WAR, what is it good for? Absolutely everything.

[luckymann]

I'm constantly trying to improve my understanding of the more advanced stats. It's not easy, there are so many different calculation methodologies from the various sources. Which kind of makes it fun, as well as a tad infuriating.

My #1 and #2 pitchers in my fictional league franchise are - on the surface at least - having amazingly similar seasons (see screenshots). Each has 20 wins, their ERAs are 0.01 different, their strikeout totals are within 1 of each other, and they have pitched almost an identical number of innings. All of which, in and of itself, is quite uncanny.

But delve a bit deeper and a completely different story emerges, with the performance of one vastly superior to that of the other according to the more advanced stats. I have gone through and the only real difference between the two is that one has given up 22 HR to the other's 14, and that same pitcher's FIP is 50 points better.

This all came about because I couldn't for the life of me work out why one has 3.6 WAR to the other's 4.9. If you can be bothered, have a look at these screenshots and tell me your assessment.

[CBeisbol]

Yes
The top player has had much better results

I don't know why this would be odd

Imagine two hitters, one with 14 home runs and one with 22 but otherwise similar numbers. There wouldn't be much question about which hitter had been better.


Edit
I'm careful to say "had been better" or "better results" because WAR doesn't say anything about who is better. Only who had better results over a specific timeframe

[Questdog]

Originally Posted by CBeisbol

Yes
The top player has had much better results

I don't know why this would be odd

Imagine two hitters, one with 14 home runs and one with 22 but otherwise similar numbers. There wouldn't be much question about which hitter had been better.


Edit
I'm careful to say "had been better" or "better results" because WAR doesn't say anything about who is better. Only who had better results over a specific timeframe

In your zeal to cover all your bases, you have left some things open to the elements.... What if the 2 batters with similar stats, but one had 14 HR and the other had 22, had faced very different quality opponents? Say the 14 HR hitter had to face pitchers that all kept the all in the park well and the 22 HR hitter had faced pitchers who all were generous with their gopher balls? Who had the better results, then?

WAR is bad, because it gives people the impression that EVERYTHING is covered by its umbrella and it is the final word on the quality of players. There are so many assumptions and corrections and the like in WAR that if it IS a close approximation of the actual value of a player's season, then it would be a miracle of miracles.

Personally, I think the old stats are much better, because they are straightforward and their shortcomings are obvious. WAR is obscure and hypothetical and tries to adjust for everything when everything cannot be adjusted for.

[CBeisbol]

Originally Posted by Questdog

There are so many assumptions and corrections and the like in WAR that if it IS a close approximation of the actual value of a player's season, then it would be a miracle of miracles.

Here's your "miracle of miracles".
https://www.beyondtheboxscore.com/20...ird-order-wins

Are you likewise impressed by the voodoo that allows you to type messages and have your device respond to them?


Of course Questdog is convinced that it would be a "miracle" if WAR got close to actually measuring what it tries to measure, but has, apparently, never taken the time to actually look into it at all. Tells a lot about the type of person they are and how much credit one should give their opinion.


They are correct about one thing though. People should better understand what WAR (and all other stats) is, and isn't.

[Matt Arnold]

One large caveat on the above that is crucially important is to note how pitcher WAR is calculated. The players differ so much because the "WAR" stat reported is a FIP-based WAR, hence showing a huge difference because their FIP values differ so much.

If you look at the above, rWAR which is a runs-based WAR metric, shows the 2 very close to equal.

But that ties into the above comments - you need to understand what you're looking at. So just because WAR values differ by over a win, does that mean one guy was worth a win more than the other guy over the course of the season? No, not necessarily. But which one would I perhaps trust a little more? I'll take the less HR.

Again, because you can probably dig in a little more. You can see that the guy who gave up more HR had a lower BABIP (.260 vs .280), had more run support (5.9 vs 4.5), and if you look at the WPA column, was worth less (1.1 vs 1.9). So I'm guessing the guy who gave up those HR got a little luckier on balls in play, maybe got luckier in getting solo shots or stranding runners, and that's why the slight differences in FIP values end up resulting in near identical overall lines.

[Bobbyraz49]

I'm not a fan of WAR stats.
Wins Above Replacements....how can you guess how replacement player could/would perform ?

[Matt Arnold]

Originally Posted by Bobbyraz49

I'm not a fan of WAR stats.
Wins Above Replacements....how can you guess how replacement player could/would perform ?

Replacement level is essentially arbitrarily set: https://library.fangraphs.com/misc/war/

It's not meant to be an actual performance, but a generic "what should you be able to expect to get for free from someone getting called up from AAA/claimed from waivers/picked off a FA pile" type of player.

[CBeisbol]

Originally Posted by Bobbyraz49

I'm not a fan of WAR stats.
Wins Above Replacements....how can you guess how replacement player could/would perform ?

