I need help! A new parameter ?

[Zagi-CRO]

Hello!

Please can you tell me does exists /in sabermetrics/ this pitcher's parameter PR

PR indicates a strong, long-term pitcher's effectivity.


PR = IP / (ERA * 9) while ERA = (9 * ER )/ IP
or
PR= (IP/9)² * (1/ER)


i.e. ---
Justin Verlander DET 186 3,63 5,693
etc....


Thanks on your coments!

[Zagi-CRO]

Tango Tiger, where are you ?

[Tango Tiger]

You don't want to know what I'm thinking.

[Zagi-CRO]

You don't want to know what I'm thinking.

I dont understand you ?

I'll ask somebody's opinion about PR...

[SABR Matt]

The continuous stream of attempts to generate new metrics coming from some folks around the net that have absolutely no logical foundation...no thought process on what is being measured...it astounds me.

Before you create a metric, ask yourself, "What am I trying to measure, and how can I best access that directly through the available data?" Ask youself whether your goals of measurement are specific enough (if your answer to the first question was "how good pitchers are..." then it's not...you need a very specific goal). Then ask yourself whether what you're trying to do has already been done and whether those previous efforts meet with the standards of scientific scrutiny or whether there is room for improvement.

You can't accomplish anything in the analysis of anything...including baseball...by slapping numbers together and hoping they show something. What are you trying to do...how best can you do it...has it been done...was it done well before. That will save us a lot of trouble and internet noise.

[Zagi-CRO]

Thank's on your coments.
I will explain my idea when I would have enough time.

[SABR Matt]

If you have a cogent explanation, I eagerly await it...I might just be missing something obvious that you see...it's happened before.

[Tango Tiger]

I was thinking exactly post #5, but Matt said it much nicer than I was thinking it.

[AstrosFan]

What are the most important new stats to come out in recent years? Any thoughts on this?

[SABR Matt]

I'm partial to my own pitching metric...but that can't really be called important until it finds a larger audience.

Otherwise, the Tango distribution is very important work, as is the leverage index, and all of the recent work on win probability.

Tom Ruane's baserunning analysis is good stuff too. Just off the top of my head.

[Zagi-CRO]

I was thinking exactly post #5, but Matt said it much nicer than I was thinking it.

Ok, I see...
But, I could try to work something, is it forbidden?
For explanation... hm... 'damned English'

On Croatian... puno bi korisnije bilo da objasnim na hrvatskom!

See my answer in topic 'Question Thread'...

[catbox_9]

While we're talking stats...I've often thought of one but it's so basic I'd assume someone came up with it years ago. It's basically the same as slugging percentage but you keep awarding bases if the runner advances due to stolen bases, hits, errors, groundouts, etc. I know there's something like this but it penalizes the runner for a CS. This metric would yield a score of 1.000 for the game in this example:
Batter singles. Advances to second on WP. CS 2-5
Batter K's
Batter is out 5-7-3 (it must be Sean Casey)
Batter reaches on K. Next batter grounds out 6-3 and the next 2 strike out.


That must exist somewhere, doesn't it?

[Zagi-CRO]

... my mistake... sorry

[Zagi-CRO]

PR = IP / (ERA * 9)

ex. IP=233,2 ERA=2,77 PR=9,354

1. Johan Santana MIN 233,2 / 2,77/ ** 9,354
2. Brandon Webb ARI 235,0/ 3,10/ ** 8,423
3. Jonathan Papelbon BOS 68,1/ 0,92/ ** 8,225
4. Roy Oswalt HOU 220,2/ 2,98/ ** 8,210
5. Bronson Arroyo CIN 240,2/ 3,29// ** 8,112
6. Chris Carpenter STL 221,2/ 3,09/ ** 7,954
7. Roy Halladay TOR 220,0/ 3,19/ ** 7,663
8. John Smoltz ATL 232,0/ 3,49/ ** 7,386
9. Carlos Zambrano CHC 214,0/ 3,41/ ** 6,973
10. Aaron Harang CIN 234,1/ 3,76/ ** 6,918
11. John Lackey LAA 217,2/ 3,56/ ** 6,779
12. Chien-Ming Wang NYY 218,0/ 3,63/ ** 6,673
13. Derek Lowe LAD 218,0/ 3,63/ ** 6,673
14. C.C. Sabathia CLE 192,2/ 3,22/ ** 6,632
15. Jason Schmidt LAD 213,1/ 3,59/ ** 6,595

PR supports long-lasting pitching /which is essential for the team/ .
Only Papelbon come on 3rd position because of 0,92 ERA!

[SABR Matt]

Mathematically...

PR = G^2 / ER

IP / 9 = G (G = Full Games) so the IP / (ERA * 9) is reduced to G / ERA. ERA = ER / G so now you have G / (ER / G) which is G^2 / ER.

Obviously, that would favor starters...particularly inning eaters...but so does PRC (Pitcher Runs Created), a Hardball Times invention which is at least somewhat DIPS (defense independent pitching statistics) compliant and is on the batter runs created scale. What you've got here is something that will naturally favor good starters, but not to greater accuracy than pre-existing metrics that have a basis in baseball logic that this PR lacks. Why should pitching skill be related in any way to the games he plays squared?

[Zagi-CRO]

Mathematically...

PR = G^2 / ER

IP / 9 = G (G = Full Games) so the IP / (ERA * 9) is reduced to G / ERA. ERA = ER / G so now you have G / (ER / G) which is G^2 / ER.

