks like it might be 31. Trout is projected for about a 5% decline in BA vs the last 3 years. But he's also projected to a walk rate dropping from 1 per 8 to 1 per 9. Then there's that 15% drop in ISO, all adding up to about a 10% drop in OPS -- doesn't seem correct to me. This is apparently "super fancy Marcel" which has both the age factor and regression to the mean -- I suppose with fewer than 900 PA over the last 3 years, Trout will (and should) be aggressively regressed towards the mean. 2020 probably requires everybody to be regressed heavily.

(Note models are models, there's no way to and no need to get them to predict accurately for every individual. Marcel looks back just 3 years I think and Trout has few PAs so heavy regression to the mean is what the model does. But we humans can look back farther and know that, for Trout, his personal mean is way, way, way above the mean -- i.e. we can use more than just those last 900 PAs and conclude he doesn't need to be regressed as heavily as Marcel seems to be doing here. The question for Trout of course is health, not skill. Presumably eventually age, bad health and lack of PT will erode the skill but, in this individual case, I don't think a 10% drop in production rate is justified.)

Anyway, Freeman may be the best player now in the sense that there might have been a year when Rafael Palmeiro graded out the best (or maybe best in AL) just because you could rely on him being there everyday and producing his 4-5 WAR a year. I think you need to greatly over-value playing time to reach such a conclusion though. Reliability at a position and the roster flexibility it creates elsewhere is nice but it's simply not that important. Let peak Larry Walker produce his 6 WAR per year, 7 WAR per 650, while missing a month every season -- still much better to have him on the team than a 4.5 WAR, 650 PA player.

Damn, a Guapo sighting.

Win Shares isn't inherently better than any other system.

It's clearly the hill Bill James has chosen to die on.

If I could get one for my team, I’d want it to be Soto.

The question there is "for how long?" If we're talking one year, it's Ohtani. As we move the window out farther, the injury risk for Ohtani eventually becomes high enough it can't be him. Given his defensive value, I'd probably take Judge over Soto for the next 3-4 years but eventually decline due to age will move Judge below Soto. Of course by then, Franco or Rutschmann or somebody may have passed Soto.

1-year draft: Ohtani
3-year draft: Ohtani
5-year draft: ??? Still probably one of Ohtani, Judge, Soto ... or Franco, Tatis, Correa, Acuna, Riley, etc. if they can take a step forward and/or get healthy
10-year draft: ???

But, in the sense that Freeman may be the best player right now, the guy with the least variance in his projection over the next 5 to 7 to 10 years might well be Soto. Injury risk is always there but, over the next 10 say, it's hard to see anything much worse than 5200 PA of a 140 OPS+ and 40 WAR (with a realistic upside of 55-60 WAR). In a strict baseball sense (i.e. money aside), there's little chance you would regret having Soto for the next 10 years.

By the way, for whatever reason, James's website won't load for me (or takes forevver) so I can't respond to his argument. But if win shares puts Freeman well ahead of ... well, any star player, not just Trout ... then it's way off in this case.

Damn, a Guapo sighting.

Is it 2010? No way Freeman was the best in 2010.

And yeah, Machado should be very much in the discussion, at least as much as Freeman. He even played every game in 2020 and he's at 4.7 WAR/year, 5.3/650.

And I only just noticed this ...

I have concluded that it is no longer reasonable to say that Mike Trout is the best player in baseball. .... This is merely an opinion, of course, and probably no better than your opinion.

Which is it? No longer reasonable or perfectly reasonable?

At some point there a long time ago, it was no longer reasonable to believe that the earth is flat. "The earth is round(ish)" is not an opinion and anybody of the "opinion" that the earth is flat is wrong. In the realm of probabilistic statements we are limited to in baseball projections/evaluation, there is no definitive right/wrong but "unreasonable" still means something like "so unlikely that the conclusion is irrational." Therefore if you are of the counter-"opinion" then you are simply ignoring the objective evidence we have and are somewhere between "wrong" and "just making #### up."

Probably it has been stated before, but perhaps new to some:

One of the issues with Win Shares was it was the same player would get about 5% fewer WS in a DH league (AL pre-2022) than with no DH. Which affects any discussion of best-MLB player using James' metric.

I haven't read the article yet, but if Betts' or Freeman's WS are equal to Trout, then Trout is actually better on an even field.

I don't see a good argument in the piece for going back 10 years into the past and 10 years into the future. Looking backward and forward 3-4 years gives you some assurance that you're not looking at some kind of fluke year. Nothing about what happened 10 years before or after a certain season tells us whether someone was legitimately great in that season.

Win Shares isn't inherently better than any other system.

Right, it's inherently worse, because it sets replacement level far too low (in a misguided attempt to avoid negative numbers or loss shares.) The differential between Freeman and Trout is 209 PA, in which Freeman produced a differential of +75 hits, -19 HR, +30 BB, +37 TB. That's negative value compared to replacement (if someone could hit negative nineteen HR.)

Stated another way, a club will gain more wins with Trout + 209 PA of a typical replacement than with Freeman, but Win Shares outright gets this wrong.

And Freddie Freeman takes over from Trout in 2020, rather than in 2022.

If you look at your list of greatest players and decide to adjust it, the result of your adjustment should not be that Freddie Freeman gets a longer reign.

The shorter the window, the choppier the list becomes. You start getting guys like Wade Boggs and Wally Berger at the top of the list for one year. 

We wouldn't want the top-5 all-time 3B, who dominated the league from 83 to 89, showing up on a list of greatest players!

Right, it's inherently worse, because it sets replacement level far too low (in a misguided attempt to avoid negative numbers or loss shares.)

I used to agree with this, but it seems like there are a ton of negative WAR players these days. Just among position players there were 26 players with -1.0 WAR or worse last year, almost one per team. And then another 60+ with at least -0.5 WAR. Not all of this can be small sample bad luck.

So teams are either playing a lot of inferior players instead of freely available guys, or replacement level is set too high.

So teams are either playing a lot of inferior players instead of freely available guys, or replacement level is set too high.

There's a fair chance that it's simply the former.

The Tigers had three sub-replacement level first base-types last year - Cabrera, Torkelson and Harold Castro. Now of those three Castro was probably the best performer, but he's also the guy who makes the least amount of sense to play - you can hope that Cabrera finds some Pujols magic or Torkelson turns into the type of player you thought you were drafting. Playing a true replacement level guy rarely makes sense, but it's still the most relavant baseline. Guys like at the level of Harold Castro are freely available even if teams would rather play worse players.

