WAR is the dumbest stat known to mankind (Trout vs. Betts)

[moonslav59]

fWAR as of today:

 

8.2 Ramirez (SS defense brings more value than RF)

7.9 Betts (Awesome number)

7.6 Trout (missed time has made up for Mookie's missed games)

6.7 Lindor

5.4 Chapman

5.1 JD (The -11.7 for defense kills his WAR.)

 

Beni is 14th at 4.3

 

Bogey is 25th at 3.8

 

Nice to see 4 in the top 25.

 

 

[kenmeister]

An ESPN article on how much JD Martinez improved the Red Sox: http://www.espn.com/mlb/story/_/id/24380539/where-red-sox-jd-martinez

[Hugh2]

Name one stat better at evaluating a player.

[harmony]

fWAR as of today:

 

8.2 Ramirez (SS defense brings more value than RF)

7.9 Betts (Awesome number)

7.6 Trout (missed time has made up for Mookie's missed games)

6.7 Lindor

5.4 Chapman

5.1 JD (The -11.7 for defense kills his WAR.)

 

Beni is 14th at 4.3

 

Bogey is 25th at 3.8

 

Nice to see 4 in the top 25.

This year Jose Ramirez has played in the field only at third base, which has an fWAR position adjustment of +2.5 runs (not the +7.5 runs at shortstop but more than the -7.5 runs in right field):

 

https://www.fangraphs.com/blogs/explaining-win-values-part-three/

[Hugh2]

WAR is Obviously a hot topic that always drums up a good conversation with conflicting views. Personally, I've always liked to use rWar for position players and Fwar for pitchers.

[cp176]

In the case of batter 'protection', there is strong statistical evidence that batters do not hit any better when they have a strong hitter behind them supposedly protecting them. There is evidence that pitchers may pitch differently to the 'protected' batters, but the batters do not reap any benefits from that.

 

People can talk about line up protection all day, but the numbers just don't bear that out.

 

You are probably right about the numbers. I certainly would never question the numbers. I am much more interested in what the players say and think with respect to having JD in the lineup as opposed to what the numbers say.

[S5Dewey]

Name one stat better at evaluating a player.

 

WAR tells us what we already know so we tend to believe it. It tells us that JDM is a better player than, say, Christian Vazquez and that Mookie is better than Brock Holt.

 

However, we as consumers tend (want?) to believe that any value expressed as an integer is concise and WAR doesn't work that way. Fangraphs in it explanation of how WAR works says that "Given the imperfections of some of the available data and the assumptions made to calculate other components, WAR works best as an approximation. A 6 WAR player might be worth between 5.0 and 7.0 WAR..." and then goes on to say that player in that group is probably an All-Star, etc.

 

That's a 17% margin for error and it means that a player with a WAR of 3.0 may actually be worth somewhere between 2.5 & 3.5. Yet when we see a player with a WAR of 3.5 we assume that he's 1/4 again as good as that 2.5 player when actually they may be the same player once the variables are accounted for. What we do know from WAR is that Mookie is better than a player with a WAR of 3.0 - but we already knew that from watching them.

 

Yes, WAR is valuable in confirming what we already know but when we're comparing two players whose WAR are within a couple of points WAR is essentially meaningless.

[mvp 78]

WAR tells us what we already know so we tend to believe it. It tells us that JDM is a better player than, say, Christian Vazquez and that Mookie is better than Brock Holt.

 

However, we as consumers tend (want?) to believe that any value expressed as an integer is concise and WAR doesn't work that way. Fangraphs in it explanation of how WAR works says that "Given the imperfections of some of the available data and the assumptions made to calculate other components, WAR works best as an approximation. A 6 WAR player might be worth between 5.0 and 7.0 WAR..." and then goes on to say that player in that group is probably an All-Star, etc.

 

That's a 17% margin for error and it means that a player with a WAR of 3.0 may actually be worth somewhere between 2.5 & 3.5. Yet when we see a player with a WAR of 3.5 we assume that he's 1/4 again as good as that 2.5 player when actually they may be the same player once the variables are accounted for. What we do know from WAR is that Mookie is better than a player with a WAR of 3.0 - but we already knew that from watching them.

