OK, so we had a lot of fun last week with a little story we called My Favorite Advanced Stats (and Why I Like Them). There are, apparently, lots and lots of questions from Brilliant Readers out there about advanced stats, how they work, what they’re missing, why they’re pointless, etc.
I consider myself a big fan of WAR. I like to consider it as one of the most appropriate ways to measure the total greatness and value of a player.
However, there is one part about WAR that I cannot really be a fan of, and that is the positional adjustment factor. Of the six components that make up Baseball Reference WAR, the other five components are all things that a player has at least some control over. Unlike those other five components, no player can control the position they play.
I think the biggest piece of evidence I have with this is Derek Jeter. As has been proven and as you have pointed out before Joe, Derek Jeter is rated not only as the worst defensive shortstop in history, but as the worst defensive player at any position in history. And yet, because he played shortstop, he is credited with 144 extra runs to his WAR. Because all were shortstops, he gets almost the same amount of runs added to his WAR as Ozzie Smith and Cal Ripken, who by fielding runs, are the 3rd best and 11th best among all players, and are the 2nd best and 4th best among shortstops, and a bigger advantage than Mark Belanger, who is listed as the best defensive shortstop and the second best at any position.
While third base is listed as another position with defensive importance, Jeter is also given a bigger boost in positional adjustment than Brooks Robinson, Adrian Beltre and Scott Rolen, who are the best, 5th best and tied for the 12th best in fielding runs among all players and are the top three third basemen.
On the other hand, Carl Yastrzemski and Barry Bonds are listed by fielding runs as the 9th best and tied for the 12th best among all positions and are the top two left fielders, but since their position is not thought of as important, lose 171 and 101 runs, respectively.
Roberto Clemente and a personal favorite of mine, Albert Pujols, are ranked by fielding runs as the 6th best and the 38th best among all positions and as the best right fielder and first baseman, respectively, but they lose 102 and 178 runs for not playing positions thought of as important defensively.
All of these examples are to say that, again, I love WAR, but I just think the positional adjustment is, at least at times, so unfair because it almost sounds like you are either inappropriately crediting a player for defense or inappropriately penalizing a player for defense, and makes some player’s WAR either higher than it probably should be or lower than it should be. Can you explain why a positional adjustment is something that is included in WAR? Is there any good reason for including it in WAR?
Im Keith Hernandez is compared to the average fielding 1B.
The average fielding SS is a much much much better fielder than the average fielding 1B.
You need a way to scale them.
Let me give you an analogy that affects your kids. Your kid is an AP Math, and that class has 20 students, all who are the best at math. The average grade is a 3.0. Your kid is a 3.0.
The regular math class has 100 kids. Their tests are easier, they cover less stuff. The average grade is a 3.0. Your neighbor is in that class, and his grade is a 3.7.
What do you do?
Well, what IS being done is that every kid in AP Math gets a 1.0 bonus to their grades. The average AP math kid, who got 3.0 in class gets an adjusted math score of 4.0.
Now, whether it should be +1.0 or +0.789 or +1.12, I don't know. But CLEARLY you can't treat the two math groups the same.
RBIs used to be looked at as an indication of being "clutch". now the stat is shot down as an indication of opportunity rather than ability. so what stat can we use? Total Bases plus sac flied with RISP? what do you suggest?
As a preamble to this question, I know FIP is a better predictor of future success than ERA but I have also read that in very select circumstances, because it does not look at bagged balls in play at all, it might miss the nuance of what the pitcher was doing (producing weak contact, etc.) My question has nothing to do with FIP but with wOBA and xwOBA - is there any nuance missed in what the hitter is doing with xwOBA in the same way that FIP might miss nuance (while still being superior)? Put another way, can having a wOBA higher than an xwOBA be a skill or just luck (better defenders, etc.)?
Both tell the story. You hit a hard shot that Kiermaier dives for an out, and it looks like it's the same as a lazy fly out. Or, if you look at HOW hard it was hit, then it's totally different.