The Tigers had three sub-replacement level first base-types last year - Cabrera, Torkelson and Harold Castro.

The Red Sox helped out here too! Dalbec and Shaw both put up -0.7 WAR, with Shaw doing so in only 19 PA!

There are also sometimes non-baseball reasons to play guys with negative WAR. In fact, Cabrera and Torkelson provide good examples. Apart from rebound/development reasons, you might play them because they're a legend on a huge contract or a guy you drafted 1-1, even if there's somebody better in AAA. (Not saying that DET had anybody better, just that they had a reason to play these guys that's not an on-field reason.)

I think the primary reason teams play guys who put up sub-replacement level numbers is because they expected those guys to play better and they waited for improvement that never came. If Justin Turner starts next season in a prolonged slump, for instance, the Dodgers are going to give him an awful long time to get out of it before sitting him down.

I think the common problem he's running into at the transition points is that Win Shares has too low a replacement level, so it overrates very good players who are extremely healthy. Since 2018, Freeman has played 162, 158, 60 (COVID year), 159, and 159 games a year. His OPS+es are, by comparison, 139, 139, 187 (COVID again), 136, and 152. Very good, but aside from the COVID year, none of those OPS+es make you say "best player in baseball", especially at first base.

I didnt read the article so feel free to criticize if I am missing something but...

It seems to me the biggest elephant in the room that no one is addressing is: How accurate are PT projections?

To put it in stark terms: Is Freeman really all that likely to continue to miss almost no games, OR is Freeman an example of sample bias because he is simply lucky enough to have avoided injury and in fact he is as likely to miss games as say Soto or Betts or the average guy who plays 150g/year.

Right? I mean until you answer that question these projections seem to be sort of guesswork.

So teams are either playing a lot of inferior players instead of freely available guys, or replacement level is set too high.

Isnt the assumption that a logical team would find the best replacement player available regardless of money. And that teams e.g. PIT KC and perhaps CIN would rather save money than do the logical baseball thing. I dont know if these bottom feeder teams really play enuf inferior guys to drag down the overall numbers that much, but could that be whats happening?

Even going by win shares, I can't see how it's Freeman. James says 30 win shares are an MVP-level season. Freeman's had one of those in the past 9 years. That's the best player in the game?

it seems like there are a ton of negative WAR players these days. Just among position players there were 26 players with -1.0 WAR or worse last year, almost one per team. And then another 60+ with at least -0.5 WAR. Not all of this can be small sample bad luck.

So teams are either playing a lot of inferior players instead of freely available guys, or replacement level is set too high.

Replacement level is a generalized approximation - it's not literally true that every team has a replacement-level player available at all positions at all times. Sometimes a particular replacement will be below the replacement level for the league as a whole.

Where teams probably err is in keeping such a below-replacement player around for a few hundred PA in a season. But although we call them freely-available, that's not always really true, there's always costs to doing the transaction, such that it's not worth it in an overall sense for such a marginal upgrade.

Even going by win shares, I can't see how it's Freeman.

His methodology is to include 9 seasons on each side of the given year with declining weights, and gives credit for seasons yet to come. Freeman is right in the presumed center of his career, and has excellent durability for playing time which matters for Win Shares' low replacement baseline. Ohtani and Judge don't have the 9 seasons of history yet, and Trout lacks the durability.

You need a baseline that isn't average. That's where "replacement level" comes in.

The method used (19-yr window) is not DESIGNED AT ALL to answer who the best player is in 2022. We won't have all the data in until 2031. So the dicussions about Freeman/Trout/etc are a related topic that needs a different metric. I like BJames' method in general, but highlighting "Trout is no longer #1" just mucked up the overall presentation.

- I agree, WS overslees the value of durable players like Pete Rose and F Freeman
- but Freeman does have a case for best player at the current time. Total Bases 2nd/6th/3rd in the past 3 years in the NL. Times on base, 1st/2nd/1st. Runs Scored, led his league 3 years running. Good base stealer, okay fielder.

The shorter the window, the choppier the list becomes. You start getting guys like Wade Boggs and Wally Berger at the top of the list for one year. 

Getting back to this for a second, not because I'm a Boggsvocate (which I am!), but because it illustrates a problem with the 19 year window. By using that long of a window, Rickey Henderson is apparently able to fend off Boggs in 1988-1989ish because Henderson had an excellent 1980-81 while Boggs was in the minors, then an excellent 1982 while Boggs was playing part time. Why in the world you want your assessment of the best player in 1989 to have anything to do with how he played in 1980-82? Why would you look at your system, see that it makes such a selection, and think, "Aha! This is the right choice!"?

I use the Boggs example here because it's one of the ones that James used and he's someone who's widely considered an all-time great. But maybe Wally Berger is a better example. Isn't something like this more interesting when it gives us some answers we don't expect? What if, sometime in the 1930s, Wally Berger was in fact the greatest player in baseball? That's pretty cool.

I guess it's so obvious that it might not even be worth pointing out but a player's stretch of greatness does not typically last 19 years. When you include that many years, you end up including what are likely a bunch of average, below average, or non-existent years in your assessment of greatness. Should it matter if player X had a average year 8-9 years ago, while player Y wasn't in the league yet? Should that really affect who you think is greater at a given time?

I hope James publishes the results or something like 3-4 years on either side of the give year.

I have to assume a year or two of prime Trout will have him retake the 2022 title? If you polled every player an executive and said, we're starting a new league from existing players, freeman would NEVER be drafted before Trout. You might pick a Soto or Harris etc but never Freeman.

I'd like to see the charts with PED players taken off. I have to assume Pujols' reign might extend backwards a couple more years rather than Bonds or A-Rod.

And finally if you take Bonds/A-Rods entire out who is the best Player without PEDs for the years in question?

Guys like Beltran, Helton, suzuki, Pujols chipper Jones, Rolen and Edmonds seem to be at the top of the lists from 98-2004

Belle, thome and Thomas appear a lot in the earlier years

The low replacement level is definitely responsible for Rose in 1970. Not doing the weighting but from 1961-1979, Baseball Reference has it:

--Yaz 94.1 WAR, 1969-71: 5.5, 9.5, 4.0
--Rose 80.8 WAR, 1969-71: 6.6, 4.8, 5.1

And finally if you take Bonds/A-Rods entire out who is the best Player without PEDs for the years in question?

