PDA

View Full Version : Comparing pitcher's value?



spideyo
12-17-2012, 01:31 PM
Is there any website or standard statistic around that ties a pitcher's performance to his contract?

With all the various praises and critiques of our (and really, everyone's) FA signings and trades this winter, I'd really like to be able to track just how much "bang for the buck" we're getting compared to the contracts we didn't sign as the season progresses.

Not being a hardcore sabremetric guy, I have no idea where to even start making such calculations

jharaldson
12-17-2012, 01:59 PM
Fangraphs.com publishes a dollar value which is a multiplier against their WAR. There is disagreement on this site if this is a valid indicator because it has shown guys with 5.00 ERA's to have significant value due to their secondary numbers remaining good but in my opinion it is the best I have seen published.

snepp
12-17-2012, 02:27 PM
The problem with FG's WAR for pitchers is that it's built upon FIP. Guys that historically outperform their FIP get underrated a bit, while guys that regularly underperform get overrated. WAR on baseball-reference is based off a form of ERA, which comes with its own set of problems.

A real rough rule of thumb, average the two together, and use $5 million per WAR.

mike wants wins
12-17-2012, 02:34 PM
Just remember, pre-FA pitchers will cost a lot less than FA pitchers, because their contracts are kept artificially low due to how the labor agreement works.

USAFChief
12-17-2012, 03:08 PM
A real rough rule of thumb, average the two together, and use $5 million per WAR.

Other wise known as a SNERRRT.

Snepp's Real Rough Rule of Thumb.

snepp
12-17-2012, 03:25 PM
I got about 3.5 CAR out of that.



(chuckles above replacement)

USAFChief
12-17-2012, 03:37 PM
I got about 3.5 CAR out of that.



(chuckles above replacement)

Of course I can only respond with the obligatory ... wait for it... bCAR or fCAR?

old nurse
12-17-2012, 04:20 PM
The best way to get a good free agent contract appears to be to have two teams interested in you.

If you strike out batters and don't walk many or give up many home runs you will have a high WAR. Little things like actually winning games means nothing in WAR. Yet if you win a lot of games, you are viewed as valuable yet Jared Weaver has a low WAR despite winning 20 games. (And do you really think a replacement player would have won 17 in his place? Fangraphs had Dempster a 3.3. His contract was nowhere near 16.5. Dan Haren with a 10 year career, averages 3.85 WAR and got 12.5 million. So what pitchers are actually paid is not based on WAR.
What players are paid is of no consequence when your team is winning

mike wants wins
12-17-2012, 05:35 PM
Wins are a team stat....would you rather sign a FA pitcher with a 7 ERA*, who's team scored 8 runs in every start, or one with a 1 ERA, who's teammates were shut out every game? Which one is more likely to help your team in the future?

*yes, I know ERA is not a perfect stat

old nurse
12-17-2012, 09:25 PM
Wins are a team stat....would you rather sign a FA pitcher with a 7 ERA*, who's team scored 8 runs in every start, or one with a 1 ERA, who's teammates were shut out every game? Which one is more likely to help your team in the future?

*yes, I know ERA is not a perfect stat


ERA does not correlate at all well to salary levels. The question of the thread was about metrics correlating to salary. War does not correlate to what actually happened in the season in terms of wins nor does it appear to correlate to salary. IMO there is no metric yet devised to tell you wether or not you got a good deal on a pitcher.

snepp
12-17-2012, 09:39 PM
Teams with players that accumulate more WAR win more games, players that accumulate more WAR end up making more money. The correlations are there, your opinions just won't allow you to see them.

old nurse
12-18-2012, 12:13 AM
Teams with players that accumulate more WAR win more games, players that accumulate more WAR end up making more money. The correlations are there, your opinions just won't allow you to see them.
In regarding correlations one must remember the number of burglaries and ice cream sales correlate. Does the metric measure what it says it does? The accuracy of the number for an individual as well as for a team may be way off. The reliability number for WAR and actual wins by a pitcher is so variable to render it moot. The 4 season long starters for the Angels finished with a record of 54-41 combined war of 6.4. According to WAR, 4 replacement pitchers would finish 48-48 with that team. In a study of replacement level pitchers across the major leagues they found the win percentage of replacement level pitchers to be .430, or 5 wins. As a descriptor of the season, WAR would be significantly off

old nurse
12-18-2012, 12:45 AM
Teams with players that accumulate more WAR win more games, players that accumulate more WAR end up making more money. The correlations are there, your opinions just won't allow you to see them.

