Let's talk about Adam Dunn destinations

[iamshack]

Except "eyes" don't always agree. Adam Dunn is the perfect example. Some people will just never admit that a hitter that strike out 25-30% of the time can be good no matter what else he does. Advanced stats allow us to have a more empirical way to say that Dunn is, in fact, a good hitter. I can hold whatever position I'd like if all of our eyes are acceptable as measurement. Since people don't like arguing over eyes, we have numbers. And when we have numbers, the best numbers are better. Batting average says Dunn sucks. OBP sometimes suggests that Dunn sucks. Yet, he does not suck.

 

We didn't need advanced stats to see that Miguel Cabrera and Mike Trout were great, but WAR told us that Trout was more valuable by quite a measure last year. It also tells us that Miggy is more valuable this year, despite being an absolutely atrocious defender.

You're saying that people's eyes said Adam Dunn wasn't a valuable player before?

 

That's nonsense.

[Chicago White Sox]

Yeah, sequencing is an interesting matter. To my knowledge, most attempts at judging the value of consistency vs. streakiness have been inconclusive, though I do specifically remember someone creating a stat called BACON (Batting Consistency) that was supposed to measure who had their hits distributed most evenly. If I recall correctly, Ichiro was the champion there.

 

Ahh, yes: http://www.plunkeveryone.com/?p=95

I would think scoring runs in a consistent manner would be more valuable, on average, than having huge deviations in run output. If true, then a player like Dunn, who has a big impact on a small number of game (when he hits a HR) and a small impact on most others, would be less valuable overall then the sum of the inputs that go into his wOBA & wRC+ stats. Maybe I'm way off base, but that's the main reason I don't like Dunn.

[Dick Allen]

Except "eyes" don't always agree. Adam Dunn is the perfect example. Some people will just never admit that a hitter that strike out 25-30% of the time can be good no matter what else he does. Advanced stats allow us to have a more empirical way to say that Dunn is, in fact, a good hitter. I can hold whatever position I'd like if all of our eyes are acceptable as measurement. Since people don't like arguing over eyes, we have numbers. And when we have numbers, the best numbers are better. Batting average says Dunn sucks. OBP sometimes suggests that Dunn sucks. Yet, he does not suck.

 

We didn't need advanced stats to see that Miguel Cabrera and Mike Trout were great, but WAR told us that Trout was more valuable by quite a measure last year. It also tells us that Miggy is more valuable this year, despite being an absolutely atrocious defender.

Has anyone ever added up all the teams players WAR and posted standings based on those? It would be interesting to see the difference between WAR standings and actual standings.

[Jake]

I remember reading a Baseball Prospectus article that was basically a walk-through of a potential manager's statistic and when they finally posted who the best managers would be with this stat, they had to juggle some things. After feeling satisfied that the stat had a logical method on paper, it also had to pass the eye test. If a stat TOTALLY turned common knowledge on its head, chances are the stat is wrong. On the other hand, if it simply produces results that are predictable with other statistics or observations, it is likewise not very useful.

 

I think things like WAR, wRC+. UBR, UZR satisfy both things. It doesn't say that players we know are s***ty are good, but it also doesn't simply reproduce things we already know. It will sometimes be unsurprising but at other times it may be surprising.

Edited July 17, 2013 by Jake

[Jake]

Has anyone ever added up all the teams players WAR and posted standings based on those? It would be interesting to see the difference between WAR standings and actual standings.

 

Current WAR totals by team

 

More details, such as projected rest of season/end of season records based on sabermetrics

[Jake]

Current WAR totals by team

 

More details, such as projected rest of season/end of season records based on sabermetrics

 

Some highlights related to White Sox - it thinks we suck, as we do. Predicts a #5 draft pick at end of season. Predicts Twins will eventually finish last in AL Central, but we will not catch Royals for third. Thinks we'll play just under .500 ball in second half.

 

This would adjust if players were traded.

[Dick Allen]

Current WAR totals by team

 

More details, such as projected rest of season/end of season records based on sabermetrics

So the Sox team WAR is the same as Cleveland's. the Angels team WAR is ome of the best in baseball. There is reason to doubt WAR as a useful, accurate tool. I like to use it, but ii really is not consistent with the standings, and I know nothing is going to be perfect, but it is significantly off in many cases.