Guys like Beltran, Helton, suzuki, Pujols chipper Jones, Rolen and Edmonds seem to be at the top of the lists from 98-2004

It's got to be Pedro.

There are also sometimes non-baseball reasons to play guys with negative WAR. In fact, Cabrera and Torkelson provide good examples. Apart from rebound/development reasons, you might play them because they're a legend on a huge contract or a guy you drafted 1-1, even if there's somebody better in AAA.

I think the primary reason teams play guys who put up sub-replacement level numbers is because they expected those guys to play better and they waited for improvement that never came. If Justin Turner starts next season in a prolonged slump, for instance, the Dodgers are going to give him an awful long time to get out of it before sitting him down.

All fair points. The Reds played Eugenio Suarez full time in 2021, when he was so bad he put up -0.7 WAR over 145 games. But his track record justified a rebound expectation. Sure enough he was worth 3.9 WAR this year (for Seattle).

I do think durability helps in small ways that aren't captured by WAR. With bench positions at such a premium, it helps to be able to pencil in a guy for 150+ starts and not have to worry about that position.

But mostly, I don't like the modern trend in MLB and NBA of guys playing less, so I'm biased towards durable players who show up every day.

it seems like there are a ton of negative WAR players these days. Just among position players there were 26 players with -1.0 WAR or worse last year, almost one per team. And then another 60+ with at least -0.5 WAR. Not all of this can be small sample bad luck.

So teams are either playing a lot of inferior players instead of freely available guys, or replacement level is set too high.

My comment about the low replacement value for Win Shares notwithstanding, I don't think WAR gives a high enough value for simple playing time. There is a very real value to occupying one or more defensive positions competently and appearing at the plate a few hundred times without becoming a black hole on offense.

When places like Fangraphs try to predict or evaluate salaries, they value stars way above their observed salaries and journeymen way below.

Why in the world you want your assessment of the best player in 1989 to have anything to do with how he played in 1980-82?

James says why in the article. It's because weightings of 1+2+3+4+5+6+7+8+9+10+9+8+7+6+5+4+3+2+1 sum to a nice round 100. Not kidding, that's his reasoning.

I don't think WAR gives a high enough value for simple playing time. There is a very real value to occupying one or more defensive positions competently and appearing at the plate a few hundred times without becoming a black hole on offense.

When places like Fangraphs try to predict or evaluate salaries, they value stars way above their observed salaries and journeymen way below.

The journeymen part can be explained by the winner's curse - whoever bids highest for a journeyman is the one who most badly misoverestimated his value.

The stars part can be explained, I think, by risk - you don't pay $40M even to someone who has delivered $40M of win value for eight years running, because there's always a chance he gets hurt or his skills diminish. Trout in particular took less AAV for long-term security.

Besides that... yes, there is organizational value to occupying a spot for playing time, and organizations do account for that. WAR shouldn't, since WAR doesn't concern itself with the friction of making transactions or occupying roster spots. But organizations do.

I used to agree with this, but it seems like there are a ton of negative WAR players these days. Just among position players there were 26 players with -1.0 WAR or worse last year, almost one per team. And then another 60+ with at least -0.5 WAR. Not all of this can be small sample bad luck.

Leaving aside the issue of exactly where "replacement level" should be set and whether talent is ever really "freely available" ... this is exactly what small sample bad luck looks like. Take a pool of 150 replacement-level players that are expected, on average, to produce 0 WAR in their small samples and there will be a chunk that produce -1 WAR, a chunk that produce -0.5 WAR on up to a chunk that produce 1 WAR. There will be teams where none of them are as bad as -0.5 or -1 only because they gave one guy 100 PAs, he stank to they gave the next guy 100 PAs and he stunk and they gave a 3rd guy 100 PAs and he stank, all due to random variation. Add those three up and you'd get -1 WAR.

As noted, teams don't KNOW this guy is an exactly 0 WAR player and that guy is an exactly -0.5 WAR player and the third guy is exactly 0.5 WAR. They decide "this is the best OF we have available right now to soak up 40 PA this month while Stanton is on the IL." Maybe he is worse than replacement and that guy the Ms just put on waivers would be 0.2 WAR better over a full season -- how would they know? Nobody's knowledge is that precise. And to the extent we want to take the term "freely available" literally, it's only true during the offseason when you can choose from a big pool of free talent. Once the season starts, it's guys in your system, whoever was waived this week and maybe a handful of guys you can get for just cash. When it comes to position players, most of these emergency replacements aren't on your 40-man so there's that cost to consider as well.

Identify the pool of replacement-level players before the season starts. At the end of the year, see how they did as a group and even put together some weighted distribution of performances. If that average comes out substantially negative or the negative tail is thicker or longer than expected then you've got something.

An informal "study" of Cub replacement players 2022 (this is gonna take a while, there were a lot of them) ... WAR PAs

Crook -0.2 9 (here WAR is possibly silly, .2 wins in 9 PAs?)
Frazier 0.1 45 (debatable that he was freely available)
Hermosillo -0.6 73
Higgins 0 229
Ortega 0.3 371
Quiroz 0.4 47
Reyes 0 193 (not expected to be rep level but a free waiver claim)
Rivas 0.5 287
Schwindel -1 292
Vargas -0.1 26
Wisdom 0.9 534
Young 0.1 22

That totals 0.4 WAR in 2128 PA. It's debatable to include Wisdom but he had the same track record as Schwindel so either they're both in or both out and the total doesn't change we just drop 826 PA. The Cubs gave Schwindel plenty of rope because he was quite productive last year. We can argue similarly that Ortega doesn't belong which brings the total back to zero; same with Reyes. So you can count that as 0 WAR in about 800 PA up to 0.4 WAR in 2100 PA.

Pitching ... I don't know if I know who's replacement, who's a prospect, heck half the guys with >min contracts I probably don't know the names of but here goes:

Brault 0.2 9
Castro -0.2 11
Dermody -0.1 1
Farrell 0.2 11
Gsellman -0.1 15 (FA at the end of spring)
Hughes 1.4 58 (our big find!)
Leiter 0.8 68
Little -0.3 1 (again, that's silly)
Menez 0.1 1
Padilla 0 2
Rucker 0.5 55 (maybe not but the Cubs did make him available in rule 5 a couple years ago)
Sampson 2.1 104 (so freely available we even waived him during the season)
St John -0.1 2
Stout 0 4

So that's 4.5 WAR in 342 innings which is awesome. Take out the good guys (Sampson, Hughes) and Rucker who's always been in the Cubs system other than his brief rule 5 sojourn to the O's and it's 0.5 WAR in 125 innings. If you add up the Cubs AAAA shuttle relievers (i.e. young guys), they had -0.1 WAR total. If you add up the Cubs' position player pitching you get -0.4 WAR in 5 IP.

