Making Sense of the 1st Half Team MVP
by FreeZorilla on Jul 15, 2010 1:00 PM EDT
Email
PrintOver the past week, the bay area's two major newspapers made their cases for awards for the Rays over the first half of the season. Before we see what the numbers have to offer, let's first review what the local representatives of the media had to say.
Marc Topkin of the St Pete Times selected the following as his MVP rankings:
1. RHP Rafael Soriano
Soriano has done less work than almost any other Ray, pitching just 322/3 innings over 33 games. But none have had more of an impact on their success. Simply put, Soriano, in an all-but perfect performance with 23 saves in 24 chances, is the one player who, thus far, has made the biggest difference.
Soriano was followed in order by David Price, Carl Crawford, Jeff Niemann, and Evan Longoria.
Roger Mooney of the Tampa Tribune selected Soriano and Crawford as Co-MVPs.
What does WAR tell us?
|
Player |
WAR |
|
Car Crawford |
4.8 |
|
Evan Longoria |
3.7 |
|
3.1 |
|
|
David Price |
2.4 |
|
1.6 |
|
|
Rafael Soriano |
0.9 |
It's worth noting that Grant Balfour has produced more WAR, and Joaquin Benoit the same WAR as Soriano. Soriano also places behind the others in FIP and xFIP, and well behind Joaquin benoit's 0.80 tERA. It is hard to argue that Soriano has performed the best in the bullpen, let alone on the team.
However there is one compelling case for Rafael Soriano, and that involves factoring leverage into the results. If we look at each players Win Probability Added which looks at the change in base/out state for each players' events, but weights them in terms of inning and margin we get the following results:
|
Player |
WPA |
|
Rafael Soriano |
3.2 |
|
Carl Crawford |
2.19 |
|
Jeff Niemann |
2.1 |
|
David Price |
2.07 |
|
Evan Longoria |
1.66 |
While Soriano, Balfour, and Benoit have all pitched extremely well, Soriano recieves a boost here because he pitches in the ninth. If he slipped up one or two more times he would have been heavily penalized, but he has indeed been lights out as the closer even if he hasn't been our best reliever. Soriano's WPA is good for tops amongst American League pitchers. At first I was critical of Soriano being included, but when telling the descriptive story of results including leverage, he has been a full win ahead of Crawford. Soriano's production could be easily replicated by Benoit, and Benoit's by Balfour. This is called bullpen chaining. The production dropoff from Crawford would be far greater. In terms of value to the team he remains my king, but I won't slam somebody for selecting someone verfied by WPA.
191 comments
|
1 recs |
Do you like this story?
Comments
One comment I left out
Crawford is not credited for his defensive wizardry in WPA as its solely an offensive stat with no consideration for defense or position.
Follow Me on Twitter @FreeZorilla
Roger Mooney of the Tampa Tribune selected Soriano and Price as Co-MVPs.
This is wrong. He selected Soriano and CC as co-MVPs.
I hope you understand how provocative and irritating a statement such as "baseball is played on the field not on paper" is. It is the kind of moronic critique that anti-sabermetric neanderthals use (along with the unfunny "mother’s basement" canard) to debunk what they do not understand.
In terms of value to the team he remains my king, but I won't slam somebody for selecting someone verfied by WPA.
If either of these guys knew what a “WPA” was, I’d agree with you. Instead, they want to talk about assholes packed full of grit.
I hope you understand how provocative and irritating a statement such as "baseball is played on the field not on paper" is. It is the kind of moronic critique that anti-sabermetric neanderthals use (along with the unfunny "mother’s basement" canard) to debunk what they do not understand.
What is WPA/LI?
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:11 PM EDT up reply actions
Can you or Slow do a piece about leverage index? I'm not sure I understand it
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:14 PM EDT up reply actions
I've deferred all stat explaining pieces to Steve
He has a gift for explaining them.
Follow Me on Twitter @FreeZorilla
Agreed
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:19 PM EDT up reply actions
"Be careful what you're good at"
I guess I got myself into this. Okay, I’ll see what I can do. Might not air until the weekend.
I love Casey Fossum. Now try and take me seriously.
by Steve Slowinski on Jul 15, 2010 1:29 PM EDT up reply actions
Your saberlibrary link was enough for me, but I've enjoyed your stats stuff so far, so I'm not going to complain.
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:30 PM EDT up reply actions
Okay, sounds good...I'll see what happens
The stats stuff has worked while there hasn’t been much else news, but I think it’ll get pushed out by real news and rumors once the games start again. We’ve covered most of the basic/important ones that we wanted to, and Glass has an article forthcoming.
I love Casey Fossum. Now try and take me seriously.
by Steve Slowinski on Jul 15, 2010 1:39 PM EDT up reply actions
Can't wait for the Wake-Take (tm)
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:41 PM EDT up reply actions
Context neutral wins
Basically you divide a player’s WPA by their leverage index, which gives you how many WPA “wins” they’ve been worth.
Check this out. It takes a bit to wrap the brain around the concept.
I love Casey Fossum. Now try and take me seriously.
by Steve Slowinski on Jul 15, 2010 1:14 PM EDT up reply actions
Here:
http://www.fangraphs.com/blogs/index.php/get-to-know-wpali/
WPA/LI (context neutral wins / game state linear weights): How many wins a player contributes to his team with the Leverage Index aspect removed, invented by Tom Tango.