You really have to decide: do you want to describe the PLAY or the PLAYER? This is what it all comes down to. Is a sweat-induced Nuke Laloosh Perfect Game the same thing as a breezy-in-control Felix Hernandez Perfect Game?
Thanks so much for your answer - it makes a lot of sense. I think in that formulation I prefer wOBA to xwOBA - it is what actually happened not what should have happened. A hit is a hit an out is an out.
To me, I prefer describing players. Wind blowing in or out, a fielder making a great play or falling down, all those are not reflective of the batter. Just to their circumstances.
I know I can find these answers somewhere, but you said to ask. WAR includes offensive and defensive contributions, correct?
How good is WAR at comparing players across eras? Trout is about the best player I have seen, but I don't think his WAR totals for his career will approach Babe Ruth or Willie Mays. Can one say that the replacement player is better in today's game, so Ruth and Mays were profoundly better than the average player, but the difference between Trout and the replacement player has narrowed? This has always perplexed me.
WAR was built to compare players within the same season. It's not built to compare players across eras, because there are certain assumptions that come into play here. You could use it as a step into those comparisons, but as I said, you need assumptions to take those leaps.
This is true of ALL stats, and so, WAR should not be held to any higher standard.
When I first bought a Bill James Baseball Abstract, I thought it was awesome. Read it every year and then the Historical Abstract came out and it was even better.
There are certain greats I wonder would get the opportunity in today's game. Would Carew & Gwynn be languishing in the minor leagues because of lack of power and speed is valued much less? Would Rob Deer be an every day starter? Would Ferguson Jenkins always said he pitched better on 3 days rest when his arm was tired, would his spin rate be too slow to crack a major league starting staff?
I love WHIP, only good pitchers have good WHIP rating. OPS is great. My Dad always said Al Oliver was an empty .300 hitter. I think Defensive WAR for 1B & Catcher leaves much to be desired.
I love advanced stats! Almost as much as I love Joe's writing. I do find the pitching ones a little more difficult, but that's likely because I never internalized how ERA is calculated. All I needed to know then was lower number better for ERA, and for SO and W larger number better lol
Have we seen a bad pitcher with great spin rate eventually become a good pitcher? I think of someone like Nick Pivetta, who every year seems like a popular pick to "break out" but never does.
Nick Pivetta is a guy who, standing on the mound, just looks like he should be a good pitcher. He's got those wide shoulders and a calm yet serious demeanor. Never understood (well, yeah, I understood but did not agree with) why the Phillies traded him.
There's one for each facet of fielding play. So, one for fielding, one for catching, one for blocking, etc. Eventually, we'll have one leaderboard that will merge them all together.
Thanks to everyone for their input. I consider myself a "numbers guy." But like most of you I have a life and don't have time to dig in to these intricacies the way I would like to. Great thread.
When Statcast debuted a few years ago, it was supposed to make defensive evaluations completely objective. What exactly can we objectively measure about individual player defense with Statcast that we couldn't before? Why haven't I seen any Statcast-based defensive rankings in the mainstream? I heard part of the reason may be that the best Statcast defensive data is owned by teams/MLB and they don't want to release it all to the public, is this true? Does fWAR and bWAR now use Statcast defensive data to evaluate players? Sorry for all the questions, but I would love a little explaining of how Statcast data "solved" the problem of evaluating individual defenders. Joe, maybe you can do a longer piece on this?