2011 class of free agent pitchers Buerle WAR 3.6 14/yr ave CJ Wilson 6.1 15 Edwin Jackson 3.9 11 Kuroda 2.2 and 10 The fifth best starter per espn that pitched last year was Bedard 2.3 4.5. I doubt you could run statistics on that small of sample to come up with a reliability number, but the millions per WAR for the best five last year are variable.

spideyo
12-18-2012, 11:31 AM
Ok...it looks like the first step is finding a good "advanced stat" glossary.

With fangraph's dollar amount...where does the actual dollar amount come from?

old nurse
12-19-2012, 12:19 PM
Ok...it looks like the first step is finding a good "advanced stat" glossary.

With fangraph's dollar amount...where does the actual dollar amount come from?

By comparing dollar ammounts from free agents that switched teams

jay
12-19-2012, 01:18 PM
In regarding correlations one must remember the number of burglaries and ice cream sales correlate. Does the metric measure what it says it does? The accuracy of the number for an individual as well as for a team may be way off. The reliability number for WAR and actual wins by a pitcher is so variable to render it moot. The 4 season long starters for the Angels finished with a record of 54-41 combined war of 6.4. According to WAR, 4 replacement pitchers would finish 48-48 with that team. In a study of replacement level pitchers across the major leagues they found the win percentage of replacement level pitchers to be .430, or 5 wins. As a descriptor of the season, WAR would be significantly off

There are so many incorrect statements in this it's hard to know where to start. You seem to be so very intent on bashing WAR as worthless, when it is largely a very good tool. Those who preach it as the end-all-be-all are equally incorrect, but continuing to cherry pick outliers to make examples is never a valid way to prove or disprove anything.

Wins are widely regarded as an extremely poor stat to measure a pitcher's skill. A "win" for a pitcher includes many things he has no control over -- ie offensive runs scored, the defense behind him, the bullpen that follows, etc -- which WAR intentionally removes to reach a more accurate measure of the pitcher's actual skill by looking at factors a pitcher can control or heavily effect. So, you're correct that WAR commonly won't equal a pitcher's wins, but it will without question provide a better measure of a pitcher's performance than "wins" -- the original question of the post.

We've already reviewed WAR's accuracy over a whole season to a team's actual record (a very strong .83-.89) with you. For the record, replacement level among the different WAR methods equates to between 43 and 52 wins on a full season... not your cited .500 record. It might help to understand how a stat works before you continually spout how ineffective it is.

jay
12-19-2012, 01:30 PM
Ok...it looks like the first step is finding a good "advanced stat" glossary.

With fangraph's dollar amount...where does the actual dollar amount come from?

oldnurse is completely wrong. FG's dollar amount uses the assumption that each win above replacement level is worth a certain dollar value (about $4.5M, I believe) on the free agent market and multiplies the player's WAR by that value. For example, Josh Willingham's 3.9 WAR from 2012 would be "worth" $17.6M.

You have to be careful to use any one year to project a contract value, as it can't account for other factors like regression, injury risk, etc (ie - Willingham had a great year, but he wouldn't get $17M/year on the FA market as most people think he'll regress at least some).

One could theoretically track the free agent deals that a front office signs and compare it back to the FG dollar value to get a rough idea if the contract provided surplus value or not.

old nurse
12-19-2012, 01:42 PM
We've already reviewed WAR's accuracy over a whole season to a team's actual record (a very strong .83-.89) with you. For the record, replacement level among the different WAR methods equates to between 43 and 52 wins on a full season... not your cited .500 record. It might help to understand how a stat works before you continually spout how ineffective it is.

So if your measurement is incorrect 15% of the time that is good?
Where in there does WAR measure the mistake pitches that end up multiple hits in the inning. The inabi;ity to hold a runner on base? The ability to make someone hit weak grounders? How good a pitcher is is again a component of skill sets that correlate with their teammate's skills. What you can do with your skill does depend on the rest of the team. To only look at one component to compare is half blind.

jay
12-19-2012, 03:56 PM
So if your measurement is incorrect 15% of the time that is good?

If you understood correlation ratios, you would know that mid .8 is an exceptionally strong correlation. It measures what it is supposed to measure and then there is a certain amount of variation (about .15 in this case) caused by other factors. If you have a better stat that can project a team's record, I'd be happy to hear about it. I don't see how you can continue to say something is worthless and inaccurate when you don't have anything better.


How good a pitcher is is again a component of skill sets that correlate with their teammate's skills. What you can do with your skill does depend on the rest of the team. To only look at one component to compare is half blind.

