Get Lost, Bill James!

I hate sabermetrics.  I’ll just get that out in the open.  I hate ‘em.  Not as bad as the Yankees, but I still hate them.  I find all of them awful, but in particular, WAR is just the worst.  It is vague, meaningless, misunderstood, and misinterpreted.  In short, it’s just like the Dickey trade – people don’t get what it means, but they want to look like they do, so they start complaining about something or other.

Here’s why I hate WAR.  Number 1, its name does an absolutely terrible job of expressing what it means.  If a statistic is called “Wins Above Replacement,” I would expect a few things out of that statistic.  1, defensive WAR plus offensive WAR would add to give total WAR, and 2, the WAR of every player on a team would give the win total of the entire team.  I have addressed numerous people about the second concern, and their response is always the same: “Well, you’ll just have to get used to it, because there’s no team WAR.”  Am I missing something?  Isn’t a team the sum of its players?  Shouldn’t the wins contributed to a team by each player add up to the amount of total team wins?

On the subject of the second concern, I don’t have any kind of answer yet, but to me, this is another problem with WAR.  The only possible explanation for the fact that offensive and defensive WAR don’t add to give total WAR is that one is weighed more heavily than the other.  However, this presents problems.  As even sabermetricians have acknowledged, 99% of major league baseball players are at least decent in the field (barring the Adam Dunn type players).  The far majority of defensive plays will be made by the fielder, regardless of whether that fielder is elite or average.  Therefore, theoretically, defense should be weighed more heavily for some players than others, but the only way to calculate how much to weigh defense would be to use that players defensive WAR, and we’ve gotten ourselves into a circle without an end.  WAR is using the same marketing tactic that Toyota uses, when it says, “90% of Toyotas sold in the last 15 years are still on the road today.”  That seems impressive at the beginning, but if you examine it, it doesn’t mean that 90% of Toyotas sold 15 years ago are still on the road.  That statement covers cars sold anytime during the last 15 years, so out of those 90% of cars still on the road, half of them have been sold in the last 7.5 years, and should definitely still be on the road.  WAR is the same – it takes into account plays made by a defensive player, but doesn’t realize that on the majority of defensive plays, any defensive player would suffice.  For example, if you compare two shortstops and find that one fielded 95% of ground balls hit his way, and one fielded 97% of balls hit his way, it doesn’t seem like a big difference.  However, what WAR fails to tell you is that any average fielder could probably have fielded about 91% of those balls.  So in reality, the difference between the fielders should be 6% to 4%, which seems like a much bigger difference.

Another great problem with WAR is the replacement player that it speaks of.  WAR doesn’t seem to realize how rare it is that players compile a stat line like Ike Davis did in 2012.  Davis hit only .227, but hit 32 home runs with 93 RBIs.  His WAR was 0.7.  In other words, if the Mets replaced Davis with a replacement level player, their first baseman would contribute 0.7 wins less than Davis did.  Ridiculous.  Ike provided more wins than that in one game when he hit 3 home runs.  For doing that, he should get credit for a full win.  A replacement player, who would probably bat about .230 with 10 home runs, doesn’t have the tools that Ike does, like raw power and pitcher intimidation factor.

WAR constantly overrates the replacement player.  In a classic example, Joe Saunders’ 2009 season, during which he went 16-7 with a 4.60 ERA, earned him a WAR of 0.3.  What the WAR people don’t realize that even pitching 186 innings in a season is something that a replacement player is probably not going to be able to do, and doing that alone should be worth more than 0.3 wins.  Saunders pitched 186 innings and won 16 games, and we’re supposed to believe that a pitcher the level of Chris Schwinden was worth just 0.3 wins less to his team?  Ridiculous!  Also, with pitchers, shouldn’t a complete game shutout go directly to the WAR total?  If you want to give a little to the defense and a little to the offense, that’s fine, but if you pitch a complete game, that’s one win gained directly by a pitcher.  The same could be said of a four home run game, or a game in which a hitter gets 6 hits.  Regardless of whether the rest of the team had been performing at replacement level, that one player directly supplied the win, so shouldn’t they get credit for it in the form of WAR?  WAR doesn’t recognize that going 16-7 isn’t something that one can just do, regardless of how much your defense or offense helped.  To go 16-7, you need a quality that a replacement player does not have, which lets you pitch deep into games, get the right hitters out, and not allow runs when the game is on the line.  A replacement player couldn’t just come up and go 16-7.  And WAR needs to recognize that.