Now if we take somebody like Little seriously ... a reliever could throw 19 replacement level innings and 1 crap inning and end up on -0.3 WAR. That's what small pitching samples look like. Meawhile, I'm not sure it's even possible for a reliever (especially a non-closer) to add 0.3 WAR or anything close to it in 1 inning so Little would have had to throw 19 innings of +0.3 WAR to get back to "replacement level."

Meanwhile pity poor Narciso Crook -- he went 2-8 with a double, a sac fly, 2 RBI and a run scored, he caught all 4 balls he got to and he even had an avg EV of 98 MPH -- what more could you possibly want out of a replacement player? -- but 0 BB and a GDP and somehow -1 DRS and the poor guy ends up with -0.2 WAR (a -14 WAR/650 pace!).

To put it in stark terms: Is Freeman really all that likely to continue to miss almost no games, OR is Freeman an example of sample bias because he is simply lucky enough to have avoided injury and in fact he is as likely to miss games as say Soto or Betts or the average guy who plays 150g/year.

The lack of injury time (IL days) being easily available (at b-r or anywhere else) remains the biggest and totally inexplicable gap in baseball data and analysis. That said, apparently the old baseball encyclopedia used to have it and good ol' Steve Treder took out the slate and calculated some basic stuff at my nudging probably around the time Druw Jones was born (OK, maybe later than that). Anyway, it was pretty obvious, and not surprisingly so, that past IL time was a pretty good indicator of future IL time and other limits to playing time. That of course doesn't mean that guys with no IL time aren't at risk of injury.

So comping Freeman to other guys who regularly play, say, 145 games a year ... no, Freeman probably doesn't have an advantage in PT projection. Compared to Mike Trout though, he almost certainly does. We should assume that there is some baseline risk of serious injury (which might increase every year in your 30s) along with standard age-related decline in PT, then add on risk for those who have been fragile in the past. A guy like Freeman would carry only that baseline risk in our projections. But without question, this needs proper, thorough analysis.

There are of course freaks like Rose and Yaz who simply never wear down. There are freaks like Molitor and Edgar who had trouble staying healthy in their 20s who become extremely durable in their 30s.

Crook -0.2 9 (here WAR is possibly silly, .2 wins in 9 PAs?)

I had to peek at this. BBref says he's -0.1 oWAR and -0.1 dWAR. The latter comes from 36 innings in the outfield rated at -1 run, which is plausible. The offensive side has got to be some kind of rounding artifact though. 0-for-9 PA would be worth -2 runs or so and -0.2 WAR, but he went 2-for-8 AB with a double, which can't be that terrible.

James says why in the article. It's because weightings of 1+2+3+4+5+6+7+8+9+10+9+8+7+6+5+4+3+2+1 sum to a nice round 100. Not kidding, that's his reasoning.

LOL. But later he explains how to make it add up for smaller time periods. :)

There are so many issues involved in something like this.

Let's start with "nobody" plays 162 games a year regularly. Sure there's Ripken and Gehrig and Billy Williams and Steve Garvey gave it a good run and Maddux didn't hit the IL until he was 40 or something. But even when you look at HoFers, what you usually find is their HoF case is primarily built on a period of about 13 years and about 8000 PAs. Reggie's a good HoF example for almost everything (exc the importance of 3000 hits) and from 22-34 he had 7806 PA and 70 (of 74) WAR. That's 145 games and 600 PAs per year. He "missed," on average, 10% of the season during his period of greatness.

That's simply what we should expect even out of the very best. That comes in all sorts of distributions obviously -- Reggie pretty consistently missed about 10 games a year in his 20s then 25-30 in his early 30s; another guy might miss a half-season along the way; Griffey or Mantle might be very durable then fall apart -- but over an extended period, you simply can't expect more than about 600 PA a year out of a top player. You get to the inner circle not by being super-durable during those 13 years (which after all will only get you about 1000 more PA) but by having another 6 excellent years outside of those 13.

In their 20s, it's probably more 1 game every 2 weeks; in their early 30s, it's probably closer to 1 game per week. For the most part, any playing time before 23 and after 35 is kinda "random" and usually not vital to an assessment of how great a player was (other than whether they are inner circle).

Which brings us to Larry Walker (and James's notion of replacement level or the importance of PT). Walker's entire career is just 8030 PA which includes a 56 PA cup of coffee. So his career is a top player's 13 seasons but in his case spread over 16 seasons. From 24-36, Walker had 6800 PAs; Reggie's "baseline" was 7800 PAs so that's an extra 1.5 seasons missed by Walker or 75-80 PAs a year. Walker was at 6.1 WAR/650, Reggie at 5.9 WAR/650. It's easy to conclude Reggie was more valuable than Walker due to greater durability and I think it's fine to make the jump to "better" especially given their rate of production was about the same. But of course WAR tells us that straight off. (Note WAA is equal.)

The question is whether Walker was less valuable than a 64 WAR player with 7800 PA. I'm sure there's an excellent real world example somewhere but nobody I looked at quite matches so I'll stay hypothetical. Anyway I assume that player is a bit more valuable than Walker because you don't have to go to the hassle of finding those extra 75-80 PAs a year. Now that's not a lot and it's not really much of a hassle, it just means your 4th OF (who's probably above-replacement) gets 450-500 PAs a year rather than 400 but presumably there are some minor follow-on effects of more PT for your 5th OF or spending $500 K less on your backup IF or something.

But the break-even point can't be very far below 64 WAR. 60 WAR or 0.3 WAR/season maybe (Vlad 23-35, 8100 PA 58 WAR; Murray 22-24, 8400 PA 60 WAR)? I don't think I'm willing to believe it's anything worse than that.

What I am absolutely sure is nowhere ####### close is Freddie Freeman 7300 PA and 49 WAR (with one season to go). Even granting Freedie another full season and 4-6 WAR (his average or 2022) to get him to 53-55 WAR ... that's still 9-11 WAR away from Walker and there is no way that 1000 extra PAs is remotely close to being worth that.