Calculating WPA/LI: WPA is divided by LI for each individual play attributed to a specific player and then the WPA/LI for the individual plays is then added up to create WPA/LI for an entire season. This is considerably different then taking a player’s WPA and dividing it by pLI.
Why you should care: Unlike standard linear weights, WPA/LI does take into account the situation. So at times when a walk would be just as valuable as a home run, WPA/LI accurately weights the walk and the home run, where linear weights would still give .13 wins to the home run and the walk .03 wins.
by R.J. Anderson on Jul 15, 2010 1:15 PM EDT up reply actions
Gracias to FZ RJ SS
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:16 PM EDT up reply actions
There's no denying Soriano's WPA is heavily inflated by leverage
Thats an opportunity only a closer gets. almost by default a near perfect closer will lead the team, but its difficult to be near perfect. it just so happens the Rays had two and almost three pitchers that fit this bill in the 1st half.
Follow Me on Twitter @FreeZorilla
I'm just saying.
Leverage is mostly out of the player’s hands. Obviously a pitcher can have some control over it by pitching well, but even then, you have teams where the setup man is better than the closer (the White Sox, for instance).
by R.J. Anderson on Jul 15, 2010 1:17 PM EDT up reply actions
That's why Soriano was quick to thank the manager
for his role as closer and subsequent All-Star appearance.
RaysRev.com or m_weber on twitter
Sure I agree
Benoit has had better results, just not leveraged opportunities.
Follow Me on Twitter @FreeZorilla
That's why FG's WAR splits the leverage in half.
Doesn’t give 100% credit to the reliever, but doesn’t ignore it.
by R.J. Anderson on Jul 15, 2010 1:20 PM EDT up reply actions
No. There were conversations about going to a chaining system.
But as far as I’m aware they’ve done it this way since introduction.
by R.J. Anderson on Jul 15, 2010 1:24 PM EDT up reply actions
I should note those conversations weren't FG based.
I think it was a thread on BTB or The Book blog where others discussed it.
by R.J. Anderson on Jul 15, 2010 1:25 PM EDT up reply actions
Where is it factored specifically when looking at a pitcher's WAR table
Follow Me on Twitter @FreeZorilla
Question based on what I just read:
If you neutralize leverage from win probability added, how does that help the statistic? Surely you would agree that different points in the game are more or less important, and the course of the game has to some degree been dictated by a building up of other events? (Example, reliever choice based on early game score)
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:22 PM EDT up reply actions
How do you mean? Without leverage why wouldn't you just use WAR?
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:26 PM EDT up reply actions
(Which seems like a better stat to me anywhoo)
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:26 PM EDT up reply actions
WAR incorporates FIP
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:29 PM EDT up reply actions
Yes, WPA does not
WAR & FIP are better talent evaluators, whereas WPA is event descriptive
Follow Me on Twitter @FreeZorilla
Which is why
WPA is superior to determine who has been most “valuable” to a team, while WAR and FIP (and other stats) are superior to determine which player is truly “best” or “most talented.”
So for the stated purpose (naming a first-half MVP), WPA actually would be a very good metric to consider.
by Not Whole View Gang survivor on Jul 15, 2010 2:52 PM EDT up reply actions
WAR is FIP based whereas WPA is more descriptive of all events
each play is measured based on the change in run expectancy. It simply looks at the change in the base and out state.
Follow Me on Twitter @FreeZorilla
So if each play is measured based on a change in run expectancy, why strip out leverage? If a team is up by 8 they probably have a shitty reliever out there.
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:29 PM EDT up reply actions
Because leverage isn't in the hands of the players.
Crawford isn’t being placed into the situations, so it’s not his fault if he’s got a .500 wOBA but only comes up with the bases loaded.
by R.J. Anderson on Jul 15, 2010 1:30 PM EDT up reply actions
I guess this goes back to the old argument about "clutch," but it seems to me if you're stripping out all outside factors just go with WAR
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:31 PM EDT up reply actions
"...if you're stripping out all outside factors just go with WAR"
I agree with this. I’m not a huge fan of WPA/LI.
I love Casey Fossum. Now try and take me seriously.
by Steve Slowinski on Jul 15, 2010 1:33 PM EDT up reply actions
High contact/low SO hitters tend to do better with runners on base than TTO players.
And since most high leverage situations have runners on base, the TTO players end up being slightly less clutch than the contact hitters. Not due to a psychological clutchiness, but due to having skills that better fit clutch situations. WPA/LI will detect that.
At this point, you should just use WAR. WPA/LI was the best there was before WAR came around.
Now it’s kinda obsolete. Or at least, it has very limited uses.
I love Casey Fossum. Now try and take me seriously.
by Steve Slowinski on Jul 15, 2010 1:31 PM EDT up reply actions
Its still more descriptive, but outside of subjective awards most of what we do is talent evaluation
Follow Me on Twitter @FreeZorilla
Its also more useful when comparing across the same team
Quality of defense is neutralized, though defensive luck isn’t
Follow Me on Twitter @FreeZorilla
True.
I just see it as one of those stats that 5 years from now has disappeared into the woodwork. I think Win Probability charts are great, and WPA will probably still be around because it’s fun, but I always felt like WPA/LI was an attempt to analyze talent before the days of WAR. Now we have better options available, and I haven’t heard people use WPA/LI when discussing players in a long time.