 

Yes, WAR is valuable in confirming what we already know but when we're comparing two players whose WAR are within a couple of points WAR is essentially meaningless.

 

You didn't answer his question tho.

[S5Dewey]

You didn't answer his question tho.

 

I did. You just didn't get it.

 

While there may be no "better" statistic to evaluate a player, WAR doesn't work either - at least beyond telling us what we already can see. It works fine to tell us what we already know (duh!) but it's not accurate enough to be used in comparing several players who are within 17% of ability.

[mvp 78]

I did. You just didn't get it.

 

While there may be no "better" statistic to evaluate a player, WAR doesn't work either - at least beyond telling us what we already can see. It works fine to tell us what we already know (duh!) but it's not accurate enough to be used in comparing several players who are within 17% of ability.

 

You didn't answer his question until you said the above bold words. You just went on a rant about WAR. You didn't say there wasn't a better one.

[S5Dewey]

You didn't answer his question until you said the above bold words. You just went on a rant about WAR. You didn't say there wasn't a better one.

 

I'm sorry. I thought anyone with a modicum of intelligence would infer that from my first post. I was apparently wrong.

[S5Dewey]

You didn't answer his question until you said the above bold words. You just went on a rant about WAR. You didn't say there wasn't a better one.

 

And BTW, it wasn't a "rant". It was a concise explanation of the fact that WAR has built in flaws that give it a 17% margin of error.

[Bellhorn04]

While there may be no "better" statistic to evaluate a player, WAR doesn't work either - at least beyond telling us what we already can see. It works fine to tell us what we already know (duh!) but it's not accurate enough to be used in comparing several players who are within 17% of ability.

 

What exactly do you mean by telling us what we already know?

[mvp 78]

I'm sorry. I thought anyone with a modicum of intelligence would infer that from my first post. I was apparently wrong.

 

Next time try to write better?

 

Not sure how anyone could infer that from your rambling.

[Hugh2]

WAR tells us what we already know so we tend to believe it. It tells us that JDM is a better player than, say, Christian Vazquez and that Mookie is better than Brock Holt.

 

However, we as consumers tend (want?) to believe that any value expressed as an integer is concise and WAR doesn't work that way. Fangraphs in it explanation of how WAR works says that "Given the imperfections of some of the available data and the assumptions made to calculate other components, WAR works best as an approximation. A 6 WAR player might be worth between 5.0 and 7.0 WAR..." and then goes on to say that player in that group is probably an All-Star, etc.

 

That's a 17% margin for error and it means that a player with a WAR of 3.0 may actually be worth somewhere between 2.5 & 3.5. Yet when we see a player with a WAR of 3.5 we assume that he's 1/4 again as good as that 2.5 player when actually they may be the same player once the variables are accounted for. What we do know from WAR is that Mookie is better than a player with a WAR of 3.0 - but we already knew that from watching them.

 

Yes, WAR is valuable in confirming what we already know but when we're comparing two players whose WAR are within a couple of points WAR is essentially meaningless.

 

So.......you can't name a stat at better evaluating a player?

[Kimmi]

An ESPN article on how much JD Martinez improved the Red Sox: http://www.espn.com/mlb/story/_/id/24380539/where-red-sox-jd-martinez

 

This paragraph is almost like the writer is channeling his inner Kimmi:

 

Of course, that ignores the trickle-down effect of adding Martinez. Maybe he provides lineup protection. Maybe he makes the other hitters better merely by his presence. Those are speculative ideas, however. One of the earliest and longest-running sabermetrics issues is the question of lineup protection, but comprehensive study after comprehensive study has failed to prove that such a thing exists. The Red Sox are better at the plate because Martinez is better than the players they had last year at his positions and because Betts is better than Betts.

[Kimmi]

Name one stat better at evaluating a player.

 

Can't be done!

[Kimmi]

I did. You just didn't get it.

 

While there may be no "better" statistic to evaluate a player, WAR doesn't work either - at least beyond telling us what we already can see. It works fine to tell us what we already know (duh!) but it's not accurate enough to be used in comparing several players who are within 17% of ability.

 

But WAR does work. What do you mean it doesn't work? The margin of error?

 

Yes, it's flawed. So are strikeouts, as I've mentioned many, times. Being flawed does not mean that it doesn't work.