Now that's not quite the question James is trying to answer (I don't think, the website won't load for me) -- he's kinda looking at projected value. But he's doing so from a weird "full career" perspective and it's hard to imagine how some mostly recent gaps in playing time can counter the fact that Trout has about 65% more WAR than Freeman. Surely the only way to get to Freeman projecting better than Trout is to focus on the last 3 years or so -- that's 900 PA and 10 WAR for Trout vs over 1600 PA and 14 WAR for Freeman. So sure, even by theoretical WAR standards, the Angels are not gonna find 700 PA and 4 WAR very easily (not for a lack of trying!) So with hindsight, you'd have to take Freeman (which would also have saved you $40 M) and if you place extra valued on PT, the gap is even bigger. And you might still take Freeman if you took it back 4 or 5 years.

But as SS notes, we don't really know how well these guys project. Freeman is probably the best bet for 600-650 PAs out there but, by the same logic, we also know that's gonna be about 4.5 WAR -- good not great. Fine, a durable 4.5 WAR is worth 5 un-durable WAR. Trout is coming off 6+ WAR in 500 PA. We still have the conclusion that about 400 PAs of Trout is as valuable as 650 PAs of Freeman. I don't think we can predict PT well enough to have much confidence that Trout won't reach 400 PAs.

But again the ultimate failure would seem to be all those guys who are as or nearly as durable as Freddie and simply better players. Why would I want a durable 4.5 WAR 1B when I could have Arenado or Machado? And it's obvious from contracts that teams agree with us. Freddie's on 6/$162 taking him through age 37 ... Xander might beat that, I'm pretty sure Correa and Turner are gonna easily beat that (though they might not get through 37). Arenado has 5/$144 left through age 36; if he passes on the opt-out, Machado has 6/$192 through 36; not the wisest decision but even Bryant got 7/$182 through 36 the same offseason as Freeman. Aaron Judge might well double Freeman's deal in total $ and run past 37. Poll 30 GMs on "who's the greatest player in the game right now" (or who's the most valuable or who would you pay the most) and not one of them will say Freeman. Hell, ask Friedman who's the greatest player on his team and I'm pretty sure he'll say Betts (10/$320 through 39 but heavily deferred), not Freeman; he might say Turner if he re-signs; he might say Kershaw out of nostalgia or Buehler if he hadn't gotten hurt. James must know he's wrong on this one.

The latter comes from 36 innings in the outfield rated at -1 run, which is plausible.

Actually it was 18 innings. Because b-r adds the "OF" line, it's easy to double count.

EDIT: On Crook's offense, I assume it's the 222 OBP and 72 GDP/650 pace that kills him. The 72 doubles and 144 RBI would be cool but that's a lot of outs.

To be clear, I don't expect WAR to "make sense" for 9 PAs or 1 IP but it's illustrative when we ask a question like "how come they didn't know that guy was a below-replacement reliever?" when the answer might well be "he threw two terrible innings in his 30 IP rather than one." The surprise breakout guy was lucky enough to throw 0 terrible innings ... he'll throw 2 next year.

First off, the more I think about it the less I can find a justification for Freeman as #1. That aside... here's a big ol' ramble on replacement level:


I think we've done a pretty good job of pegging replacement level for position players, but maybe a less than great job of explaining why it's what it is and complications in assuming that it's a constant.
For position players, it moves throughout the baseball year. It is easier to find a cromulent player now (when the labor market is fairly liquid) than in May. Those shifts are not equal by position either -- replacement level for a first baseman is lower relative to other positions in-season than it is out-of-season, because, while first isn't the hardest position to learn, there is an adjustment period in doing so (there's stickiness). I contend that this is a bigger issue at the ends of defensive spectrum than in the middle.
Also:
teams aren't always (short term) win maximizing, which I think is implied with this kind of methodology. Part of this stems from arb clock and 40 man roster management, which can run counter to win maximizing strategies
as salaries have risen, I think clubs are more likely to stick with established commodities than unknown players, partly because of sunk cost fallacies, partly because struggling but established players have (on average) higher "good outcomes" than similiar unknown players, by dint of past accomplishments as evidence that they have ever been capable of reaching those heights

Pitchers are another story. Not only is it harder (imo) to assert how good someone is (compared b-ref v fangraphs WARs for a FIP oriented reason), but role is hugely important with pitchers in a way that lacks an analogue on the hitter side. Starting versus relieving. Openers. Guys on the AAA/Maj shuttle. Win state comes into play in a way we don't need to contend with for hitters. It's tough.

And that gets to one last ramble around role that does apply to both pitchers and hitters... You deploy star players in a way that serves to maximize how they can help you (by and large). You deploy replacement level talents in a way governed by what your healthy stars (and average players and key backups and...) can do - they are spackle, they are afterthought. There are a lot of above replacement level players that don't see much big league time because they lack tactical applications or much of a ceiling or things like that. So, this slightly distorts what we see as well, in terms of who fills those scrub innings.

no. 47 is a wonderful analytic piece with more insight that almost anything Bill James has come up with lately. Also some nice micro bits re: Crook etc. That was enjoyable.

So comping Freeman to other guys who regularly play, say, 145 games a year ... no, Freeman probably doesn't have an advantage in PT projection. Compared to Mike Trout though, he almost certainly does.

Right so I guess my next question is: if we account for that does Freeman's advantage vs other plausible candidates e.g. Judge Soto etc. go down? And by how much? if you could guess. thanks.

...replacement level for a first baseman is lower relative to other positions in-season than it is out-of-season, because, while first isn't the hardest position to learn, there is an adjustment period in doing so (there's stickiness).

dk: you make a number of excellent insights, but this part I dont understand: arent we to assume that replacement players are already playing the position they are being asked to fill? ANd why would first base take some getting used to and not the other positions? You probably have a good answer but it would be helpful if you could elaborate.

Good thread. I personally think that Ohtani is the best player in baseball, but maybe that is just because he is unique and cool.

I think we've done a pretty good job of pegging replacement level for position players

It's been a while since I really dug into what I can find of the WAR methodology, but I've always thought replacement level is set too high on the defensive side. IIRC, replacement level hitting is defined as 1 run per 27 outs worse than average, and at that point it's clear the hitter is terrible by traditional metrics. But with (again, IIRC) replacement level fielding defined as league-average, is it really obvious that, for example, a 40th or 45th percentile fielder should be replaced? Should half the league really be below replacement level? Sure, it's easy to shake the Glove Tree and have a lot of players fall out, but if none of them can hit at all, what team is going to make that tradeoff?