Wonder if you know of the astounding stats of Eddie Stanky and how do you think he would be viewed in today's game. I often look at stats for a player's birthday on Baseball Reference, and last Sept. 3, I clicked on Stanky's page. It truly is astounding. Stanky hit all of 29 career homers during an 11-year career from 1943-53, and yet three times he led the National League in walks: 148 walks in 1945; 137 in 1946; 144 in 1950. By comparison, Willie Mays had only 1 season with more than 100 walks and that was late in his career. How did Stanky do this? Why wouldn't pitchers throw to him? He must have had some incredible 12-15 pitch at bats. Naturally, with all those walks, Stanky's OBP was terrific. In three seasons, he had an OBP over .436, including .460 with the Giants in 1950. .460! Joe DiMaggio never had a season with a .460 OBP. For his career, Stanky only had a .268 batting average, but he had a .410 OBP. Now, nobody would ever say Stanky was better than Ichiro Suzuki, but Suzuki was a career .311/.350. He only had one season with an OBP over .400. Back then, Stanky, nicknamed "The Brat," was viewed as a gritty role player, although he did finish third in MVP voting in 1950. After all, he only had a .268 career BA and 29 homers. But with a career .410 OBP, including a season at .460, how would he be viewed in today's game?
Joe made a reference that you discount batting average compared to OBP and slugging. In today's game, it seems like Stanky with his .268 BA, would be a much more valuable player than a Tim Anderson, who has hit more than .300 the last 4 seasons, including a league-leading .335 in 2019. But the Sox shortstop can't get a walk to save his life. He has a career .316 OBP. It seems to support your case about batting average compared to OBP. Anyway, as Joe says, Eddie Stanky is my version of going down the rabbit hole. Thanks for the reply.
How much do you value FIP and with it, fWAR? I get that on a purely data-driven level strikeouts, walks, and home runs are the only things a pitcher fully controls, but I also know that these players are not robots. A pitcher is going to attack hitters very differently if he has the 1999 New York Mets behind him (or any other excellent defense) as opposed to the 2022 Nationals. That's going to affect pitch selection, how "perfect" a pitcher feels he has to make each individual pitch, etc. I've just always felt that it's best to not ENTIRELY separate pitching from defense because defense informs pitching to a degree (and vice versa), so I try to take FIP and fWAR for pitchers with a grain of salt.
1. In this enlightened day and age, why do we keep referring to "innings pitched" as the indicator of a pitcher's workload and capability? Not all pitches are created (or thrown) equally, and not all innings are created equal. Is there not a better way for teams/fans to have a better understanding of workload and stress induced on a pitcher's arm? Seems science has surpassed the archaic way of just counting how many innings.
2. I'd love a better understanding of how to determine a reliever's effectiveness and value. Saves and holds don't do it, and honestly neither does ERA. Some combination of WPA, HLI, strikeout rate, strand rate, pitch count (to thus measure efficiency), WHIP, walk rate, etc. Surely there's a way to condense all of that into a single metric that can really grade out who the top guys are.
1. IP is ubiquitous. But we track batters faced and we track Pitches Thrown. If someone wants to use those, they are readily available, at Savant, Fangraphs, Reference.
So why do teams still refer to an "innings limit" as a way of managing workloads for pitchers? Is it just because it's a simple enough (ubiquitous, as you say) way for everyone to understand the target?
I consider myself a big fan of WAR. I like to consider it as one of the most appropriate ways to measure the total greatness and value of a player.
However, there is one part about WAR that I cannot really be a fan of, and that is the positional adjustment factor. Of the six components that make up Baseball Reference WAR, the other five components are all things that a player has at least some control over. Unlike those other five components, no player can control the position they play.
I think the biggest piece of evidence I have with this is Derek Jeter. As has been proven and as you have pointed out before Joe, Derek Jeter is rated not only as the worst defensive shortstop in history, but as the worst defensive player at any position in history. And yet, because he played shortstop, he is credited with 144 extra runs to his WAR. Because all were shortstops, he gets almost the same amount of runs added to his WAR as Ozzie Smith and Cal Ripken, who by fielding runs, are the 3rd best and 11th best among all players, and are the 2nd best and 4th best among shortstops, and a bigger advantage than Mark Belanger, who is listed as the best defensive shortstop and the second best at any position.
While third base is listed as another position with defensive importance, Jeter is also given a bigger boost in positional adjustment than Brooks Robinson, Adrian Beltre and Scott Rolen, who are the best, 5th best and tied for the 12th best in fielding runs among all players and are the top three third basemen.