I love Casey Fossum. Now try and take me seriously.
by Steve Slowinski on Jul 15, 2010 1:45 PM EDT up reply actions
I'm still confused between RE24 and WPA/LI
RE24 considers base state and outs
WPA considers the same plus margin and inning
WPA/LI considers ?
Follow Me on Twitter @FreeZorilla
That's part of my problem with WPA/LI...
It’s confusing and not very intuitive. It makes my brain hurt every time I try conceptualizing it.
I’d need to look up more to answer that correctly. I’m not entirely sure how to articulate the differences.
I love Casey Fossum. Now try and take me seriously.
by Steve Slowinski on Jul 15, 2010 1:49 PM EDT up reply actions
Can you compare it to RE24
I know you or Tango attempted this yesterday. Specificailly what context is included in each.
Follow Me on Twitter @FreeZorilla
Sure.
Contexts included in RE24: runners on base, outs in inning
Contexts included in WPA: runners on base, outs in inning, inning, score
Leverage is the importance of each situation. We’re familiar with WPA Leverage because that’s what Fangraphs shows. You could also talk about RE24 leverage, though.
Ignoring context, a single is worth about .5 runs more than the average event, a HR 1.4 runs more, and an out .3 runs less. In general, a WPA leverage of 2 means a single is going to help win the game with the power of 1 run, a HR 2.8 runs, and an out -.6 runs. It doubles each event’s importance.
Now, everything doesn’t get doubled with a leverage of 2. In different situations different events are worth more or less relative to each other. Take the runner on second in a tie game in the bottom of the ninth example. A single and HR are worth the same. Maybe it’s 6 runs and an out is -4 runs (huge numbers because it’s a hugely important situation, say an LI of 10). WPA hands out those huge numbers as is. WPA/LI will divide each by the LI of the situation. So a single is worth .6 runs, so is a HR, and an out is -.4 runs. WPA/LI makes it so that the credit/blame you can get for each plate appearance is the same, but adjusts the relative importance of each outcome.
That help? Where can I elaborate more?
Hmmm
Context of RE24 vs WPA/LI. I think I am comfortable with RE24 and WPA. I have a hard time distinguishing between RE24 and WPA/LI outside of the walkoff example (I think?)
Follow Me on Twitter @FreeZorilla
It's not just the contexts that are different/relevant here.
There’s another dimension: importance.
WPA and WPA/LI have the same context. RE24 and RE24/LI have the same context.
By using WPA instead of wOBA you’re taking into account both the context and importance of a situation. WPA/LI (and RE24/LI) keep the context, but strip away the importance.
Let me see if I can find some actual numbers to use as an example. Off to the Book Blog…
Leverage defines importance (inning/margin)
Re24 only looks at the change in base/out state, or am I wrong?
WPA looks at the same thing but adjusts it for importance as defined by Li.
Am I missing a fundamental difference between RE24 and WPA/LI beyond WPA/LI indentifying a walkoff situation?
Follow Me on Twitter @FreeZorilla
The leverage used for WPA accounts for all four contextual factors, not just the additional two beyond RE24.
runners on base, outs in inning, inning, score
Every situation will see the linear weights change, not just a walk-off situation. Bases empty, a walk is as good as a single. Runner on second, the double and triple show less of a gap. Runner on first, single and double show a wider gap. Two outs, all events are more tightly clustered. Two outs and up by ten runs, all events are worth basically the same (because they don’t affect the outcome of the game.)
And thats all in both RE and WPA
The difference between them is RE24 strips out inning and score. But isn’t that the same thing as WPA/li?
Follow Me on Twitter @FreeZorilla
Yes, RE24 strips out inning and score.
But WPA/LI doesn’t strip out any of the contexts, it just strips out the IMPORTANCE of those contexts. The contexts remain in order to calculate a context-dependent wOBA.
I’m saying the same thing again and again, because that’s how I see it in my head. I’ll try to find different ways of saying it from other people. One link below…
It seems as though take RE24 and dividing by 10 (10 runs/win)
would get you pretty close to to WPA/LI
Follow Me on Twitter @FreeZorilla
Not really.
RE24/10 would get you pretty close to WPA. You’d just be ignoring inning and score. And RE24/10 is a Fangraphs stat: REW
(I’d ignored that RE24 is in runs and WPA is in wins, so good that you brought that up. Something to consider, but not much difference in what they’re measuring; just a difference in scale. I’ve really been comparing REW and WPA)
Feel free to cut me off at any point now
RE24/10 would get you pretty close to WPA. You’d just be ignoring inning and score.
By stripping out LI, from WPA, aren’t we replicating that?
Follow Me on Twitter @FreeZorilla
Haha.
Stripping out LI ignore the IMPORTANCE of the inning and score, but not the situation.
Ok, here’s something that might help:
RE24 has obvious ways that changing your approach matters: one out with a runner on third is different from two outs and bases loaded.
But adding in inning and score is less obvious. Here’s an example, though: let’s say there’s a runner on third and one out. If it’s the first inning in a tie game, you want to hit a sac fly. But if it’s the ninth inning and you’re down two, screw the sac fly, you just want to get on base.
Ignoring the importance of those two situations, the approach of the batter is different and the relative values of getting on base vs. hitting fly balls (making outs) are very different. That’s what WPA/LI looks at.
Hmmm.
Say the LI is 4.0, that means the situation is 4 times as important as the typical situation. A run scored here is four times as important to winning as a run scored in a typical situation. WPA will give the team credit for 4 runs for each run scored. Why? Because the win probability changed four times as much as in a typical situation. Four runs didn’t actually scored, but that one runs was as important as one normal run.