On a side note, I would greatly appreciate any pointers on how to calculate (or even reliably estimate) dWAR from publicly available metrics. I don't think I've ever seen it fully explained, but I'd like to be able to separate the overall components by position.

47. I just saw Dan S answer a question about the quality of ZIPs using a very similar construct.

It's been a while since I really dug into what I can find of the WAR methodology, but I've always thought replacement level is set too high on the defensive side. IIRC, replacement level hitting is defined as 1 run per 27 outs worse than average, and at that point it's clear the hitter is terrible by traditional metrics. But with (again, IIRC) replacement level fielding defined as league-average, is it really obvious that, for example, a 40th or 45th percentile fielder should be replaced? Should half the league really be below replacement level? Sure, it's easy to shake the Glove Tree and have a lot of players fall out, but if none of them can hit at all, what team is going to make that tradeoff?

On a side note, I would greatly appreciate any pointers on how to calculate (or even reliably estimate) dWAR from publicly available metrics. I don't think I've ever seen it fully explained, but I'd like to be able to separate the overall components by position.

I believe dWAR on bbRef is just Rfield and RPos then converted from runs to wins - it's not going to give you something exact, but you could just add in Rrep to that to make a dWAR versus league wide replacement level.

Really though, oWAR and dWAR are more obfuscating than anything - it's very strange that the positional adjustment is included in both while the replacement level adjustment is only included in oWAR. I prefer to just group Rrep and Rpos, which together give you a 'just showing up' sort of credit.

thanks for the kind words, sunday:
replacement level comes from the pool of position players, with position level adjustments based on how hard that spot is to defend.

from there, i see three forces impacting ease in taking on a position (i'm simplifying): prior experience, similarity to past roles, and aging (which doesn't matter too much here, since i'm talking about short term transitioning). for past experience, we see players generally moving rightward (from hard to easier) positions over the course of their careers - shortstops become second basemen, third basemen move to first, etc... so players tend to have more experience at positions that are harder, that they'd be worse at (after whatever transition/learning period is complete). for similarity to past roles, i think of this in two ways: 1) the skills involved (why transitioning between outfield spots is easier than, say, moving from third to an outfield corner), 2) the level of difficulty. with this second bit, it's easier to transition between positions in the middle of the defensive spectrum than to outlier spots. catcher for sure, which i think you could reasonably say isn't part of the spectrum, but also short (despite its similarity to second, it's just hard) and first.

put all of this together and you get that first base is the toughest non-catcher/pitcher transition IN THE SHORT TERM (all of them assuredly have some kind of transition period). now, it's not rocket science and you can obviously pick it up (it's the easiest position, after all) but this is one of the reasons why you get weird, sub-replacement first basemen more often than you might expect (see harold castro for detroit this year).

no. 47 is a wonderful analytic piece with more insight that almost anything Bill James has come up with lately. Also some nice micro bits re: Crook etc. That was enjoyable.

Thanks. FTR, I don't consider anything I do here to be "analysis" (that's what I did for a living), it's back of the envelope stuff, at best the sort of stuff you do in planning out a genuine data analysis. And nitpicking myself, Reyes definitely shouldn't be in the "freely available" bucket -- he was a waiver claim but the Cubs took on something like $1-1.5 M in salary. Since he came in at 0 WAR, it doesn't really matter though.

Vargas is an interesting example too. For ages 25-27, he had about 250 PA for the DBacks, 0.7 WAR, -0.2 WAA ... that's good bench production. At age 28, in the covid year, he managed to play for 3 teams (Ari, Min, CHC) and put up -0.6 WAR in just 54 PA -- in 54 PA, he wiped out all his solid work in AZ. In 2021, he again played with 3 teams (CHC, Pit, Ari) for -0.4 WAR in 83 PA. In 2022, he was back on the Cubs (via some transaction so obscure b-r doesn't even have it) for -0.1 in 26 PA then went onto the Nats where he put up +1.5 WAR in 196 PA. Where did that come from?

So Vargas is now 31, has 600 career PA, 1.2 WAR, -0.8 WAA. On the surface at least, that's a solid bench player -- not good enough to start but a 1 WAR player not a 0 WAR player. We could supplement that with his minors performance but he seems a pretty popular 29th man so I assume he does pretty well there. Is he a 1-WAR player or a 0-WAR player? If you looked at AZ 2019 or WSN 2022, you want him on your bench or even starting; in his other 200 PA spread across 4+ years, he's been below-replacement. That's what 11 small sub-samples from a not very large 600 PA sample look like. Nobody can say with much confidence whether Vargas is 1-WAR, 0-WAR or even below-replacement. Interestingly, he just passed 3 years of service time and the Nats signed him for $975 K already.

if we account for that does Freeman's advantage vs other plausible candidates e.g. Judge Soto etc. go down? And by how much?

I think I've answered that with the Larry Walker bit. My guess is that a durable 4.5 WAR player (Freeman) is maybe as valuable as the guy who puts up 5 WAR while missing about 1/3 of a season. That's assuming you can "guarantee" Freeman will be healthy and that the other player "always" misses 1/3 of a season. So Judge has had two seasons like that and still managed to put up 5.5-6 WAR; he's also had 3 full seasons and put up 25 WAR, numbers Freeman has no hope of reaching. When Judge is fragile, he puts up more WAR than Freeman; when Judge is healthy, he blows Freeman out of the water. I don't think that one's close.

Over the last 3 years, Freeman had nearly twice as many PA as Trout; also Trout produced at a lower rate of "just" 7.3 WAR/650 in that span. If we took those numbers seriously and used them to project PT and performance, Trout would project to just something like 3.5-4 WAR in a year and leave you needing a replacement for 200-300 PA. So I'd take 650 PA of 4.5-5 WAR over 400 PA of 3.5-4 WAR -- basically the equivalent of finding an average player to replace our projected Trout.

And as I've said, there are durable players that have shown themselves to be (a bit) more productive than Freeman. Maybe you can make an argument for Freeman that, as a 1B, his chances of remaining durable are higher and that, in comparison to SS/3B, his bat is easier to move to DH as he ages so he'd be a safer bet on a long-term contract. Maybe but I'd take Soto, Machado, Arenado and maybe Turner ... and obviously Ohtani and Judge ... and probably Correa, Lindor and maybe Seager ... I've got my doubts on Semien and Acuna's an interesting question. I'll definitely take Freeman over Renndon though. :-)

Anyway, I think there are clearly 10-15 better players in the game right now (not even considering somebody like Alcantara) and Freeman's contract status compared with others makes it clear real baseball decision-makers agree with that. But, yes, it's possible that Trout is now fragile enough and reduced enough that he's down at Freeman's level.