Another example of the absurdity of WAR can be seen by comparing different seasons for hitters. For example, as stated above, Ike Davis had a WAR of 0.7 in 2012.  Carlos Lee also had a WAR of 0.7 in 2005.  In both of those seasons, the player hit 32 home runs, but in 2012, Davis batted .227, while Lee batted .265.  Lee, with no disrespect to Ike, who was battling Valley Fever, should not be grouped with a hitter who hit .227 while he hit .265.  If you want to pin the difference on defense, you can try, but it just exposes another great fault of WAR: weighting components for different players.  Carlos Lee had a oWAR of 1.6 in 2005, which in itself is absurd, because 32 home runs and a .265 average is worth more than 1.6 wins, and he had a dWAR of -1.6.  Davis had an oWAR of 0.9, and a dWAR of -1.1, which is also absurd, since his fielding percentage was 13th – that’s better than half – among MLB first basemen.  If the two WARs are combined for each player, Davis ends up with -0.2, and Lee with 0.0.  However, this was not the case.  In Lee’s case, we can deduce that his dWAR was valued as roughly half as important as his oWAR, since when you subtract the dWAR, about half of the original oWAR is left.  However, with Davis, the dWAR was valued as roughly 1/6 as important as his oWAR.  Why?  Lee played about 180 more innings in the field than Davis did, but that still doesn’t account for a difference that big.  Therefore, the only explanation that I can think of is that the WAR formula noticed something unimportant in the stats – perhaps that Lee played multiple positions, none of them well, or perhaps that first basemen are thought to play the easiest position in the field – and turned it into a gross misrepresentation of how valuable a player is to his team.  WAR has to realize a few things – 1, that replacement players aren’t as good as they think, and 2, that defense, as explained, is mostly unimportant.  Until the sabermetricians learn that, WAR will continue to be unimportant.

Update: Read the author’s response piece here

11 Comments

Summary: “In my completely subjective opinion, these players are worth more wins than WAR says they are, therefore WAR is worthless.(because in my world home runs and batting average are the only measures of a player’s offensive value).”

No. What I’m saying is that WAR doesn’t include any intangibles like pitching around a hitter or having to change pitchers because of a hitter. How many games could the Mets have won last year because the opposing team used their lefty reliever to face Davis, then couldn’t use him at another critical juncture of the game?

Then make that clear. You are only talking about batting average, wins, home runs, and errors. The whole reason sabermetrics was created is these basic stats weren’t good enough.

Those aren’t intangibles–one could measure the number of times a lefty reliever came in to face Ike Davis and how much it (likely) cost them later in the game. I’d be willing to wager the total is around a win or less.

As for pitching around a hitter, the hitter is more likely to take a walk, which is valuable and is factored into WAR.

WAR is far from a perfect measure of a player’s value, but it’s still a useful approximation of how many wins a player provides his team.

The Mets lost the game in which Ike Davis hit three home runs.

But that’s because Chris Young, on that day, probably pitched no better than a replacement player, and no one who hit before Ike could get on base. If everyone performs at the level they’re supposed to, Ike’s performance wins that game.

Because I’m bored, snowed in, and have been talking with others about WAR today, a few answers to your questions:

“If a statistic is called “Wins Above Replacement,” I would expect a few things out of that statistic. 1, defensive WAR plus offensive WAR would add to give total WAR, and 2, the WAR of every player on a team would give the win total of the entire team.”