On the other hand, Carl Yastrzemski and Barry Bonds are listed by fielding runs as the 9th best and tied for the 12th best among all positions and are the top two left fielders, but since their position is not thought of as important, lose 171 and 101 runs, respectively.
Roberto Clemente and a personal favorite of mine, Albert Pujols, are ranked by fielding runs as the 6th best and the 38th best among all positions and as the best right fielder and first baseman, respectively, but they lose 102 and 178 runs for not playing positions thought of as important defensively.
All of these examples are to say that, again, I love WAR, but I just think the positional adjustment is, at least at times, so unfair because it almost sounds like you are either inappropriately crediting a player for defense or inappropriately penalizing a player for defense, and makes some player’s WAR either higher than it probably should be or lower than it should be. Can you explain why a positional adjustment is something that is included in WAR? Is there any good reason for including it in WAR?
Jeter is compared to the average fielding SS.
Im Keith Hernandez is compared to the average fielding 1B.
The average fielding SS is a much much much better fielder than the average fielding 1B.
You need a way to scale them.
Let me give you an analogy that affects your kids. Your kid is an AP Math, and that class has 20 students, all who are the best at math. The average grade is a 3.0. Your kid is a 3.0.
The regular math class has 100 kids. Their tests are easier, they cover less stuff. The average grade is a 3.0. Your neighbor is in that class, and his grade is a 3.7.
What do you do?
Well, what IS being done is that every kid in AP Math gets a 1.0 bonus to their grades. The average AP math kid, who got 3.0 in class gets an adjusted math score of 4.0.
Now, whether it should be +1.0 or +0.789 or +1.12, I don't know. But CLEARLY you can't treat the two math groups the same.
ERA+ - i don't have a good understanding - what's considered average, what's good, what's great. and why.
So one look can give an instant impression.
I prefer ERA-, not ERA+.
If the league ERA is 4.00 and the pitcher's ERA is 3.00, that's a 4/3 = 133 ERA+ or 75 ERA-.
League average is naturally 100 in either case.
RBIs used to be looked at as an indication of being "clutch". now the stat is shot down as an indication of opportunity rather than ability. so what stat can we use? Total Bases plus sac flied with RISP? what do you suggest?
I'm not really a fan of "clutch", which is mostly Random Variation.
As a preamble to this question, I know FIP is a better predictor of future success than ERA but I have also read that in very select circumstances, because it does not look at bagged balls in play at all, it might miss the nuance of what the pitcher was doing (producing weak contact, etc.) My question has nothing to do with FIP but with wOBA and xwOBA - is there any nuance missed in what the hitter is doing with xwOBA in the same way that FIP might miss nuance (while still being superior)? Put another way, can having a wOBA higher than an xwOBA be a skill or just luck (better defenders, etc.)?
Both tell the story. You hit a hard shot that Kiermaier dives for an out, and it looks like it's the same as a lazy fly out. Or, if you look at HOW hard it was hit, then it's totally different.
You really have to decide: do you want to describe the PLAY or the PLAYER? This is what it all comes down to. Is a sweat-induced Nuke Laloosh Perfect Game the same thing as a breezy-in-control Felix Hernandez Perfect Game?
You tell me, it's your decision.
Thanks so much for your answer - it makes a lot of sense. I think in that formulation I prefer wOBA to xwOBA - it is what actually happened not what should have happened. A hit is a hit an out is an out.
Right, to you.
To me, I prefer describing players. Wind blowing in or out, a fielder making a great play or falling down, all those are not reflective of the batter. Just to their circumstances.
To each his own!
I know I can find these answers somewhere, but you said to ask. WAR includes offensive and defensive contributions, correct?
How good is WAR at comparing players across eras? Trout is about the best player I have seen, but I don't think his WAR totals for his career will approach Babe Ruth or Willie Mays. Can one say that the replacement player is better in today's game, so Ruth and Mays were profoundly better than the average player, but the difference between Trout and the replacement player has narrowed? This has always perplexed me.