Well, you don't guess

You define it

[CBeisbol]

Originally Posted by Matt Arnold

Replacement level is essentially arbitrarily set: https://library.fangraphs.com/misc/war/

It's not meant to be an actual performance, but a generic "what should you be able to expect to get for free from someone getting called up from AAA/claimed from waivers/picked off a FA pile" type of player.

It's not "arbitrary". Not at all

Prior to this year, at any given time during the season, there were 750 players in the major leagues (plus players on the IL, waivers, etc). Then there where thousands more in the minors.

So, we're looking at the (upper) tail end of a bell curve when we look at all baseball players. The players closer to the (upper) tail (the Mike Trout's, Mookie Betts', etc) have the most value. With players having less value as you move away from the tail. At some point, where there are more players than MLB roster spots, is replacement level. At the tail end, one can't just say, if we can't get Trout, we'll get someone else who's just as good. Because there is no one just as good. As you move down the tail, there are other players who are just as good as each other. But there are few of them, they are a scarce resource. As you move farther down they become more plentiful-less scarce. Eventually, you reach a point where there are so many equally valuable players, that any individual player has no value. If you can't get that player, you can get another one who's just as good. You don't need to trade for these players because you have them in AAA or they have just been DFA'd by another team.

Anyone who plays OOTP knows who these players are and has a general idea where replacement level for their league is

[Questdog]

Originally Posted by CBeisbol

Here's your "miracle of miracles".
https://www.beyondtheboxscore.com/20...ird-order-wins

Are you likewise impressed by the voodoo that allows you to type messages and have your device respond to them?


Of course Questdog is convinced that it would be a "miracle" if WAR got close to actually measuring what it tries to measure, but has, apparently, never taken the time to actually look into it at all. Tells a lot about the type of person they are and how much credit one should give their opinion.


They are correct about one thing though. People should better understand what WAR (and all other stats) is, and isn't.

ANY conglomeration of statistical analyses can get a good correlation to team wins.... That does not mean that any of them are correct on an individual basis.

[Matt Arnold]

Originally Posted by CBeisbol

It's not "arbitrary". Not at all

Prior to this year, at any given time during the season, there were 750 players in the major leagues (plus players on the IL, waivers, etc). Then there where thousands more in the minors.

So, we're looking at the (upper) tail end of a bell curve when we look at all baseball players. The players closer to the (upper) tail (the Mike Trout's, Mookie Betts', etc) have the most value. With players having less value as you move away from the tail. At some point, where there are more players than MLB roster spots, is replacement level. At the tail end, one can't just say, if we can't get Trout, we'll get someone else who's just as good. Because there is no one just as good. As you move down the tail, there are other players who are just as good as each other. But there are few of them, they are a scarce resource. As you move farther down they become more plentiful-less scarce. Eventually, you reach a point where there are so many equally valuable players, that any individual player has no value. If you can't get that player, you can get another one who's just as good. You don't need to trade for these players because you have them in AAA or they have just been DFA'd by another team.

Anyone who plays OOTP knows who these players are and has a general idea where replacement level for their league is

I mean arbitrary in the sense that the actual replacement level value is in many ways picked more so that numbers would scale and look correct, rather than some deep-seeded calculation of exactly the 750 best players in the game at any point in time or anything like that. For example, see this article here which talks about when BBRef and Fangraphs unified that piece of their model: https://blogs.fangraphs.com/unifying-replacement-level/

Okay, it's not just numbers picked out of the blue, there is certainly some analysis going on to decide the levels and make sure the numbers line up with observations. But the actual value that the replacement level is set at is more of a fixed agreed-upon number rather than something which is perhaps more dynamic in its calculations.

[CBeisbol]

Originally Posted by Questdog

ANY conglomeration of statistical analyses can get a good correlation to team wins.... That does not mean that any of them are correct on an individual basis.

LOL
Nice word salad. The favorite dish of people who are out of their element

Put your money where your mouth is and provide a conglomeration of traditional stats that correlate with team wins as well as WAR does

Be the hero that ends WAR forever

[Questdog]

Originally Posted by CBeisbol

LOL
Nice word salad. The favorite dish of people who are out of their element

Put your money where your mouth is and provide a conglomeration of traditional stats that correlate with team wins as well as WAR does

Be the hero that ends WAR forever

Runs Created + ERA.....
OBP + OBP allowed....
Runs scored plus runs allowed...

I have no desire to end WAR; just for people to realize its limitations and not use it for definitive statements....

[BIG17EASY]

I feel like I'm reading a less-nuanced, unedited version of Keith Law's Smart Baseball. Anyone who doesn't think WAR is a better evaluation of a player than the old "baseball card stats," they need to read that book. While WAR has its flaws, it's one of the better all-encompassing stats available today.

[CBeisbol]

Originally Posted by Questdog

Runs Created + ERA.....
OBP + OBP allowed....
Runs scored plus runs allowed...