1) Depends what WAR you’re looking at, but it probably doesn’t because you have to factor in position. A shortstop contributing 5 def+off WAR is more valuable than a first baseman with the same production. Or if position is already factored in, then you can’t add them because you’d be double-counting the position adjustment.
2) As you note, it’s called Wins Above Replacement. A team made entirely of replacement players would still win some games. You add up the team’s WAR and you get the marginal wins they’d provide over a team of minimally-skilled players.

“The only possible explanation for the fact that offensive and defensive WAR don’t add to give total WAR is that one is weighed more heavily than the other… For example, if you compare two shortstops and find that one fielded 95% of ground balls hit his way, and one fielded 97% of balls hit his way, it doesn’t seem like a big difference. However, what WAR fails to tell you is that any average fielder could probably have fielded about 91% of those balls.”

Of course it tells you that. That’s the whole idea behind the “Above Replacement” part of Wins Above Replacement.

“Ike provided more wins than that in one game when he hit 3 home runs. For doing that, he should get credit for a full win.”

Okay, sure, start him at one win, and then start deducting points for when he goes 0-for-4 and costs his team wins.

“What the WAR people don’t realize that even pitching 186 innings in a season is something that a replacement player is probably not going to be able to do, and doing that alone should be worth more than 0.3 wins.”

Of course a replacement player could. If the team runs out of money and can only spend the minimum for one of their rotation spots, a replacement level player will throw those innings (or a combination of multiple replacement level players as the team gets tired of one guy’s terrible performance and swaps him for another). Those innings don’t disappear if Joe Saunders doesn’t throw them. Someone’s gotta throw them. The mere act of throwing those innings cannot possibly be worth anything.

“Another example of the absurdity of WAR can be seen by comparing different seasons for hitters. For example, as stated above, Ike Davis had a WAR of 0.7 in 2012. Carlos Lee also had a WAR of 0.7 in 2005. In both of those seasons, the player hit 32 home runs, but in 2012, Davis batted .227, while Lee batted .265. ”

1) It’s about more than batting average, and 2) You can’t just compare raw stats. The league’s run environment can change year-to-year. Context is important.

Because I’m bored, snowed in, and have been talking with others about WAR today, a few answers to your questions:

“If a statistic is called “Wins Above Replacement,” I would expect a few things out of that statistic. 1, defensive WAR plus offensive WAR would add to give total WAR, and 2, the WAR of every player on a team would give the win total of the entire team.”

1) Depends what WAR you’re looking at, but it probably doesn’t because you have to factor in position. A shortstop contributing 5 def+off WAR is more valuable than a first baseman with the same production. Or if position is already factored in, then you can’t add them because you’d be double-counting the position adjustment.
2) As you note, it’s called Wins Above Replacement. A team made entirely of replacement players would still win some games. You add up the team’s WAR and you get the marginal wins they’d provide over a team of minimally-skilled players.

“The only possible explanation for the fact that offensive and defensive WAR don’t add to give total WAR is that one is weighed more heavily than the other… For example, if you compare two shortstops and find that one fielded 95% of ground balls hit his way, and one fielded 97% of balls hit his way, it doesn’t seem like a big difference. However, what WAR fails to tell you is that any average fielder could probably have fielded about 91% of those balls.”

Of course it tells you that. That’s the whole idea behind the “Above Replacement” part of Wins Above Replacement.

“Ike provided more wins than that in one game when he hit 3 home runs. For doing that, he should get credit for a full win.”

Okay, sure, start him at one win, and then start deducting points for when he goes 0-for-4 and costs his team wins.

“What the WAR people don’t realize that even pitching 186 innings in a season is something that a replacement player is probably not going to be able to do, and doing that alone should be worth more than 0.3 wins.”

Of course a replacement player could. If the team runs out of money and can only spend the minimum for one of their rotation spots, a replacement level player will throw those innings (or a combination of multiple replacement level players as the team gets tired of one guy’s terrible performance and swaps him for another). Those innings don’t disappear if Joe Saunders doesn’t throw them. Someone’s gotta throw them. The mere act of throwing those innings cannot possibly be worth anything.