None of that is a shot at Freeman -- he's a very fine player, he might well end up in the HoF.

James seems to be trying to answer a really odd question -- "after it's all said and done, who will we say was the great player (from a career perspective) playing the best in 2022?" It's sort of ... Chipper and Griffey were both great players but, if we lined them up by ages, then you'd obviously rather have Griffey in his 20s. But at some point along the way, you'd rather have Chipper. Now let's add Thome, Bagwell, Thomas, Trammell, Piazza, Biggio, Alomar and Raines to our comparison pool. (Ignore the elephant in the game.) Who was the best great player of 2003? But don't answer that using just 2003 stats. It wasn't Griffey, he'd fallen apart. Maybe it was Chipper. Maybe there's a year in there where Biggio or Bagwell sneaks through.

Somebody around here does something similar with pitchers, using a moving window to say "for 2003, the best active pitcher was..." Once the stats are in the book, it would be easy enough to calculate a 5 year moving window of WAR where, for year T, the window is defined as the 5-year window of years T-2 through T+2 and say "this was the best player of 2003." Based on what you folks say, James seems to be doing that only over 19 years (which sounds nuts to me) but also, for current players, imputing (i.e. filling in) the missing years from those 19 and then ... well I still don't know how you get to Freeman as the greatest.

But sure, just as Chipper (or Thome or, heck, Bobby Abreu) was a better player than Griffey in 2003 because Griffey couldn't stay on the field, Freeman might be better than Trout right now. And if I had to decide which to pay tens of millions over the next 5 years then I'd have no choice but to trust my highly uncertain projections of their futures. But we have the luxury of sitting on our uncertainty, admit we don't have a good idea which of these guys will have more PT in the future, hope Trout stays healthy and sit back and wait to see how it turns out.

Griffey v Chipper is a sort of interesting question -- which career do we "prefer" or is "better" or what have you.

The totals are close: Chipper 85 WAR, 53 WAA; Griffey 84 WAR, 46 WAA. I'm surprised to see that Griffey ended up with 700 more PA but a lot of that WAA difference is really the league difference (which is put into Rrep in bWAR which is annoying in this sort of comp). That is the difference in Rrep is 64 in Griffey's favor but, based purely on PT, the difference should be only about 22 in Griffey's favor ... so about 42 of that is league difference ... Griffey is being compared to a higher average than Chipper then receiving that adjustment in Rrep/WAR, not WAA. So the WAA difference is "really" only about 2.5 to 3 which is essentially Griffey adding one extra shitty season at the end.

Griffey was more power-oriented, Chipper more BA-oriented but the overall offesive production is close, Chipper winning on OBP. They come out very close on defense -- 3B and CF are simillar in Rpos and Griffey the very good young CF became Griffey the very bad old CF while Chipper was (by DRS) not nearly as bad as his reputation.

The difference somes in career shape of course. Due to injuries at the start of his career, Chipper didn't get started until 23 by which time Griffey had 2400 PA under his belt. But Chipper was never a below-average player and made up most of that ground in his 30s.

Griffey 19-30: 7300 PA, 76 WAR
Chipper 23-34: 7500 PA, 60 WAR

Griffey 31-36: 2200 PA, 7 WAR (13 oWAR)
Chipper 35-40: 3100 PA, 25 WAR

Griffey 37-40: 1800 PA, 1 WAR

In its way, a classic peak/prime vs career argument. Early Griffey teams had a better chance of winning a division/pennant, late Chipper teams still had an excellent player to help them to a division/pennant. In the end, the durability difference gets obscured by Griffey hanging on but was anyway just 1100 PA in their effective period. I'm not clear whether James is saying Chipper was the greater player because of those extra 1100 PAs or he's just saying that, by the early 30s, we had seen enough missed time by Griffey what we obviously should choose Chipper for ages 33-40. Regardless I don't see how that would make Chipper the greatest player of the early 2000s ... but at 35-36, he did have 7-WAR seasons so I'm not ruling it out either.

Chipper was certainly one of the greatest players for ages 31+ -- 40 WAR, 25 WAA, never below-average -- and, without question, by 2003 you'd take him over Griffey (maybe even the 38-yo Bonds) but that doesn't necessarily mean he was a serious candidate for best player in the game for those years.

"replacment defense" ... dWAR was an unfortunate choice of acronym. But there's actually nothing in bWAR that defines "replacement level" offense or defense. An oWAR of 0 seems to imply such a thing but really an oWAR of 0 is "the player would have 0 WAR if he was an average defender at his position." In that sense, it's easy to conceptualize an "assumption" that a replacement level player is a horrible hitter and an average defender but there is actually no such definition or assumption of such in bWAR. A replacement-level player is just one whose offense and defense combined, relative to average, happens to negatively balance out their playing time. Rrep is awarded at a rate of about 1 run per 30 PA so a 100 OPS+ hitter playing an average or slightly worse 1B or a 80 OPS+ hitter playing an average SS or a 100 OPS+ hitter playing late Bernie/Griffey defense in CF might all be replacement level or close to it.

So, in theory, replacement level over a full season could be -2 wins on offense with average defense or -1 win on offense and -1 win on defense or 0 wins on offense and -2 on defense. But in practice, we see an awful lot more of the first one than the last one, especially among actual replacement players (as opposed to a former star like Griffey/Bernie/Miggy/Pujols). Generally, if you can hit to a 100 OPS+, they'll find a spot for you in the majors, even if you are a 1B/DH type, even if it's just as a bench bat. Meanwhile guys who can play an adequate SS and speedsters who can play a solid CF but can't hit a lick are a dime a dozen. Defensive production is just easier to find -- there are maybe 2-3 Vogelbachs stuck in the minors; there are about 50 Vargasses plus another 50 young IFs. So a simplification that a replacement player is a crappy hitter and average defender is reasonably close to the truth.

So a simplification that a replacement player is a crappy hitter and average defender is reasonably close to the truth.

Agreed, though maybe for a slightly different reason. There's more value wrt short benches in more versatile guys than slightly above average bats who don't do much else well (and arguably tactically, given pinch running/def replacement opportunities) ... but I think that's overwhelmed by a tendency of players to slide down the defensive spectrum to a place where they can be at least average / to make way for better prospects or players to do what they do best. (Which is part of what makes those guys a dime a dozen?) Another contributing factor is the pull of other markets, like Japan or Korea for better bat guys versus glove men.