WAR was built to compare players within the same season. It's not built to compare players across eras, because there are certain assumptions that come into play here. You could use it as a step into those comparisons, but as I said, you need assumptions to take those leaps.
This is true of ALL stats, and so, WAR should not be held to any higher standard.
When I first bought a Bill James Baseball Abstract, I thought it was awesome. Read it every year and then the Historical Abstract came out and it was even better.
There are certain greats I wonder would get the opportunity in today's game. Would Carew & Gwynn be languishing in the minor leagues because of lack of power and speed is valued much less? Would Rob Deer be an every day starter? Would Ferguson Jenkins always said he pitched better on 3 days rest when his arm was tired, would his spin rate be too slow to crack a major league starting staff?
I love WHIP, only good pitchers have good WHIP rating. OPS is great. My Dad always said Al Oliver was an empty .300 hitter. I think Defensive WAR for 1B & Catcher leaves much to be desired.
I love advanced stats! Almost as much as I love Joe's writing. I do find the pitching ones a little more difficult, but that's likely because I never internalized how ERA is calculated. All I needed to know then was lower number better for ERA, and for SO and W larger number better lol
Have we seen a bad pitcher with great spin rate eventually become a good pitcher? I think of someone like Nick Pivetta, who every year seems like a popular pick to "break out" but never does.
Nick Pivetta is a guy who, standing on the mound, just looks like he should be a good pitcher. He's got those wide shoulders and a calm yet serious demeanor. Never understood (well, yeah, I understood but did not agree with) why the Phillies traded him.
Is there an absolute best defensive stat that tells the most about a player's defensive value, and if so what is it? And if there isn't, why not?
Thanks.
Anything you find on Savant.
https://baseballsavant.mlb.com/leaderboard/outs_above_average?type=Fielder&startYear=2016&endYear=2022&split=no&team=&range=year&min=q&pos=&roles=&viz=hide
Thank you. So if I'm understanding you, there is not ONE best defensive stat, but a few.
There's one for each facet of fielding play. So, one for fielding, one for catching, one for blocking, etc. Eventually, we'll have one leaderboard that will merge them all together.
Thanks. Much appreciated.
Thanks to everyone for their input. I consider myself a "numbers guy." But like most of you I have a life and don't have time to dig in to these intricacies the way I would like to. Great thread.
When Statcast debuted a few years ago, it was supposed to make defensive evaluations completely objective. What exactly can we objectively measure about individual player defense with Statcast that we couldn't before? Why haven't I seen any Statcast-based defensive rankings in the mainstream? I heard part of the reason may be that the best Statcast defensive data is owned by teams/MLB and they don't want to release it all to the public, is this true? Does fWAR and bWAR now use Statcast defensive data to evaluate players? Sorry for all the questions, but I would love a little explaining of how Statcast data "solved" the problem of evaluating individual defenders. Joe, maybe you can do a longer piece on this?
It's available on Savant:
https://baseballsavant.mlb.com/leaderboard/outs_above_average?type=Fielder&startYear=2016&endYear=2022&split=no&team=&range=year&min=q&pos=&roles=&viz=hide
Fangraphs does use it.
And here's an article I wrote three years ago:
https://technology.mlblogs.com/introducing-infield-outs-above-average-6467e61a98dc
Thank you!