[Kimmi]

S5, what stat isn't flawed?

 

And why are you willing to accept some flawed stats but not others?

[Hugh2]

S5, what stat isn't flawed?

 

And why are you willing to accept some flawed stats but not others?

 

I don't think anyone has ever said WAR was flawed, the point is there isn't a better singular stat.

[Kimmi]

I don't think anyone has ever said WAR was flawed, the point is there isn't a better singular stat.

 

No one has ever said that WAR was flawed?

 

But to your point, I agree completely.

[Hugh2]

Wasn't*

[Kimmi]

Wasn't*

 

Ah. I should have known.

[S5Dewey]

What exactly do you mean by telling us what we already know?

 

We already know that, for example, Jose Ramirez, Mike Trout, and Mookie Betts are some of the best players in MLB this year.

 

What WAR doesn't tell us is which of them is the better player because they all fall within that 17% margin of error.

[S5Dewey]

So.......you can't name a stat at better evaluating a player?

 

Given the number of data points we have (all the players in MLB for how many years?) it's hard to say that any stat with that many data points and a 17% margin of error is a "good stat". I'll give you this though, it's the best bad stat currently in use.

[moonslav59]

Given the number of data points we have (all the players in MLB for how many years?) it's hard to say that any stat with that many data points and a 17% margin of error is a "good stat". I'll give you this though, it's the best bad stat currently in use.

 

No stats came close to capturing everything, so yes, all stats were and still are bad, in the sense that none are very close to being perfect.

 

WAR is an attempt to do better than what we had before.

 

It's not perfect. It's not close to perfect, but in my opinion, it's closer to perfect than it is to the next closest single stat- for what that is worth.

[vegasbob]

You would think that the proof of WAR and its variations for position players and pitchers applied over a 10 year period, correlated for results in the season standings would provide some consistency of results. Better WAR cumulatively, better standings results. As many have mentioned the playoffs , whether 1 game, 5 games or 7 game series are subject to numerous variances and almost by definitions are small samples. But there still would be some correlation even if with wider variances to full seasons. I am not the stat geek, but what do you who know the numbers say to that?

[Kimmi]

You would think that the proof of WAR and its variations for position players and pitchers applied over a 10 year period, correlated for results in the season standings would provide some consistency of results. Better WAR cumulatively, better standings results. As many have mentioned the playoffs , whether 1 game, 5 games or 7 game series are subject to numerous variances and almost by definitions are small samples. But there still would be some correlation even if with wider variances to full seasons. I am not the stat geek, but what do you who know the numbers say to that?

 

The numbers say that WAR correlates to actual team wins with an r^2 of roughly .87, depending on which version of WAR you use and which sample of data you look at. That's pretty darn good. The correlation should only get stronger as improvements in technology and tweaks to WAR itself make the stat even better.

 

Here's the thing about stat geeks. They are not going to find a stat (or anything) that works and then rest on their laurels. They understand and acknowledge the shortcomings, and they will continue to find ways to make improvements.

[Bellhorn04]

We already know that, for example, Jose Ramirez, Mike Trout, and Mookie Betts are some of the best players in MLB this year.

 

What WAR doesn't tell us is which of them is the better player because they all fall within that 17% margin of error.

 

You're talking about players that everyone knows are having great seasons.

 

But I also look at the WAR for players that I haven't seen play and don't really know much about.

[S5Dewey]

You're talking about players that everyone knows are having great seasons.

 

But I also look at the WAR for players that I haven't seen play and don't really know much about.

 

Yes, my examples talked about players who are having great seasons but the same principle is true for players who are having mediocre or even bad seasons

 

I'm not at all sure where the WAR line is between good, mediocre, and bad players are but I think it's safe to say say that there are more mediocre players than good or "bad" ones. WAR doesn't define which of those mediocre players are better or worse than others in that group because due to that 17% thing they're all statistically the same. Yet we've become so dependent on WAR that we tend to think of a player with a WAR of 3.5 to be a better player than someone with a WAR of 2.5 when, even according to Fangraphs, there's no evidence to support that.

 

Look, I didn't title this thread and that title doesn't reflect my beliefs. I think WAR is useful. It's just not as useful as most people think it is.