Walt reminds of the 'argument' in the NBJHBA that because Biggio had passed Griffey, who had been the best player in baseball, Biggio was now the best player in baseball. The whole point was the comparison to Griffey, no young player was ever considered.

There's more value wrt short benches in more versatile guys than slightly above average bats who don't do much else well

I dont understand what you're saying here. Assuming the premise is true: versatility is preferred. WHy should those inf/of or whatever guys be avg defender/below avg batter? Why? You cant find an inf/of who hits and plays a crappy defense? I mean you're saying this likes its so obvious but why?

(and arguably tactically, given pinch running/def replacement opportunities)

Ive read this 17 times and am getting a stroke. "arguably tactically"? and given PH..? what? Are you saying there's more chances to be a def. substitute than a PH? I doubt it but I have no idea what you're saying here.

Walt reminds of the 'argument' in the NBJHBA that because Biggio had passed Griffey, who had been the best player in baseball, Biggio was now the best player in baseball. The whole point was the comparison to Griffey, no young player was ever considered.

Or older player - like Bonds. I remember thinking in the early 90s that Griffey would eventually be the best player in the game, because of his youth, but Bonds was better at the time. Thing is, Griffey never passed Bonds. By the time Bonds was forced out of the league, Griffey wasn’t much of a player either.

sorry, this is what i get for writing when i've a spare minute here or there, let me try again (written in similar circumstances, thus could also be jabberwocky and internal shorthand).

You cant find an inf/of who hits and plays a crappy defense?
this is pretty easy to find for near free through most of the defensive spectrum! (if hits means ops+ in, say, the 90s.)

ok - so, i think i need to explain why i think:
* that the premise is true that teams tend to play players at positions where they are near average or better
and
* why should those inf/of or whatever guys be avg defender/below avg batter?

i'll focus on the latter.

the bench (versus off-bench replacements) needs to be primarily composed of people who can cover multiple spots (if we had 60 man active rosters, this would be a different conversation). so, the reserve infielder has to be able to handle short; the 4th of has to be able to handle center, and so on. there are exceptions to this, but this is true in the main. thus, if you have 2 players who would have near equal value if each got 500 pa in the bigs, but one is a decent hit, weak glove middle infielder (let's call him esteban quiroz) and another has better d but less bat (let's call him kevin newman) - teams will not only be more likely to play the latter, but they'd also be more likely to keep guys like newman on a 40 man roster because there are more scenarios where you see them as a viable fill in in an emergency, as newman meets minimum requirements defensively at more spots. (and you're viewing this guy, whoever he is, a replacement type, you view them more in terms of how they can handle contingency scenarios versus how are they as an everyday player in the medium or longer term.)

as for tactical stuff - maybe i should leave this bit alone, it's more of an aside - but. i'm already contending that bench players are worse than starters with the bat but not much worse (or maybe better) defensively (less likely to apply at short, some at catcher and center). so, what are the ways you use bench guys besides as replacement starters? pinch hitters, pinch runners, pinch, uh, fielders. (defensive replacements). the bench player is often a worse hitter likely a worse hitter than the starter (at their position or those at easier spots), even before the pinch hitter penalty (players hit worse as pinch hitters, ceteris paribus, than when starting the game) - the places where you can get a benefit mostly surround the platoon advantage and things like using a corner infielder/outfielder to hit for a shortstop or catcher or whatever. otoh, if you're looking for a pinch runner or a defensive upgrade, it's often not hard to find a bench option than is better than the starting option (the fourth outfielder is a better fielder than the worst start corner outfielder, the fastest reserve of or if will likely be faster than a starter with below average speed). there are more types of tactical uses for the kinds of players who we see on benches than for the guys we affix labels like aaaa player to. (who are above replacement level players that don't fit mlb needs.)

In practice, replacement level at different positions seems also to involve demographics. In the past decade the Rangers have had a string of young fast CF as fourth outfielders: Leonys Martin, Delino DeShields Jr, Carlos Tocci, Leody Taveras; their careers tend to be over before they're 30. But they've had a similar string of backup catchers who are old guys: Yorvit Torrealba, Geovany Soto, JP Arencibia, Chris Gimenez, Bobby Wilson, Jeff Mathis for a couple of years who must have been fifty and was batting .096. Are most clubs like that? Or is that a hallmark of a bad team.

The principle is the same, I guess, "free" talent at a basic rock-bottom level, but the talent search draws on different pools of availability.

DerK: this is a much more well thought out argument. I dont necessarily see that it should be true for the overall bench player but most people seem to think it is true in practice.

For instance, 2 inn. of defense is roughly akin to one PH appearance. Is a replacement more likely to do 2 inn of def then 1 PH? Im not sure why. Are replacements more often used when you're ahead? Also there's the issue of speed and as you suggest better defenders tend to have more speed. BUt yeah I understand your argument much better. thanks.

was it true in the old days E.g. the 60s or 70s. Were replacment players decent def/bad off?

i should defer to people who aren't talking out of their posteriors -- but!

catcher is a position where plus defense is more valued than at other spots and, critically, one where you can't shift top defenders to a tougher catcher-like position, so good defenders are quite likely to start as well. last i looked at this at a macro level, starters were better than average with the glove, primary backups a hair worse than average, other players kind of a hole defensively (they were most of the negative that counteracted the starters being positive, since this is a zero-sum exercise). also, backups were slightly older than starters (like a year on average?) and other catchers were quite a bit younger.

setting aside rambles about how the good d/bad o backup catcher is a longstanding archetype, exacerbated by the use of backup catchers as defacto extra coaches bitd (the explosion in coaching staff size reduces the need here) and where catching skills as perceived by teams decline at higher ages than they do at our positions (whether because the lack of dependence on range, which peaks early, or because catching defense might be more knowledge or experience based than we see at other positions)... i'm guessing texas' choices as specific to that team's internal preferences and dynamics. :) same for fourth outfielders, where i think a "tweener type" is a more common 4th outfielder type - though i haven't studied this and i don't know how the dh complicates things.

as for the 60s/70s, dunno. i think roster dynamics were fairly different then, as bullpens were smaller, financial/40 man considerations mattered less, and there were more opportunities for specialized position player roles.

i think the key idea with all of this is that how you use and roster your bench is a function of how you use your best players and roster/financial rules - the rest trickles down.