You can also think of the runs-per-win number changing. Usually 10 runs = 1 win. But with a LI of 4, that drops to 2.5 runs = 1 win.
Likewise, WPA will credit a player with 4 PAs in this situation. If he gets a hit, it’s like going 4/4. If the LI is only .5, it’s like going .5 for .5.
WPA/LI strips the importance away, so that a run in every situation counts the same and each PA counts as exactly 1 PA. (But WPA/LI doesn’t ignore the context, customizing wOBA for each situation.)
Also, this:
There are many ways to have a leverage of 3. Maybe one is bottom of ninth, no outs, runner on 2nd, tie game. Maybe another is top of eighth, one out, down three.
That the LI is 3 means these situations are both 3 times as important as the typical situation. But they obviously aren’t the same situation. In the first, a walk is next to useless — it doesn’t advance the runner and will never need to score. In the second, a walk sores a run. In the first, a home run is not more advantageous than a double or (most) singles. In the second, a home run is way better than a double.
So let’s say you have to send a guy to the plate who’s merely “okay” in each situation. In the first, you’d want his skills to be heavy in getting hits, but you don’t need power or walks. In the second, you’d want his skills to be high in home runs and walks.
Those last two paragraphs are what WPA/LI is all about. The situations are different, so the value of each event shifts. The LI is 3, but WPA/LI doesn’t care about that part of it. It just sees one plate appearance and says, given this situation, what are the relative values of outs/singles/doubles/etc.
Here's a solid article.
Math and concept heavy, but good. And at this point in the discussion, you can’t really avoid being heavy ;)
http://www.insidethebook.com/ee/index.php/site/article/game_state_version_of_woba/
another one
Comments are really good for this one (not so much the first one).
http://www.insidethebook.com/ee/index.php/site/comments/wpa_li/
WPA/LI considers all the same things as WPA.
It’s just that instead of making plate appearances more or less important according to the situation, it makes all plate appearances equally important.
Like I wrote below WPA/LI is basically a version of wOBA where the weights of the inputs change depending on the situation.
WAR uses wOBA, basically, which treats all singles the same.
That’s pretty good. But using WPA/LI accounts for the fact that the value of the single isn’t the same relative to other events in every situation. Two examples:
- With the bases empty, a walk is as good as a single. WPA/LI knows this, wOBA does not.
- With a runner on second and a tie game in the bottom of the ninth, a single is (nearly) as good as a home runs. WPA/LI knows this, wOBA does not.
Let's just put it this way
He blows a save and loses, and that WPA drops by at least 0.50, so the leverage swings both ways.
Bad Left Hook - The SB Nation boxing blog
"Baseball is played on the field, not on a calculator."
Yes
Soriano recieves a boost here because he pitches in the ninth. If he slipped up one or two more times he would have been heavily penalized, but he has indeed been lights out as the closer even if he hasn’t been our best reliever. Soriano’s WPA is good for tops amongst American League pitchers
Follow Me on Twitter @FreeZorilla
Love when the Most Valuable Player debate
goes all algorithmic. LOL
For my money, I think factoring leverage is important when it comes to most valuable player as opposed to best player on team.
RaysRev.com or m_weber on twitter
Funny you mention money, b/c on a team with constraints Soriano can only meet the demands on his contract
Follow Me on Twitter @FreeZorilla
I firmly believe that arguing the semantics of the phrase "most valuable player" is the least productive thing anyone can do.
To me it’s Craw
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:13 PM EDT up reply actions
Lets keep this thread on topic please?
Mark Topkin reporting that TGN1 has requested that the thread stay on topic
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:16 PM EDT up reply actions
I disagree...
I wouldn’t say Soriano is the best player on the Rays. But, I would argue that he was the most valuable player for the 1st half. It may be semantics, but it’s a valid point in the discussion.
RaysRev.com or m_weber on twitter
The stat WAR can be directly translated into value
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:19 PM EDT up reply actions
And WPA overvalues closers.
The answer is probably somewhere in the middle.
I love Casey Fossum. Now try and take me seriously.
by Steve Slowinski on Jul 15, 2010 1:36 PM EDT up reply actions
Sure
To be it goes back to what you are looking for. WAR/FIP measures process performance. WPA measures the relative value of results. RJ is absolutely right about opportunity beng out of the players’ hands while performance isn’t.
Follow Me on Twitter @FreeZorilla
Just because its out of the players hands doesn't mean its not relevant to the value the player provided in the past.
It may not be predictive but it is relevant to MVP discussions.
Fair enough...yeah, I can agree that there's a case to be made for Soriano.
Personally, I prefer to recognize CC for his efforts. Considering how much he’s done for this franchise and for the team this year, I want to make sure he gets his due shout-out.
I love Casey Fossum. Now try and take me seriously.
by Steve Slowinski on Jul 15, 2010 1:46 PM EDT up reply actions
Even without the stats.
The eye test shows that Crawford has been far more valuable then any reliever. Thats what shocked me about the writers decision since they base their selection on the eye test.
Most would agree, and again Crawford's WPA does not consider his defense.
Follow Me on Twitter @FreeZorilla
Leverage, but RJ suggested leverage was included in WAR which I did not realize
Follow Me on Twitter @FreeZorilla
Also, since WAR is partially a counting stat, wouldn't it follow that since relievers are the least used players they would be penalized?
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 3:31 PM EDT up reply actions
But that seems justifiable when determining value
But since its a counting stat, it seems counting the leverage of the appearance should be considered.
Follow Me on Twitter @FreeZorilla
I don't disagree with that, but its clearly a penalty for relievers
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 3:39 PM EDT up reply actions
But, like FreeZorilla said...
It doesn’t tranlate leverage. Which is just as important when it comes to value.
RaysRev.com or m_weber on twitter
I know what your saying
But James Shields’ WAR is higher than Soriano’s. I don’t think anyone can argue Shields has been more valuable than Soriano to the team’s success.
Sure and thats b/c WAR for pitchers is FIP based
a metric that favors Shields. Thats a valid point. WAR works better for hitters IMO
Follow Me on Twitter @FreeZorilla
Good point.
If you use his ERA, his WAR is about 1, not drastically lower, but lower.
Rally’s WAR, which uses ERA but adjusts for quality of defense, has him at, uh, 0 WAR. 0 WAR with a 4.87 ERA? I don’t get that. 4.87 is not near replacement level.
Also, just want to say its great to have you around to further the discussion
Few can challenge viewpoints, remain open-minded and be as respectful as you do. One of the least intimidating people to ask questions of, and your answers are some of the easiest to understand. Slowinski is like a budding Kalkman.
(wipes chin)
Follow Me on Twitter @FreeZorilla
It's nice to have intelligent discussion about these things. A place where saber noobs aren't shunned and feel like they can ask questions.
I like this blog.
I should warn you FZ used to have Zobrist pajamas, but he know owns Sky Kalkman pajamas.
And before TGN1 asks how I know, sf1 told me.
As you can always expect come from behind victory is when you least expect it.
GOD IT'S A CREEP-FEST IN HERE
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 7:12 PM EDT up reply actions
Can't this be, in some ways, a better way to measure performance to date
At least when you are comparing pitchers on the same team. Meaning, all the pitchers on the Rays pitch in front of the same defense, and we are trying to figure out how valuable the player has been so far. In essence, what I’m saying is: since an MVP is retrospective judgment, can’t we use Rally WAR to judge pitchers’ performance to-date? My thinking is the amount runs they let cross is what really matters to this point. Rays’ pitchers pitch in front of the same defense and same fields for most part. Whether or not they are playing over their head or whether they will regress to the mean shouldn’t matter. Just trying to think a little differently here.
I agree somewhat when measuring performance-to-date
You’ve neutralized the quality of the fielders, but not the luck nor the sequencing of events. I woudl say it doesn’t work at all with relievers who are constantly shuffled in and out of jams and not fully responsible for runs scored, but hey base/outs states help with that!
Follow Me on Twitter @FreeZorilla
True, I guess the element of luck is still there when using ERA as a metric
So BABIPA could still largely affect Rally WAR. Like Shields’ 2010 BABIPA is .342, when his career average is .318. So, even though you’re pitching in front of the same defense (mostly) compared to guys on the same team, lucky hits may still be falling. I hear what you’re saying.
I do think that could be argued. If you replace Sori with say, Matt Lindstrom, there is less of a drop off in actual wins than if you replaced Shields with Jason Marquis or something.
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:23 PM EDT up reply actions 1 recs
HO HO HO
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:25 PM EDT up reply actions
I think one could argue that.
What if a crappy reliever had pitched instead of Soriano, bumping Balfour and/or Benoit into the closer role and everyone else up a notch on the bullpen ladder. How many games would that have cost the Rays?
Now put a crappy starter in place of James Shields. How many games would that have cost the Rays?
There’s no obvious answer, but that’s exactly the point — they’re close.
This is accurate.
The argument is not who is the most “talented” player or who is the “best”, its about who is the most valuable. Luck with the situations presented to you and how you perform in them can’t be disregarded just because its luck, at least not in the context of this discussion. The numbers show it, given Soriano’s leverage and his performance he has been the most valuable.
he's been more valuable than half of CC.
He has not been more valuable than CC the entire player.
by rglass44 on Jul 15, 2010 2:02 PM EDT up reply actions 1 recs
WPA
I guess “one number shows it”. I prefer to factor in leverage as much as possible when looking at past contributed value.
Gotcha.
Two problems to consider with using straight WPA, however.
- Bullpen chaining: If you flipped Soriano and Benoit, Benoit would have the huge WPA and Soriano would be merely ok. The Rays might lose a fraction of a win because of this. So Soriano isn’t really that valuable — having the quality innings in the closer role has been valuable, but a number of Rays pitchers could have provided that.
- Replacement level. A replacement level reliever will post a 0 WPA. But replacement level position players and starters would be at about -1 WPA right now. That bumps Crawford/Niemann/Price up to about 3.2 WPA over replacement, right beside Soriano. And that’s not considering the defensive exploits of Crawford, which would increase his WPA.
I guess we get into a semantics argument here over what MVP means.
1) I prefer not to look at the guys behind a player when determining his value. You can’t take away from Soriano’s contributions just because somebody else on the team could have done the same thing. The fact is, he is in the closer role and Benoit is not.
2) I agree in this case. The defense of Crawford can’t be discredited.
I agree that you shouldn't look at the ACTUAL player behind someone to determine his value.
Poor example on my part.
But I do think you have to look at SOME baseline level of performance to compare every player to. This is usually replacement level, for good reasons. So take Soriano, plus a typical bullpen #2 through 7. Now remove Soriano, bump everyone up a spot, and throw the replacement reliever in at the end (he’s not going to close games just because he took the closer’s roster spot).
That calculation is hard. And it’s really hard to do with WPA rather than simply ERA or FIP. Suffice it to say, it knocks a guy like Soriano down a few notches.
I’ll pass on the semantics argument, too. ; ) I may point out potential issues with a line of thinking, though.
No problem with comparing him to replacement level.
I think you would have compare a replacement level player given Soriano’s same leverage situations, without considering bullpen chaining. At least in my view, thats how value is determined.
You could actually chain position players, too, it just matters a lot less.
Say CC goes down. Jennings or Joyce would step in, both being 2 win players. That’s a loss of 3 wins, not CC’s 5 WAR number.
But that’s not really fair to CC, the Rays being stacked in the OF and all. The USUAL backup outfielder might be more like a 1 win player. And since that guy no longer can come off the bench, your pinch-hitter is worse, costing maybe another .5 WAR. So CC is a 4.5 WAR guy after bumping everyone up a notch in role.
Same thing could be done for Soriano. On a USUAL team, there’s someone decent to take over the closer’s role. It’s not a Balfour with a 2.50 ERA, but maybe someone with a 3.75 ERA. The AAA shmuck goes at the end. In that case you don’t lose the full 3.2 WPA, but maybe half of it.
Anyway, I’ll leave it at that unless something new pops up. Not trying to force this down anyone’s throat, just trying to explain how I see it as well as I can.
This is the point I was trying to get at yesterday.
Very interesting. This may be a stupid question, but do organizations do any type of chaining or look at actual replacements to see where they are “thin” at a position? For the Rays, we know that OF depth is a strength (in theory). However, catcher, or 3B may lack anything behind Longo, Jaso/Shoppach etc. Would Friedman and others look at numbers to see what I see using names? I’m interested in the multi-level strategy or analysis teams do, specifically when entertaining trades.
If we put assumed Longo and CC were equal players, at the same age and salary, I’d think we were better off dealing CC due to what we have behind him. How do you quantify that when determining which prospects to deal or get in a trade?
As you can always expect come from behind victory is when you least expect it.
I think Topkin and Mooney just want to illustrate how this team is different from 2008 and 2009
I think they are basing their MVP on impact due to never really having a legitimate closer on the Rays. Howell was very good last year, even considering the fall of towards the latter part of the year. But I think they are taking it from the perspective of: “He has been the difference compared to Rays’ teams of the past.”
I’m not agreeing with their point-of-view, but that’s just what I take away from their opinions. IMO, CC is the MVP.
.
These same guys were around when the Tampa BBWAA gave Gerald Williams (0.5 WAR and -1.50 WPA) and Jason Bartlett (1.7 WAR and -1.40 WPA) team MVP awards. So frankly, I don’t think they know what the hell they’re talking about with these decisions.
by R.J. Anderson on Jul 15, 2010 1:20 PM EDT up reply actions
Why does who they pick as MVP bother you so much?
It’s like you take it as a personal insult.
Because they are bad at their job
And Topkin’s remarks about how people like Keith Law “weren’t paying close attention” in his Bartlett defense piece was about the most ridiculous thing I’ve seen.
by R.J. Anderson on Jul 15, 2010 1:25 PM EDT up reply actions
Not to speak for RJ
But it bothers the hell out of me because this is what the average fan sees and reads. They see the beat writer who follows the Rays saying that Barlett is the team MVP and now they bash other players who are better and hold an average SS to a level he doesn’t deserve. If we had traded Bartlett in the off season there are fans who would be screaming about boycotting the Rays, firing Friedman because he’s an idiot and harping about the cheapness of Sternberg because he just traded the team MVP. When in reality, it would have been because we had received something that was as or more valuable than an average MLB SS.
by ReyL on Jul 15, 2010 2:15 PM EDT up reply actions 1 recs
I was going to say something to this effect
They appeal to the average fan by not choosing the obvious choice and being different. This differentiation is mistakenly interpreted as “expertise” by the average fan. Then they just repeat what they hear. Good point.
This is a great piece FreeZo and a perfect example of why I am always excited to read your articles.
Like most of us, i am sure your first reaction was incredulity when you saw that Topkin picked Soriano as first half MVP. But instead of just bashing you dug into the stats and were able to come up with a reasonable statistical explanation for his pick.
He's back!
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:40 PM EDT up reply actions
What was it like going to school with Firpo Marberry?
by R.J. Anderson on Jul 15, 2010 1:46 PM EDT up reply actions
i was looking at all time WAR leaders and saw Delahanty
and noticed he diied in the middle of the 1903 season while still playing, so i googled him and saw he was thrown off a train drunk and fell into Niagra Falls—check it out
But his WAR numbers are pretty amazing
by sternfan1 on Jul 15, 2010 1:49 PM EDT up reply actions 1 recs
LL had a thread about ridiculous baseball deaths a while back.
I believe he was included.
Also some nut who hatched his entire family.
by R.J. Anderson on Jul 15, 2010 1:51 PM EDT up reply actions
A ruptured inguinal hernia.
Fucking ouch.
Maddon's Mission
Make you want to kill him, then make you want to love him. Sly.
by Jonah Keri on Jun 19, 2010 10:31 PM EDT
What a strange season
RE24 north of 20, but a WPA of -.02. Terribly unclutch season -1.41
Follow Me on Twitter @FreeZorilla
No, if you are WAR leader is a 3 win player, you've got a team that shares, rather than one glory hog
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 2:43 PM EDT up reply actions
Agreed 100%, but really the purpose of my comment was to make fun of Aryan for his "you're" gaff
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 3:16 PM EDT up reply actions
BR's WAR isn't identical to FG's
But Gossage had 7 WAR in 1975.
by R.J. Anderson on Jul 15, 2010 1:50 PM EDT up reply actions
I don't think any teams have a reliever leading their team in WAR
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:38 PM EDT up reply actions
Then I don't get your point
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:41 PM EDT up reply actions
I've just been dragged kicking and screaming into the No-Spin Zone
The caller said the boy, after removing the bulb from its socket, left the building and threw the bulb on the ground. When the bulb broke, the caller said the boy screamed "I am the cat and I am here to steal."
by Top Gun Numba 1 on Jul 15, 2010 1:43 PM EDT up reply actions
Benoit
I have found myself wondering lately how Benoit would do as a closer if, God forbid, Soriano got hurt. Benoit has been an absolute beast out of the pen. We don’t ever get the pill to Soriano in many of those games without Benoit.
I will absolutely slam these choices.
They’re awful, and I guarantee they weren’t checking the WPA leaderboard. I also think the value of WPA is grossly overstated.
Look at the all-star game
Evan Longoria had a two out double with nobody on, a leadoff walk, and started a fine defensive double play. Brian McCann had two plate appearances.
1) Runners at the corners, two outs, he flew out
2) 2 out bases clearing double
Those are 2 great opportunites, of which he took advantage of one. Longoria had more successful individual plate appearances, but was not afforded the high leverage at-bats. No one has a problem naming McCann player of the game, it is not unreasonable to apply the same standard to a season.
I can understand why one finds these subjective awards uninteresting and meaningless, but upon further inspection Soriano is a somewhat reasonable choice, though not my own personal one.
Follow Me on Twitter @FreeZorilla
I agree with this logic totally
Especially with the comparison from the ASG. I think it is completely reasonable to consider Soriano as team MVP. It wasn’t like the 2008 Barlett season MVP, where we just happened to lose some games when he was hurt. I’d still go with CC for first half MVP, but I can understand where they are coming from. Again, I think they were judging this on a player that has made a needed impact, and something the Rays have never really had. This team has never had a shut down closer like Soriano, and I think Topkin is getting caught up in same sensationalism over that.
I agree that it's very overrated in terms of valuing players.
But, as the cliche goes, it’s a “story stat”. When your guy tightens up watching a game, those are high leverage situations. Soriano has come through in those situations. In other words, when these writers have been sitting on the edge of their seats, Soriano has succeeded, making them very happy. No, they aren’t looking at the WPA leaderboard, but their guys and WPA are measuring the same thing.
It’s just that their guts are really bad analysts in any other sense.
I'm sure they never checked WPA. They just went with their gut, which is what they always do
But choosing Soriano isn’t asinine, even though I don’t agree with it either. Now if they chose John Jaso, then that would be another story. Sadly, I could have seen them doing this.
Is WAR the proper metric to use?
A lot of WAR is from defensive value, but how accurate is a halfseasons worth of defensive numbers?
My Mvp- CC
Second I’d put Soriano. Quantitatively he isn’t that high but solidifying the bullpen was very important and Soriano did just that. Doing that was important psychologically as well as performance. So really you can say my “second” MVP is a co-Soriano and Friedman tagteam
Go Gators!!
by matthan on Jul 15, 2010 2:52 PM EDT via mobile reply actions
Here's where I'm a big fan of WAR because of its framework, not necessarily because of the individual calculations.
Keep the numbers you have faith in: hitting runs, position adjustment, playing time runs, whatever. Fill in the rest, like fielding your position, with whatever you want (although people can argue with you.) Boom, something useful. You can debate each piece of the puzzle point by point instead of having traditional “I know you are but what am I” style discussions.
My Co-MVPs = Jaso and Shoppach.
Because anything that isn’t Navarro is extremely valuable to me.
Sign lady must die.
dude, this one's over and so is the season
by sternfan1 on Jun 19, 2010 11:38 PM EDT up reply actions
WPA can be a little misleading because of replacement level.
A replacement level reliever will post about 0 WPA, while replacement level hitters and starters will post about -2 WPA over a full season (I think I have that right). So half way through the season, full time starters and position players should have 1 win added to their WPA when comparing them to relievers.
The trend seems be the writers comparing players to the position from previous years.
That’s what the Bartlett and Soriano picks have in common. They did a great job filling a position of need. The logic being that the team is much better this year because this player is so much better than who was there before. It’s flawed logic, but I can see where it is coming from.
Agreed, it's flawed logic...
The media tends to overrate players by judging them by the difference between the current and the old, instead of just looking at the current player in a vacuum. It happens in baseball just like it happens when a new QB comes in and replaces an awful one.
If the previous years team had a major problem, and a player comes in and solves that problem, the media (or common fan) generally overrates the player’s value. That’s why I can see why these guys keep getting picked over the consistently great players.
We know RE24 pulls from the base/out run expectancy matrix
We are in agreement here. I am under the assumption (I think this is where I am wrong) that WPA simply pulls from the RE matrix and applies the leverage index to weight the event. Is there a seperate matrix it draws upon? if not, and we unweight it by stripping LI, it seems we are back to RE24.