Wonder if you know of the astounding stats of Eddie Stanky and how do you think he would be viewed in today's game. I often look at stats for a player's birthday on Baseball Reference, and last Sept. 3, I clicked on Stanky's page. It truly is astounding. Stanky hit all of 29 career homers during an 11-year career from 1943-53, and yet three times he led the National League in walks: 148 walks in 1945; 137 in 1946; 144 in 1950. By comparison, Willie Mays had only 1 season with more than 100 walks and that was late in his career. How did Stanky do this? Why wouldn't pitchers throw to him? He must have had some incredible 12-15 pitch at bats. Naturally, with all those walks, Stanky's OBP was terrific. In three seasons, he had an OBP over .436, including .460 with the Giants in 1950. .460! Joe DiMaggio never had a season with a .460 OBP. For his career, Stanky only had a .268 batting average, but he had a .410 OBP. Now, nobody would ever say Stanky was better than Ichiro Suzuki, but Suzuki was a career .311/.350. He only had one season with an OBP over .400. Back then, Stanky, nicknamed "The Brat," was viewed as a gritty role player, although he did finish third in MVP voting in 1950. After all, he only had a .268 career BA and 29 homers. But with a career .410 OBP, including a season at .460, how would he be viewed in today's game?
I don't think I can give you a good historical perspective here. He'd certainly be appreciated by clubs, players, and fans.
I think the big difference is that a player like this would have been called up earlier in his career.
Joe made a reference that you discount batting average compared to OBP and slugging. In today's game, it seems like Stanky with his .268 BA, would be a much more valuable player than a Tim Anderson, who has hit more than .300 the last 4 seasons, including a league-leading .335 in 2019. But the Sox shortstop can't get a walk to save his life. He has a career .316 OBP. It seems to support your case about batting average compared to OBP. Anyway, as Joe says, Eddie Stanky is my version of going down the rabbit hole. Thanks for the reply.
How much do you value FIP and with it, fWAR? I get that on a purely data-driven level strikeouts, walks, and home runs are the only things a pitcher fully controls, but I also know that these players are not robots. A pitcher is going to attack hitters very differently if he has the 1999 New York Mets behind him (or any other excellent defense) as opposed to the 2022 Nationals. That's going to affect pitch selection, how "perfect" a pitcher feels he has to make each individual pitch, etc. I've just always felt that it's best to not ENTIRELY separate pitching from defense because defense informs pitching to a degree (and vice versa), so I try to take FIP and fWAR for pitchers with a grain of salt.
FIP misses out on things.
ERA (or RA/9) overfits on things.
And in terms of the lesser-of-two-evils, FIP wins. ERA really (really really) clouds your judgement of a pitcher.
You can read more here:
http://tangotiger.com/index.php/site/comments/when-does-past-era-become-more-predictive-of-future-era-than-past-fip
How much does the desire for a one rating to rule them all detract from a true understanding of the game?
Sorry, I don't get the question.
It is now noon, ET. I'll check back here in one hour, then will shut myself down.
All the questions have been tremendous, and the environment has been extremely hospitable. Thank you for this thread Poz, much appreciated.
Two things:
1. In this enlightened day and age, why do we keep referring to "innings pitched" as the indicator of a pitcher's workload and capability? Not all pitches are created (or thrown) equally, and not all innings are created equal. Is there not a better way for teams/fans to have a better understanding of workload and stress induced on a pitcher's arm? Seems science has surpassed the archaic way of just counting how many innings.
2. I'd love a better understanding of how to determine a reliever's effectiveness and value. Saves and holds don't do it, and honestly neither does ERA. Some combination of WPA, HLI, strikeout rate, strand rate, pitch count (to thus measure efficiency), WHIP, walk rate, etc. Surely there's a way to condense all of that into a single metric that can really grade out who the top guys are.
2. I talk about the challenges for relievers here:
https://joeposnanski.substack.com/p/subscriber-bonus-ask-about-a-stat/comment/13813368
1. IP is ubiquitous. But we track batters faced and we track Pitches Thrown. If someone wants to use those, they are readily available, at Savant, Fangraphs, Reference.
So why do teams still refer to an "innings limit" as a way of managing workloads for pitchers? Is it just because it's a simple enough (ubiquitous, as you say) way for everyone to understand the target?
Clubs refer to pitch limits, first and foremost. Then they might refer to times thru order.
The innings limit might be discussed at the seasonal view, and at that point, that's just a proxy for pitches thrown.