[hammerhead johnson]

The biggest work in progress IMO is evaluating catcher defense, though there are some really promising stats coming out in that regard too. The only issue is that they aren't being updated at the rate that other defensive statistics are (I don't believe any numbers for the most sophisticated catcher defense statistics are available yet for this season).

 

Right. I don't even pay attention to catcher WAR. Ron Santo has a higher WAR (71) than any catcher in baseball history with the exception of Johnny Bench.

 

Top 5

1. Johnny Bench 75

2. Pudge Rodriguez 70

3. Gary Carter 69

4. Carlton Fisk 69

5. Mike Piazza 64

[Jake]

So the Sox team WAR is the same as Cleveland's. the Angels team WAR is ome of the best in baseball. There is reason to doubt WAR as a useful, accurate tool. I like to use it, but ii really is not consistent with the standings, and I know nothing is going to be perfect, but it is significantly off in many cases.

 

If it simply predicted the standings, it wouldn't be useful. I just went over that. It is meant to be more useful than the standings - which is to say that WAR alone suggests that the Indians have really overachieved (does anyone disagree with this?) and the White Sox have underachieved (does anyone disagree with this?).

[Eminor3rd]

And why is that? Because that is the way that these stats were biased. Change the importance of a piece of the puzzle, and their values could change drastically.

 

"Biased" is the wrong word though. Run values that are attached to both offensive and defensive metrics are based on large sample correlations of historical data. No one is arbitrarily "deciding" what things are worth -- they are studying how events impact the game and weighting them mathematically.

[Jake]

Right. I don't even pay attention to catcher WAR. Ron Santo has a higher WAR (71) than any catcher in baseball history with the exception of Johnny Bench.

 

Top 5

1. Johnny Bench 75

2. Pudge Rodriguez 70

3. Gary Carter 69

4. Carlton Fisk 69

5. Mike Piazza 64

 

Yeah, and all historical data on defenders is extremely limited. The implementation of UZR in 2002 makes everything a bit different and far more accurate than the TZ pre-2002. Catchers are even harder to measure without video-based data.

[chw42]

Yeah, and all historical data on defenders is extremely limited. The implementation of UZR in 2002 makes everything a bit different and far more accurate than the TZ pre-2002. Catchers are even harder to measure without video-based data.

 

Pretty much.

 

The WAR for all those catchers is offense only. No defense incorporated at all.

[Eminor3rd]

So the Sox team WAR is the same as Cleveland's. the Angels team WAR is ome of the best in baseball. There is reason to doubt WAR as a useful, accurate tool. I like to use it, but ii really is not consistent with the standings, and I know nothing is going to be perfect, but it is significantly off in many cases.

 

Again -- it comes down to using a stat to answer the question it was meant to answer. All linear weights-based metrics, including WAR, are designed to make a descriptive, context-neutral evaluation of performance so as to be able to compare such performances across different situations (players, leagues, eras, strategies, etc.) It is essentially using math to find an ultimate common denominator. What WAR does is tell you how many wins would be created by a performance in an entirely average chain of events. This never actually occurs, but it's useful because now we can make comparisons.

 

Projections seek to do the same thing -- describe the average outcome. What those standings say is this: "Here is what actually happened and here is what the player's performances should have produced on average . It essentially is what happened versus what the teams "earned." Then, they take it one step further and say "if the players continue to play as they have, their results should regress toward what their performance 'earns,' and they should end up closer to this "

 

If you think these projections are claiming to be a crystal ball, you are mistaken, and you should read these articles (one of them involving the White Sox even!) by old Dave Cameron:

 

http://www.fangraphs.com/blogs/of-projecti...nd-predictions/

http://www.fangraphs.com/blogs/the-differe...nd-projections/

 

Anyone who has watched or played sports knows that the winner is often not the most deserving or most talented -- that is, I believe, at the very core of what makes sports interesting -- and these types of standings strive to give us a deeper look at team performance by not just showing us the winners, but also showing us who has "played the best." It's interesting information for fans like all of us, who tend to spend a ton of time picking apart every aspect of our team and its plight.

[Jake]

Again -- it comes down to using a stat to answer the question it was meant to answer. All linear weights-based metrics, including WAR, are designed to make a descriptive, context-neutral evaluation of performance so as to be able to compare such performances across different situations (players, leagues, eras, strategies, etc.) It is essentially using math to find an ultimate common denominator. What WAR does is tell you how many wins would be created by a performance in an entirely average chain of events. This never actually occurs, but it's useful because now we can make comparisons.

 

Projections seek to do the same thing -- describe the average outcome. What those standings say is this: "Here is what actually happened and here is what the player's performances should have produced on average . It essentially is what happened versus what the teams "earned." Then, they take it one step further and say "if the players continue to play as they have, their results should regress toward what their performance 'earns,' and they should end up closer to this "

 

If you think these projections are claiming to be a crystal ball, you are mistaken, and you should read these articles (one of them involving the White Sox even!) by old Dave Cameron:

 

http://www.fangraphs.com/blogs/of-projecti...nd-predictions/

http://www.fangraphs.com/blogs/the-differe...nd-projections/

 

Anyone who has watched or played sports knows that the winner is often not the most deserving or most talented -- that is, I believe, at the very core of what makes sports interesting -- and these types of standings strive to give us a deeper look at team performance by not just showing us the winners, but also showing us who has "played the best." It's interesting information for fans like all of us, who tend to spend a ton of time picking apart every aspect of our team and its plight.

 

Great post. A very lucid explanation of what these statistics are all about.

[Dick Allen]

Again -- it comes down to using a stat to answer the question it was meant to answer. All linear weights-based metrics, including WAR, are designed to make a descriptive, context-neutral evaluation of performance so as to be able to compare such performances across different situations (players, leagues, eras, strategies, etc.) It is essentially using math to find an ultimate common denominator. What WAR does is tell you how many wins would be created by a performance in an entirely average chain of events. This never actually occurs, but it's useful because now we can make comparisons.

 

Projections seek to do the same thing -- describe the average outcome. What those standings say is this: "Here is what actually happened and here is what the player's performances should have produced on average . It essentially is what happened versus what the teams "earned." Then, they take it one step further and say "if the players continue to play as they have, their results should regress toward what their performance 'earns,' and they should end up closer to this "

 

If you think these projections are claiming to be a crystal ball, you are mistaken, and you should read these articles (one of them involving the White Sox even!) by old Dave Cameron:

 

http://www.fangraphs.com/blogs/of-projecti...nd-predictions/

http://www.fangraphs.com/blogs/the-differe...nd-projections/

 

Anyone who has watched or played sports knows that the winner is often not the most deserving or most talented -- that is, I believe, at the very core of what makes sports interesting -- and these types of standings strive to give us a deeper look at team performance by not just showing us the winners, but also showing us who has "played the best." It's interesting information for fans like all of us, who tend to spend a ton of time picking apart every aspect of our team and its plight.

No but if you say X performance is worth 5 wins above a replacement, and Y performance is worth 5 wins above a replacement, their perforances should be similar. I know the most deserving team doesn't always win, but over a course of the current 95 or so to 162, it does tend to even out. You win some you should lose, you lose some you should win. WAR is obviously flawed. WAR suggests 8 teams have played worse than the White Sox so far this year. And Cleveland, a team that is way in front of the Sox has an identical team WAR. If Chris Sale has a deservingly high WAR because he gets no runs support, the fact that no runs are being scored, should lower the other WARS accordingly. It is a fun nimber, and I will continue to look at it,but it's actual accuracy has to be questioned.

 

WAR suggests an absolute. A 2 War player should get you 1 more win than a 1 war player. It should be reasonably reflected in the standings.

Edited July 18, 2013 by Dick Allen

[iamshack]

No but if you say X performance is worth 5 wins above a replacement, and Y performance is worth 5 wins above a replacement, their perforances should be similar. I know the most deserving team doesn't always win, but over a course of the current 95 or so to 162, it does tend to even out. You win some you should lose, you lose some you should win. WAR is obviously flawed. WAR suggests 8 teams have played worse than the White Sox so far this year. And Cleveland, a team that is way in front of the Sox has an identical team WAR. If Chris Sale has a deservingly high WAR because he gets no runs support, the fact that no runs are being scored, should lower the other WARS accordingly. It is a fun nimber, and I will continue to look at it,but it's actual accuracy has to be questioned.

This is assuming that the standings capture the best playing teams...perhaps the teams that are winning are benefitting from something other than the quantifiable results that we believe drives winning baseball teams...

[chw42]

This is assuming that the standings capture the best playing teams...perhaps the teams that are winning are benefitting from something other than the quantifiable results that we believe drives winning baseball teams...

 

TWTW

[Jake]

No but if you say X performance is worth 5 wins above a replacement, and Y performance is worth 5 wins above a replacement, their perforances should be similar. I know the most deserving team doesn't always win, but over a course of the current 95 or so to 162, it does tend to even out. You win some you should lose, you lose some you should win. WAR is obviously flawed. WAR suggests 8 teams have played worse than the White Sox so far this year. And Cleveland, a team that is way in front of the Sox has an identical team WAR. If Chris Sale has a deservingly high WAR because he gets no runs support, the fact that no runs are being scored, should lower the other WARS accordingly. It is a fun nimber, and I will continue to look at it,but it's actual accuracy has to be questioned.

 

Read those articles (including the one linked that talks about the White Sox). WAR and like-minded statistics assume a luck-less environment. The easy way to think about this is if you win every third game 10-0 but lose the other two 5-4, your team will look like a damn good team statistically because it outscores its opponents by so much. Unfortunately, your results will be bad. Fortunately, baseball runs are not scored in such a non-normal distribution. When metrics like these vary from the records, there is usually one of two things at play:

 

1. Luck has been on a team's side. A team with a negative run differential but good record, for instance, usually is just lucky. If they are lucky, a given team's fall is likely to begin on any given day. Their fall isn't necessarily going to be as UNlucky as their lucky run is, but it would be most likely to occur in a way commensurate with how the advanced metrics thought they would play.

 

So if Cleveland has been lucky and is in truth the 20th best team in MLB, the most likely result of their season is not the 20th best overall record. The good luck already happened and in real life, bad luck doesn't automatically catch up immediately. What is likely is that they will have the 20th best record from now till the end of the season, provided they make no changes to the team. Adding or subtracting players can change that calculus.

 

2. The other possibility is things like coaching. A good manager (and this is the hardest thing to measure empirically) can influence how timely the runs a team scores are. Advanced metrics basically assume that you'll score clutch runs at an average rate and un-clutch runs at an average rate. A good coach may be able to motivate or otherwise influence his team to perform better at clutch time than other times. Statistically, this would basically mean that the team's runs are being used more efficiently than average. It is easy to confuse this with luck!

 

The White Sox article suggests that our over-performance of projections has to do with more than luck -- in large part, it can be chalked up to tremendous health. We have begun to believe that this tremendous health is a product of the White Sox and not just the players. Some combination of Kenny Williams, Rick Hahn, Don Cooper, and Herm Schneider have led to our players (and pitchers especially) to be far healthier than what could otherwise be predicted. On its face, this would make the White Sox look lucky when in fact they've had a relatively unpredictable factor contributing to their success. This is particularly true when a player outperforms his previous health when he comes here, since projection systems can only guess about player health based on that player's own track record of health.

 

[witesoxfan]

Regarding catchers, there is work being done to figure out pitch framing. There's a great article on Grantland regarding the topic and why Jose Molina is such a stud and Ryan Doumit sucks balls.

 

http://www.grantland.com/story/_/id/927575...e-molina-others

 

It's interesting, and could very easily be the next big breakthrough regarding advanced fielding stats for catchers.

[Dick Allen]

Read those articles (including the one linked that talks about the White Sox). WAR and like-minded statistics assume a luck-less environment. The easy way to think about this is if you win every third game 10-0 but lose the other two 5-4, your team will look like a damn good team statistically because it outscores its opponents by so much. Unfortunately, your results will be bad. Fortunately, baseball runs are not scored in such a non-normal distribution. When metrics like these vary from the records, there is usually one of two things at play:

 

1. Luck has been on a team's side. A team with a negative run differential but good record, for instance, usually is just lucky. If they are lucky, a given team's fall is likely to begin on any given day. Their fall isn't necessarily going to be as UNlucky as their lucky run is, but it would be most likely to occur in a way commensurate with how the advanced metrics thought they would play.

 

So if Cleveland has been lucky and is in truth the 20th best team in MLB, the most likely result of their season is not the 20th best overall record. The good luck already happened and in real life, bad luck doesn't automatically catch up immediately. What is likely is that they will have the 20th best record from now till the end of the season, provided they make no changes to the team. Adding or subtracting players can change that calculus.

 

2. The other possibility is things like coaching. A good manager (and this is the hardest thing to measure empirically) can influence how timely the runs a team scores are. Advanced metrics basically assume that you'll score clutch runs at an average rate and un-clutch runs at an average rate. A good coach may be able to motivate or otherwise influence his team to perform better at clutch time than other times. Statistically, this would basically mean that the team's runs are being used more efficiently than average. It is easy to confuse this with luck!

 

The White Sox article suggests that our over-performance of projections has to do with more than luck -- in large part, it can be chalked up to tremendous health. We have begun to believe that this tremendous health is a product of the White Sox and not just the players. Some combination of Kenny Williams, Rick Hahn, Don Cooper, and Herm Schneider have led to our players (and pitchers especially) to be far healthier than what could otherwise be predicted. On its face, this would make the White Sox look lucky when in fact they've had a relatively unpredictable factor contributing to their success. This is particularly true when a player outperforms his previous health when he comes here, since projection systems can only guess about player health based on that player's own track record of health.

But these aren't projections, this is actual WAR accumulated this season. Cleveland is 14 games ahead of the White Sox, yet their total WAR is identical.To suggest they are even in performance but that much better based on managing or luck doesn't make sense. Just like all stats, basic and most advanced, it is flawed. Perhaps very flawed. I take it to mean not all exactly equal WAR players are exactly equal. Some must be far better than others.

Edited July 18, 2013 by Dick Allen

[GreenSox]

I certainly think issues such as lazy defense, bad base running and poor managing has contributed to half of our misery this year. I don't know the extent to which WAR measures those factors. The team isn't very good, but the starting pitching is good, the front of the bullpen was too prior to Crain's injury. Certainly not a horrendous team as it has played, although a below average team.

Edited July 18, 2013 by GreenSox

[Eminor3rd]

But these aren't projections, this is actual WAR accumulated this season. Cleveland is 14 games ahead of the White Sox, yet their total WAR is identical.To suggest they are even in performance but that much better based on managing or luck doesn't make sense. Just like all stats, basic and most advanced, it is flawed. Perhaps very flawed. I take it to mean not all exactly equal WAR players are exactly equal. Some must be far better than others.

 

I don't think you understood the beginning part of Jake's post. The reason it is off is because there is nothing like a normal distribution of events in real life. Also, it is a fallacy to assume that because someone will regress, they will regress completely to the original projection or true talent -- just that they will regress TOWARD it. Past events, whether likely or not, are in the bank, so to speak, there's no law that says it all has to even out. That's the gambler's fallacy. It's like if a coin flip is 50/50 and you flip 5 heads in a row, its not more likely that the next 5 will be tails -- it's 50/50 every time.

 

Again, take a look at those articles. Especially the second one.

[Eminor3rd]

I certainly think issues such as lazy defense, bad base running and poor managing has contributed to half of our misery this year. I don't know the extent to which WAR measures those factors. The team isn't very good, but the starting pitching is good, the front of the bullpen was too prior to Crain's injury. Certainly not a horrendous team as it has played, although a below average team.

 

It mostly doesn't -- team WAR is just the sum or individual player WAR. So anything that happens that has to do with how the players interact with one another or how different streaks or sequences of events occur will NOT be reflected. These are the factors that make reality reality and so it cannot be predicted entirely by a mathematical formula.

[Eminor3rd]

But these aren't projections, this is actual WAR accumulated this season. Cleveland is 14 games ahead of the White Sox, yet their total WAR is identical.To suggest they are even in performance but that much better based on managing or luck doesn't make sense. Just like all stats, basic and most advanced, it is flawed. Perhaps very flawed. I take it to mean not all exactly equal WAR players are exactly equal. Some must be far better than others.

 

Going further with the coin flip analogy:

 

If you flip a coin ten times, you would PROJECT 5 heads and 5 tails, meaning of all possible individual outcomes, it is the most likely. However, it is actually far more likely that the true outcome will be something other than 5H+5T, but not specifically anything else -- no individual combination is more likely to occur. Does that make sense?

[hammerhead johnson]

I'd only use WAR to evaluate individuals (other than catchers).

 

This thread is Hawk Harrelson approved.