I fundamentally disagree with you and I have no idea why you'd want to assess a pitcher's performance (the question in this post) based on his teammates. If David Price takes the mound with Little Leaguers behind him, he has no control over that but his standard stats would be far, far worse... and you certainly can't tell me that means he's a worse pitcher. Many of the advanced stats intend to remove those such things (defense, park factors, league, etc) so that you're comparing pitchers on a even apple to apple level.

old nurse
12-19-2012, 04:42 PM
I fundamentally disagree with you and I have no idea why you'd want to assess a pitcher's performance (the question in this post) based on his teammates. If David Price takes the mound with Little Leaguers behind him, he has no control over that but his standard stats would be far, far worse... and you certainly can't tell me that means he's a worse pitcher. Many of the advanced stats intend to remove those such things (defense, park factors, league, etc) so that you're comparing pitchers on a even apple to apple level.

You don't need much for statistics to understand the value of a David Price nor to understand that Hector Noesi had a bad season. The issue would be assessing the relative values of the middle of the pack. Baseball remaina a team sport. Not many pitchers have every skill. The factors of your home park and defense of the players behid the pitcher would influence a team's opinion of the value of the player. Putting a player in a position to succede will influence the outcomes measured by WAR.
David Price would pitch completely differently to a major league team with major leagures behind him than a bunch of little leaguers. How your team plays behind you will change how you pitch. You cannot isolate that.
The pitcher in WAR is deducted for giving up a home run. Is there a stat that says every home run was a mistake pitch?

ANd yes 85% correlation is good, not excellent. It is leaving a lot of room for error. The variabilty of WAR from year to year is either the player is inconsistent, or the stat is.

jay
12-19-2012, 05:24 PM
You're absolutely right that an individual team should value players slightly differently based on their park and other factor they know (ie strong infield defense), but that's not what WAR tries to do. WAR intentionally removes those factors to provide an unbiased and equally comparable measure of a player's contributions. If you're faulting WAR for not providing 30 different valuations or being able to see inside a GM's head, then you win on that point. You're also right that certain things probably can't be quantified, but I'm not sure how you'd ever prove (let alone measure) an individual player's perception of his defense. We can come up with impossibilities all day long.

Regardless, I still haven't seen you provide something that provides a better measure of a player's (or team's) performance.

old nurse
12-19-2012, 06:28 PM
You're absolutely right that an individual team should value players slightly differently based on their park and other factor they know (ie strong infield defense), but that's not what WAR tries to do. WAR intentionally removes those factors to provide an unbiased and equally comparable measure of a player's contributions. If you're faulting WAR for not providing 30 different valuations or being able to see inside a GM's head, then you win on that point. You're also right that certain things probably can't be quantified, but I'm not sure how you'd ever prove (let alone measure) an individual player's perception of his defense. We can come up with impossibilities all day long.

Regardless, I still haven't seen you provide something that provides a better measure of a player's (or team's) performance.

Easily measured is the abscense of the variability of the batter, park and defense is velocity, movement of the ball and control. Those are the only independent variables that are truly under a pitcher's control. Stikeouts are influenced by the eye and judgement of the batter. While ball 4 may be a mistake on the pitcher's part, the other three may be an attempt to fool the hitter into swinging. The fourth one may as well be also depending on the batter. To propose a relative measure of skill level independent of variables of a skill that is dependent on variables iss a difficult task. For entertainment purposes you can discuss WAR,,,, fip, xfip, but it does not translate to value when you look at the contracts.
The origional question was how do you tell you have good value, or bang for your buck in terms of pitching. In terms of pitching, those that fall into the category of a WAR less than two you have no clue as an independent factor. Does the limitations of the average pitcher fit into your team's composition and ballpark. If as a team it works you got bang for your buck, if it does not work, you did not.

In addition, accumulated WAR for a team correlates to ..83 for a team win total. It does not mean that the individual stat is accurate for the player. WAR is not a predictor of future performance rather a look back on what was. If it only describing with an .83 correlation, then no it is not a good measure.

Top Gun
12-19-2012, 06:41 PM
CBS Sports' Jon Heyman reports that the Cubs have agreed to a deal with Carlos Villanueva.

Terms of the deal are not yet known, but, assuming it's not too out of line, it's another sneaky addition to the rotation by the Cubs. Villanueva was shifted back and forth from starter to reliever the last two seasons by the Blue Jays, but he pitched really well out of the rotation in 2012 before a couple late-season blowups inflated his ERA. All told, he put up a 4.16 ERA and 112/46 K/BB ratio over 125 1/3 innings this past season.
Related: Cubs (http://www.rotoworld.com/teams/clubhouse/mlb/chc/cubs)

Source: Jon Heyman on Twitte (https://twitter.com/JonHeymanCBS/status/281544807850065920)

Maverick
12-19-2012, 07:28 PM
CBS Sports' Jon Heyman reports that the Cubs have agreed to a deal with Carlos Villanueva.

Terms of the deal are not yet known, but, assuming it's not too out of line, it's another sneaky addition to the rotation by the Cubs. Villanueva was shifted back and forth from starter to reliever the last two seasons by the Blue Jays, but he pitched really well out of the rotation in 2012 before a couple late-season blowups inflated his ERA. All told, he put up a 4.16 ERA and 112/46 K/BB ratio over 125 1/3 innings this past season.
Related: Cubs (http://www.rotoworld.com/teams/clubhouse/mlb/chc/cubs)

Source: Jon Heyman on Twitte (https://twitter.com/JonHeymanCBS/status/281544807850065920)

Dude.....spam much...is it really necessary to post this in every topic?

jay
12-19-2012, 09:12 PM
The origional question was how do you tell you have good value, or bang for your buck in terms of pitching.

If it's a free agent deal you're looking to assess, I would in fact say WAR can be used as one measurement through something like FG's dollar values. It's a more accurate assessment of what the player/pitcher contributed on his own. The WAR that a team expects a player to generate should definitely correlate pretty well to what they'd be willing to pay.


WAR is not a predictor of future performance rather a look back on what was.

There is no perfect predictor or magic 8 ball anywhere in this world and WAR doesn't claim to be either. However, it's a better predictor of a player's future performance than things like wins, RBIs, etc.

I still haven't seen you cite a real measurement that exists as a better way to measure a player's value or contributions.

old nurse
12-19-2012, 10:02 PM
I still haven't seen you cite a real measurement that exists as a better way to measure a player's value or contributions.

Please reread my above post as I thought I explained it perfectly clea my position on assessing a pitcher's value with no outside influences.

I notice you quickly dropped the correlation to reality bit with WAR, pun intended.

I went to ESPN. Took the top ranked players from the free agent class from this year that have signed. I divided the average per year from the contract and divided it by Fangraph's WAR value for last year. Nothing stands out as a developing pattern. Yes the sample size is small enough to not be significant yet Upton 4.5 mil per War , Hamilton's 5.7 Grienke 4.8, Kuroda 3.8 and Haren 7.2 It would lead one to believe there would be no pattern.

glunn
12-20-2012, 01:22 AM
I just stumbled on this site (http://baseballplayersalaries.com/) where you can type in a pitcher's name and get all sorts of analysis about what he is worth.

jay
12-20-2012, 07:56 AM
Yes the sample size is small enough to not be significant yet Upton 4.5 mil per War , Hamilton's 5.7 Grienke 4.8, Kuroda 3.8 and Haren 7.2 It would lead one to believe there would be no pattern.

A free agent deal should be based off 'expected WAR' going forward. So, of course you'll see TONS of variation based on what different teams expect a player to do going forward. You also see other factors affect how much a team is willing to pay for a free agent or give up in a trade (see: Blue Jays, Toronto and Dickey, R.A.)... which makes sense since going from 86 wins to 90 wins is incredibly more valuable than something like 66 to 70. You literally can't come up with a perfectly correlated pattern since there literally isn't one. WAR (and FG dollars) at least provides a baseline designed to equally measure all players as best possible.

I agree with you that velocity, movement, and control would be great things to further assess a pitcher, but the point is you haven't named a stat that exists today that can be used as a better alternative to WAR.

old nurse
12-20-2012, 08:06 AM
A free agent deal should be based off 'expected WAR' going forward. So, of course you'll see TONS of variation based on what different teams expect a player to do going forward. You also see other factors affect how much a team is willing to pay for a free agent or give up in a trade (see: Blue Jays, Toronto and Dickey, R.A.)... which makes sense since going from 86 wins to 90 wins is incredibly more valuable than something like 66 to 70. You literally can't come up with a perfectly correlated pattern since there literally isn't one. WAR (and FG dollars) at least provides a baseline designed to equally measure all players as best possible.

I agree with you that velocity, movement, and control would be great things to further assess a pitcher, but the point is you haven't named a stat that exists today that can be used as a better alternative to WAR.

Where do I find the stat expected WAR? If that is what the basis for payment is than Spideyo has no shot at finding out if his team has gotten a good deal based on current WAR. Thank you for proving my point.