Show your work

You don't have any work, of course, because you haven't done it

Doesn't appear you understand it well enough to do it

None of the measures you listed work on the player level
What was Mike Trout's Runs created plus ERA for example?
Or runs scored plus runs allowed?
Or OBP and OBP allowed?

I have no desire to end WAR; just for people to realize its limitations and not use it for definitive statements....

Or any desire to understand it, apparently.

So you're definitive statement about WAR being a miracle wasn't worth the non-existent paper it was printed on

[Questdog]

Originally Posted by CBeisbol

Show your work

You don't have any work, of course, because you haven't done it

Doesn't appear you understand it well enough to do it

None of the measures you listed work on the player level
What was Mike Trout's Runs created plus ERA for example?
Or runs scored plus runs allowed?
Or OBP and OBP allowed?


Or any desire to understand it, apparently.

So you're definitive statement about WAR being a miracle wasn't worth the non-existent paper it was printed on

I have studied baseball and its statistics since before you were born (probably). For you to claim that I do not understand something is 1) arrogant on your part and 2) nonsensical, because you know nothing of my knowledge.

For a team's WAR to correlate to a team's wins, both pitchers and hitters WAR must be used.
To say that any other system must correlate to team wins without using both hitter and pitching stats is absurd.

I think the general regard you have acquired on the boards for yourself is a very good indicator of the insight you have provided to the community....

[mytreds]

WAR is a stat made by a human from other stats. The stats used by the human's WAR formula is heavily subjective to what the human believes makes up a "good" ball player. While it can be useful, to abscribe WAR to the level of gospel and "cannot ever be argued against" is pretty absurd. It is a stat like any other, and is not more important or better than any other. The difference, however, is whereas a stat, such as batting average is a true objective stat, WAR is a purely subjective stat.

WAR is the exterior of car. Gives you the general idea of what a player looks like and how he should perform. Open the hood to find what is really going on with more individual stats.

[dkgo]

Even that's not true because human subjectivity decides what is a hit or an error.

WAR is not perfect and no one who knows what they are talking about will claim that it is. The defensive component especially has lots of room for improvement. But if you are looking for one number to measure how good a player has been, WAR is one of the best that we have and is obviously better than a stat like batting average.

[CBeisbol]

Originally Posted by Questdog

I have studied baseball and its statistics since before you were born (probably). For you to claim that I do not understand something is 1) arrogant on your part and 2) nonsensical, because you know nothing of my knowledge.

Ahh. Dunning and Kruger would be proud.

That you are old is absolutely irrelevant to your understanding of WAR

You have absolutely demonstrated that you do not understand even the basics about how WAR works

And you've twisted yourself into knots trying to hide from the untenable situation you've put yourself into.

You first claimed that it would be "the miracle of miracles" if WAR came close to measuring player value. Then, when shown indisputable proof that it does you changed your argument to "ANY conglomeration" of stats can be shown to correlate to team wins. Of course, you failed to provide any evidence that this is true.

For a team's WAR to correlate to a team's wins, both pitchers and hitters WAR must be used.
To say that any other system must correlate to team wins without using both hitter and pitching stats is absurd.

Not sure if this is your non-understanding of this or you're understanding that you're not making any sense and trying to weasel out of it

I again, invite you to show how the stats you listed, at the player level, can be aggregated to match team wins as closely as WAR does

Show how Mike Trout's runs created and, uhhh, Andrew Heaney's ERA can tell us how much each player has contributed to the Angels team wins in 2020

We know you won't
First, because they don't
And second, because you wouldn't know how even if they did

I think the general regard you have acquired on the boards for yourself is a very good indicator of the insight you have provided to the community....

My popularity is as relevant as your age to this discussion
That you bring it up, again, shows you don't know enough about the discussion to know that you don't know enough about the discussion.

[CBeisbol]

Originally Posted by mytreds

WAR is a stat made by a human from other stats. The stats used by the human's WAR formula is heavily subjective to what the human believes makes up a "good" ball player.

Not true
It objectively measures what leads to runs and wins

While it can be useful

It is extremely useful
https://www.beyondtheboxscore.com/20...ird-order-wins

to abscribe WAR to the level of gospel and "cannot ever be argued against" is pretty absurd.

Agreed
And the only people I see who ever make any mention of this are people trying to argue against WAR.

t is a stat like any other, and is not more important or better than any other. The difference, however, is whereas a stat, such as batting average is a true objective stat, WAR is a purely subjective stat.

Again, this is not close to being true
Anyone who calculates WAR using the same formula will, assuming they make no mistakes, arrive at the same answer. Just as anyone calculating batting average will.

Gives you the general idea of what a player looks like and how he should perform.

Nope. Entirely incorrect. WAR says absolutely nothing about how a player should perform
It only tells you how they did perform