“Another example of the absurdity of WAR can be seen by comparing different seasons for hitters. For example, as stated above, Ike Davis had a WAR of 0.7 in 2012. Carlos Lee also had a WAR of 0.7 in 2005. In both of those seasons, the player hit 32 home runs, but in 2012, Davis batted .227, while Lee batted .265. ”

1) It’s about more than batting average, and 2) You can’t just compare raw stats. The league’s run environment can change year-to-year. Context is important.

Just want to address a few things here. You are a bit off the mark in my opinion.

“Here’s why I hate WAR. Number 1, its name does an absolutely terrible job of expressing what it means. If a statistic is called “Wins Above Replacement,” I would expect a few things out of that statistic. 1, defensive WAR plus offensive WAR would add to give total WAR, and 2, the WAR of every player on a team would give the win total of the entire team. I have addressed numerous people about the second concern, and their response is always the same: “Well, you’ll just have to get used to it, because there’s no team WAR.” Am I missing something? Isn’t a team the sum of its players? Shouldn’t the wins contributed to a team by each player add up to the amount of total team wins?”
1. Offense is more valuable than defense. They can’t be weighted the same.
2. It’s called “wins above replacement” because it gives you the difference between the value of the particular player, and the average Triple-A replacement player.
3. A team of 25 replacement players would NOT go 0-165. They would win somewhere around 40-50 games. That’s why all the numbers don’t add up to a team’s wins because it adds up to how many more wins the team would than a team of 25 replacement players.

“Saunders pitched 186 innings and won 16 games, and we’re supposed to believe that a pitcher the level of Chris Schwinden was worth just 0.3 wins less to his team? ”
Yes, Saunders won 16 games but had a not-so-hot ERA of 4.60. That is one of the main reasons why his WAR is so low. His wins are almost completely meaningless.

” To go 16-7, you need a quality that a replacement player does not have, which lets you pitch deep into games, get the right hitters out, and not allow runs when the game is on the line. A replacement player couldn’t just come up and go 16-7. And WAR needs to recognize that.”
Again… wins?

“Davis had an oWAR of 0.9, and a dWAR of -1.1, which is also absurd, since his fielding percentage was 13th – that’s better than half – among MLB first basemen.”
Errors?

“If the two WARs are combined for each player, Davis ends up with -0.2, and Lee with 0.0. However, this was not the case. In Lee’s case, we can deduce that his dWAR was valued as roughly half as important as his oWAR, since when you subtract the dWAR, about half of the original oWAR is left. However, with Davis, the dWAR was valued as roughly 1/6 as important as his oWAR. Why? Lee played about 180 more innings in the field than Davis did, but that still doesn’t account for a difference that big. ”
Positional adjustments. Defense is more valuable at particular positions.

Overall, I think you are off. And using errors and wins, two things that even many fans that don’t like sabermetrics disregard now. In 2006, Randy Johnson went 17-11 with a 5.00 ERA. Obviously he had a great career before that but do you think he was good that year? Because he won 17 games? I think you’re looking at this the wrong way. I’m not convinced.

I only made it so far into this before I had to stop. Here’s what I don’t get:

Before I would decide to go on a public rampage about something, I would do some basic fact-finding of the subject I’m about to rant about. All you have to do is roll your mouse over either stat at Baseball-Reference and it TELLS YOU WHY oWAR + dWAR does not = WAR.

Holy crap, dude.

There is almost nothing in this piece that is accurate.

– You have an entire paragraph about offense and never once mentioned OBP.
– You mentioned a pitcher’s win total four (FOUR!) times in once paragraph. A pitcher’s win total is useless.
– Fielding percentage? AHAHAHAHAHAHAHAHAHAHAHA

I can’t believe I read most of that.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

%d bloggers like this: