Good Thread Idea on the Sun Deck--Clutch Project

[mth123]

Originally Posted by Spring~Fields

Adam Dunn	 1.47
Jerry Hairston	 1.15
Jay Bruce	 0.87
Edwin Encarnaci	 0.71
Joey Votto	 0.57
Brandon Phillips 0.25
Andy Phillips	 0.11
Jolbert Cabrera	 0.10
Ryan Freel	 0.10
Scott Hatteberg	 0.06
Paul Janish	 0.05


Jeff Keppinger	 -0.04
Juan Castro	 -0.13 
Dave Ross	 -0.14 
Norris Hopper	 -0.36 
Javier Valentin	 -0.39 
Ken Griffey Jr.	 -0.39 
Paul Bako	 -0.65 
Corey Patterson	 -1.64

Interesting numbers. I'm not real up on the current thinking on WPA. At first glance it kind of reinforces what I've already believed. This team has (if it keeps Dunn) most of the pieces in place and simply needs to limit the playing time of the players who drag it down. As this list shows, this year those guys are Patterson, Griffey and the catchers. I'm a little surprised at the numbers for Keppinger (I thougt it would be higher) and Janish (I thought it would be lower).

Griffey's numbers are interesting. He has been better lately. My theory is that the DH stint on the AL trip rejuvinated his legs a bit which is showing up as better range and increased power. I wonder if it will last now that he is back out in the field (I'm guessing it just takes too much out of his legs which has been the source of his success on both offense and defense IMO). I'd think based on the recent results that Griffey could do well as a DH for a couple more years. If he could DH, by season's end he may be near the top of this list. I hope he and any potential AL trading partners realize that.

Some questions for the stat gurus. Is the stat accepted as valid? Does defense factor in? Does anyone have the numbers for pitchers this year?

[jojo]

Originally Posted by SteelSD

Only if beating said horse is somehow a celebration of its death.

What do you have against WPA?

[Spring~Fields]

Originally Posted by mth123

Interesting numbers. I'm not real up on the current thinking on WPA. At first glance it kind of reinforces what I've already believed. This team has (if it keeps Dunn) most of the pieces in place and simply needs to limit the playing time of the players who drag it down. As this list shows, this year those guys are Patterson, Griffey and the catchers. I'm a little surprised at the numbers for Keppinger (I thougt it would be higher) and Janish (I thought it would be lower).

Griffey's numbers are interesting. He has been better lately. My theory is that the DH stint on the AL trip rejuvinated his legs a bit which is showing up as better range and increased power. I wonder if it will last now that he is back out in the field (I'm guessing it just takes too much out of his legs which has been the source of his success on both offense and defense IMO). I'd think based on the recent results that Griffey could do well as a DH for a couple more years. If he could DH, by season's end he may be near the top of this list. I hope he and any potential AL trading partners realize that.

Some questions for the stat gurus. Is the stat accepted as valid? Does defense factor in? Does anyone have the numbers for pitchers this year?

I was hoping that maybe JoJo could help us out.

I am wondering what happens to the offense if one decideds to put the negative number players in high PA spots too often?

I was wondering if Keppinger was an input error or typo myself.

[oneupper]

Originally Posted by mth123

Some questions for the stat gurus. Is the stat accepted as valid? Does defense factor in? Does anyone have the numbers for pitchers this year?

I'm not a stat guru, but I find WPA to be very entertaining and couple of years ago I contacted Dave Studeman from the Hardball Times to see if he could process the REDS games and post the results here in REDSZONE. It worked for about two weeks, but it was too much work for Dave, so we called it off.

Here's an article he wrote in the HBT, which explains a bit more about WPA:

http://www.hardballtimes.com/main/ar...n-probability/

The stat is tossed around, but I wouldn't say it is general "accepted", because as far as I know, there hasn't been an agreement on a standard table of Win Probability per game situation.

There is historical data...that is, how many times a team has won starting from a certain situation. Tangotiger has published it here:

http://winexp.walkoffbalk.com/expectancy/search

It's fun to see, in a certain moment of a game, how likely a comeback is.

But some situations have small sample sizes, so generally a standarized table is used. Making a standarized table involves assumptions, such as average runs per game, home field advantage (or not),park factors...etc. There is no consensus on those assumptions, so the tables will vary depending on who calculates it.
But it shouldn't vary THAT much from one source to another. This is a counting stat and the player either performed (and gained WPA) or did not (and lost WPA).

As far as I have seen the fangraph.com calculation is a very simple one. Defense is NOT factored in, as in a player will get credit when reaching on a error (which probably isn't fair). When Dave was doing the REDS games, he would distribute WPA from a non-routine play between the defense and the pitcher, to make it fairer. But this was subjective and involved much more work on his part. You could get an idea of how much defense was helping or hurting the team.

Eliminating defense may not be as "accurate" but it's objective. Player hit ball..good things happened...he gets credit. And its a LOT easier. Fangraphs.com tracks WPA IN REAL TIME in these cute little graphs which I in particular enjoy very much.

http://www.fangraphs.com/scoreboard....ate=2008-07-05

The pitchers WPA is also tracked. Its just the opposite of the hitters WPA.
As can be expected, late inning relievers can generate a lot WPA, and the stat in several forms and variation is used to track relievers peformance.

The same page that has hitter's performance has the pitcher's WPA

http://www.fangraphs.com/scoreboard....ate=2008-07-05

Volquez leads the starters and Cordero the relievers. No surprise there.
There are other interesting stats on the page, such as pLI...which shows how "leveraged" (critical) the situations for the players were.
(high for a reliever, low for a starter, high for PH hitter, for example).

enough for now...