Follow Me on Twitter @FreeZorilla
2nd and 3rd two outs bases clearing single
Situation 1: team is leading 4-0 in the 3rd
Situation 2: teams is trailing by 2 in the 7th
Can you walk through the exercise in determining RE24, WPA, and WPA/LI for each? This shoudl clear it up.
Follow Me on Twitter @FreeZorilla
Also, keep in mind that there are two diferent things here.
You can have the current run/win expectancy or the CHANE in the run/win expectancy. The RE24 and WPA numbers are the changes. If you let me start with run/win expectancy charts that others have done and I just have to compute the changes, it’s a lot easier…
Ok, hopefully this will help:
(Note, I’m changing the first situation to up 3-0, because the win expectancy table i’m using doesn’t go beyond a 5 run lead.)
First, the context-neutral single. In linear weights runs, it’s +.5 runs. In wins, that’s about +.05 wins.
Next, the RE24 single. Using this table (because it’s what I could find quickly), the run expectancy of each situation is .634 runs. With the two-run single, that scores two runs and leaves a run expectancy of .251 for a change of 2.251 – .634 = +1.617 runs, or +.16 wins.
Next, the WPA single. Using this table (because it’s what I could find quickly), the win expectancy of the first situation (assuming home team) is .877. With a two-run single, it moves to .952, for a change of .075 wins. The second situation goes from .260 to .551 for a change of .291. So in these two situations, a single is worth +.075 wins and +.291 wins, respectively.
Before we hit leverage, you can see how much context matters. Context-neutral, a single is about +.05 wins. Given this base-out situation, it’s worth about +.16 wins. Given the two addition inning-score situations, it’s worth +.08 wins or +.29 wins.
Now, leverage. From this table, the LI of situation 1 is 1.0 (strangely enough). The LI of situation 2 is 3.6. WPA/LI for these situations is simply the change in win expectancy divided by the leverage. In situation 1, this is .08/1.0 = .08. In situation 2, this is .29/3.6 = .08.
The fact that those both work out to be +.08 wins was not required — given that they had the same base-out situation was the same, we would have expected them to be similar, but not the same. Notice they are both MORE than the context-neutral .05 value for the single. Why? Because relative to other possible events, a single is better with runners on second and third. Also notice that they are both LESS than the RE24 .16 value of the single. Why? Because the importance of the situation has been removed. But not the context. WPA/LI measures situational hitting. In these situations, a single is relatively more important than a walk, double, etc. WPA/LI figures this out, giving .08 wins for the single instead of the context-neutral .05 wins. But it doesn’t give the hugely high .29 wins — that would be crediting a hitter for the importance of the situation, which WPA/LI is stripping away.
It’s not often addressed, and Fangraphs doesn’t have this stat, but there is also a “leverage” associated with RE24. It would be calculated similarly to win probability leverage, but using the RE24 table, not the WPA table. For 2nd/3rd two outs, maybe it’s 2.0 (making it up, here). If you divide RE24 by this leverage, you’ll get a number like WPA/LI but that does not consider the context of inning/score. With my made up numbers, this “RE24/LEV24” number is .16/2 = .08 (would probably be close to the other .08’s, but doesn’t have to be the same.)
Helpful?
by Sky Kalkman on Jul 16, 2010 9:58 AM EDT up reply actions 1 recs
Very much so
I have it now, thanks for your patience and time.
Follow Me on Twitter @FreeZorilla
by FreeZorilla on Jul 16, 2010 10:19 AM EDT up reply actions
Sweet!
Through all this explaining of WPA/LI, I think I like the stat even more than before.
by Sky Kalkman on Jul 16, 2010 10:25 AM EDT up reply actions
Errr...
Leverage doesn’t just get you from RE24 to WPA, it takes you all the way from context-neutral linear weights to WPA.
Yes, the RE24 table is used to derive WPA; you work backwards from all the possible 9th inning situations. But leverage is calculated after the fact; it’s not calculated first and then used to figure out WPA.
To compute leverage, you basically put yourself in a particular situation and see what each event would do. Single? +.2 WPA. HR? +.4 WPA. Out? .3 WPA. A weighted average of those events would give you +/ 0, by definition. But if you do the weighted standard deviation of those potential outcomes, that’s leverage (well, one way to do it, and then you have to re-scale so 1 is average). In a high-leverage situation, the potential swings are large. In a low-leverage situation, the potential swings are small.
So… first RE24, use that to calculate WP, then use that to calculate LI. To get WPA/LI, you divide each events WPA by its LI and add up those values from every situation.
Actually, that might be another hangup: WPA/LI is calculated for each event, then summed. You don’t take a full season’s WPA and divide by the full season’s average LI.
Also Tango had this curious comment yesterday
Well said any B-R.com corrects that minor issue with RE24 that Fangraphs reports.
by tangotiger on Jul 14, 2010 9:03 AM PDT up reply actions
Any idea what this refers to? I’ve always used FG RE24
Follow Me on Twitter @FreeZorilla
Comments For This Post Are Closed
Secondary Sidebar
User Tools
Please read our Community Guidelines.
FanPosts
Community blog posts and discussion.
Recommended FanPosts
Recent FanPosts
FanShots
Quick hits of video, photos, quotes, chats, links and lists that you find around the web.
Recent FanShots
DRB Suggestion Box
Certain photos copyright © 2012 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