Obviously, that would favor starters...particularly inning eaters...but so does PRC (Pitcher Runs Created), a Hardball Times invention which is at least somewhat DIPS (defense independent pitching statistics) compliant and is on the batter runs created scale. What you've got here is something that will naturally favor good starters, but not to greater accuracy than pre-existing metrics that have a basis in baseball logic that this PR lacks. Why should pitching skill be related in any way to the games he plays squared?

it's 5 p.m in Croatia.... a time to write /after work!/

You have right. PR = 1/ER * (IP/9)²

And if add W- win in formula, we have this:

PRW=( IP * W) / (ERA * 162)

[Zagi-CRO]

Results for PRW 2006.:

1. Johan Santana MIN 233,2 / 2,77/ **9,874/ 19
2. Brandon Webb ARI 235,0 / 3,10/ **7,487/ 16
3. Chien-Ming Wang NYY 218,0 / 3,63/ **7,043/ 19
4. Roy Oswalt HOU 220,2 / 2,98/ **6,842/ 15
5. Roy Halladay TOR 220,0 / 3,19/ **6,811/ 16
6. Chris Carpenter STL 221,2 / 3,09/ **6,628/ 15
7. John Smoltz ATL 232,0 / 3,49/ **6,565/ 16
8. Bronson Arroyo CIN 240,2 / 3,29/ **6,309/ 14
9. Carlos Zambrano CHC 214,0 / 3,41/ **6,198/ 16
10. Aaron Harang CIN 234,1 / 3,76/ **6,149/ 16
11. Derek Lowe LAD 218,0 / 3,63/ **5,931/ 16
12. Barry Zito SF 221,0 / 3,83/ **5,699/ 16
13. Kenny Rogers DET 204,0 / 3,84/ **5,575/ 17
14. Justin Verlander DET 186,0 / 3,63/ **5,377/ 17
15. Jon Garland CWS 211,1 / 4,51/ **5,201/ 18

The results seems good --leading Santana and Webb. It's ok.
Papelbon is out form list because of fewer wins:
93. Jonathan Papelbon BOS 68,1/ 0,92/ **1,828/ 4


PRW=( IP * W) / (ERA * 162)
PRW = (IP*W) / ((ER/IP*9)*162)
PRW= ( IP/9)² * W/(ER*162)

What is a purpose of this formula?
To predict CY Award during baseball season!!

[Zagi-CRO]

How can I put tabele in post instead of mishmash with numbers ?

[SABR Matt]

use the code tag..it turns the font into a constant-width font (I believe Courrier New) so you can line things up correctly

[SABR Matt]

As for the newest version of this...if your mission is to predict who the baseball writers will pick as the Cy Young winner...you're using the correct statistics...though I don't see the point in predicting something like that.

[Tango Tiger]

The number you are creating has no meaning on its own. "Santana is 9874!" It means nothing. If you want it to mean somethign, you have to associate it to something else.

For example, Woolner has PAP, which is somethng like (Pitches minus 100)^3.

This really means nothing. It gives you an exponential scale that is off the charts if the pitch count in a game goes high. But, I could do (Hits minus 1)^8. What does that mean? Nothing. It glorifies someone who gets alot of 3 and 4 hit games, but, it's nothing.

But, if you can associate PAP or my crazy contraption here to something else, like future injury or future performance changes, fine. That's good.

But, doing a mish-mash of something that gives you back, as a unit, "innings pitch squared per run allowed".... it's nothing. As is most of the junk stats I've seen.

[SABR Matt]

Ah...that's the element I was missing from my objection earlier Tango...relationships. My own metrics always relate observed statistics to run scoring or the winning of games. Your work has related all kinds of things to all kinds of other things but there is always a motivating, driving purpose behind anything you do. I confess to being somewhat mentally fatigued by the bulk of statistical crysstal meth I've seen around here (home brewed out of poisons and random elements...get it. ). It's not that I want to discourage folks from playing around with the data and learning about baseball. It's not that I'm being some sort of elitist punk...it's just that folks around around the web don't seem to be starting their investigations of baseball by trying to answer real-world questions with real logical thought.

[AstrosFan]

I've always read that the most important element of statistical studies is the ability to predict the future, not rate the past.

[Mariano_Rivera]

Ah...that's the element I was missing from my objection earlier Tango...relationships. My own metrics always relate observed statistics to run scoring or the winning of games. Your work has related all kinds of things to all kinds of other things but there is always a motivating, driving purpose behind anything you do. I confess to being somewhat mentally fatigued by the bulk of statistical crysstal meth I've seen around here (home brewed out of poisons and random elements...get it. ). It's not that I want to discourage folks from playing around with the data and learning about baseball. It's not that I'm being some sort of elitist punk...it's just that folks around around the web don't seem to be starting their investigations of baseball by trying to answer real-world questions with real logical thought.

Personally I think my RCA metric has some meaning

For those few of you that care this is RCA

RCA=(Expected Runs Scored/Expected Innings Pitched)*9

Expected Runs Scored=(SO* -.287)+(IFFB*-.243)+(GB* -.101)+(FB*0.035)+(LD*0.356)+(BB*0.315)+(HBP*0.342) +109

Expected IP=Expected Outs/3

Expected Outs=(SO*1.00)+(BB*0.00)+(HBP*0.00)+(GB*.772)+(LD* .1 88)+(FB*.812)+(IFFB*.962)

[Zagi-CRO]

Finnaly:

PR = IP / (ERA * 9)
---- ERA= ER/IP*9
PR= IP / (ER/IP*9 *9)
PR= IP ²/ ER /9²
PR = 1/ER * (IP/9)²

PRW= IP * W / ERA / 162
PRW = (IP * W) / (ER/IP *9 * 162)
---- 162= 9*9*2
PRW= IP² * W / (ER * 9 * 9 * 9 * 2)
PRW = (IP / 9)² * W / (ER * 18)
PRW = 1/ER * (IP/9)² * W /18
---- PR=1/ER * (IP/9)²
PRW = PR * W / 18
If the player has 18 W then his PRW=PR !!