[SteelSD]

Originally Posted by jojo

What do you have against WPA?

As long as no one's attempting to tie WPA to player performance, nothing.

[Spring~Fields]

Originally Posted by SteelSD

As long as no one's attempting to tie WPA to player performance, nothing.

How about if they try to tie it to management performance?

[oneupper]

Originally Posted by Spring~Fields

I am wondering what happens to the offense if one decideds to put the negative number players in high PA spots too often?

I was wondering if Keppinger was an input error or typo myself.

WPA isn't really predictive nor does it try to be. A few "clutch" games and a player's WPA can vary quite a bit. Brandon Phillips, for example, had accumulated a LOT of negative WPA before he went on his recent hot streak.

That said, better hitters tend to have better WPAs over the course of a season.

Kepp is not an error...he's been neutral WPA this year. GIDPs KILL. But as a reference, Alex Gonazalez was -1.07 in 2007. Not making outs is important.

As for high PA spots...OBP is a better. Use that.

And since this is WIN probability, a winning team will have an overall positive WPA, while losing teams...well...they have to distribute the negative.

WPA shows how a player PERFORMED situation-wise during a certain period.
It says if that player WAS clutch, but you can't really use it to determine if a player IS clutch (as in...performs better under pressure situations).

[Spring~Fields]

Originally Posted by oneupper

WPA isn't really predictive nor does it try to be. A few "clutch" games and a player's WPA can vary quite a bit. Brandon Phillips, for example, had accumulated a LOT of negative WPA before he went on his recent hot streak.

That said, better hitters tend to have better WPAs over the course of a season.

Kepp is not an error...he's been neutral WPA this year. GIDPs KILL. But as a reference, Alex Gonazalez was -1.07 in 2007. Not making outs is important.

As for high PA spots...OBP is a better. Use that.

And since this is WIN probability, a winning team will have an overall positive WPA, while losing teams...well...they have to distribute the negative.

WPA shows how a player PERFORMED situation-wise during a certain period.
It says if that player WAS clutch, but you can't really use it to determine if a player IS clutch (as in...performs better under pressure situations).

That is truly very interesting and helpful, thank you.

I thought that I saw a correlation between WPA and OBP. They will predict a fequency won't they? Or probability? Of course not always, in a game of failures.

[jojo]

Originally Posted by SteelSD

As long as no one's attempting to tie WPA to player performance, nothing.

And if they did, why would it bother you?

[SteelSD]

Originally Posted by jojo

And if they did, why would it bother you?

Because it's a junk metric that attempts to assign situational importance to an event while disregarding that a baseball game isn't played in a linear fashion.

[jojo]

Originally Posted by SteelSD

Because it's a junk metric that attempts to assign situational importance to an event while disregarding that a baseball game isn't played in a linear fashion.

Basically it just looks at historic play by play data to assign an average win probability to a specific event in a specific context-i.e. a single with bases loaded in the bottom of the eighth while tied increases the team's chances of winning on average over all similar situations in the last gazillion games by X therefore the average value of that single in "wins" is likely "X".

It actually doesn't require that a baseball game be played in a "linear fashion". Since it's largely derived from the application of Markov chains to the play by play data, the sequence of events that lead to a particular state is in fact wholly unimportant.

In other words, the focus is strictly upon the context of the event and the win value associated with the probability of the outcome (i.e. what state you are in, what potential states you can transition into, and the likelihood of each of those possible transitions).

It's really just leveraging the mountains of available play by play data by applying basic probability theory to generate a win expectancy matrix which can be tweaked depending upon the specific players involved (i.e. the probabilities change depending upon whether Coffey is pitching with bases loaded in the 9th or Cordero is on the mound).

[SteelSD]

Originally Posted by jojo

Basically it just looks at historic play by play data to assign an average win probability to a specific event in a specific context-i.e. a single with bases loaded in the bottom of the eighth while tied increases the team's chances of winning on average over all similar situations in the last gazillion games by X therefore the average value of that single in "wins" is likely "X".

It actually doesn't require that a baseball game be played in a "linear fashion". Since it's largely derived from the application of Markov chains to the play by play data, the sequence of events that lead to a particular state is in fact wholly unimportant.

In other words, the focus is strictly upon the context of the event and the win value associated with the probability of the outcome (i.e. what state you are in, what potential states you can transition into, and the likelihood of each of those possible transitions).

It's really just leveraging the mountains of available play by play data by applying basic probability theory to generate a win expectancy matrix which can be tweaked depending upon the specific players involved (i.e. the probabilities change depending upon whether Coffey is pitching with bases loaded in the 9th or Cordero is on the mound).

jojo, I understand WPA quite well. I was here to debunk it the first time Studeman tried to push it on us as something more than a random assignment of game-state event "importance". It's a junk metric that tells us nothing about actual player value.

[RedsManRick]

I must say that I don't put any value in WPA either. I refuse to believe that a HR in the 1st inning should be worth possibly 10x less than a HR in the bottom of the 9th when the exact same number of runs are produced. That 9th inning homer didn't contribute any more to the win. The order and timing in which runs score is basically irrelevant, so long as the totals are what they are. Put another way, the value of the 9th inning HR is built on the events that led up to it -- yet the interaction effect is not distributed across the events, only assigned to the most recent one, making nearly every single plate appearance of increasing value.

It's like BA w/ RISP. Yes, it measures something and that something is sort of relevant, but I just don't see its utility. It doesn't tell me anything that increases my understanding of player value, past performance, or likely future performance.