Closer - Reliever Evaluation

[Zagi-CRO]

It's interesting see who are the best closers now /based on simple math/:

 [b]
AL Closers[/b]
0.42	Joakim Soria
0.32	George Sherrill
0.29	Mariano Rivera
0.15	Troy Percival
0.13	Miguel Batista

[b]
NL Closers[/b]
0.19	Jason Isringhausen
0.17	Takashi Saito
0.03	Francisco Cordero
0.02	Rudy Seanez
0.01	Brian Shouse

[skyking162]

It's interesting see who are the best closers now /based on simple math/:

 [b]
AL Closers[/b]
0.42	Joakim Soria
0.32	George Sherrill
0.29	Mariano Rivera
0.15	Troy Percival
0.13	Miguel Batista

[b]
NL Closers[/b]
0.19	Jason Isringhausen
0.17	Takashi Saito
0.03	Francisco Cordero
0.02	Rudy Seanez
0.01	Brian Shouse

It might be simple, but we don't know if it's sounds unless you explain your methods. Then we can disagree and you can improve your calculations.

[RubeBaker]

Please don't be like ESPN and try to use a single formula to grade who's best at what. It's never as simple a single formula or statistic. But I'm with SkyKing here, please show your work to receive full credit.

[SamtheBravesFan]

My guess is that his stats are weighted towards runs and hits allowed and strikeouts.

[Zagi-CRO]

Ok, ok... a math is quite simple.

For the closers:
MVCl=(20*SV +SO+IP - 60*(WHIP +ERA))/162

0.42	Joakim Soria
0.39	Jason Isringhausen
0.32	George Sherrill
0.29	Mariano Rivera
0.17	Takashi Saito

The same case with 162 /like the batter's math/...

Soria is the leader.
Explanation:
Soria has 4 SV /only Isringshausen has 5/, he has the most SO /7/, an ERA 0.00, WHIP 0.40, IP=5 /also a very good/.




For the relievers:
MVRl=(2.75*(SO+IP) - 25* (WHIP +ERA))/81

0.15	Dan Wheeler
0.15	Matt Albers
0.15	John Grabow
0.13	Santiago Casilla
0.12	Brian Bruney

Explanation:
They all have an ERA=0.00, WHIP about 0.50, the most IP and SO /except Wheeler/.

SBF - I accept your suggestion about using runs allowed in the math. How? Have you an idea?

I know that very simple but I could not find anything better!!

[Zagi-CRO]

My guess is that his stats are weighted towards runs and hits allowed and strikeouts.

It's a good idea! Runs and hits allowed... like regressive factor.

[Zagi-CRO]

I made comparison between the results for MVCl with R and without R:

MVCl=(20*SV + SO+IP -R - 60*(WHIP +ERA))/162

          includ. R   without R
Jose Valverde	5.15	5.28
Takashi Saito	4.85	4.91
Fran Cordero	4.70	4.84
Jona Papelbon	4.40	4.47
Joe Nathan	4.32	4.41
Fran Rodriguez	4.27	4.40
Bobby Jenks	4.21	4.33
Trevor Hoffman	4.16	4.29
Joe Borowski	3.67	3.91
Billy Wagner	3.58	3.72

[SamtheBravesFan]

I think you should remove saves from your formula. The only thing that signifies, basically, is that a reliever finished a game.

[RuthMayBond]

I think you should remove saves from your formula. The only thing that signifies, basically, is that a reliever finished a game.

One save definitely should NOT carry TWENTY times the weight of an inning pitched

[SamtheBravesFan]

One save definitely should NOT carry TWENTY times the weight of an inning pitched

... I didn't even realize that, but you're right! The only thing that registered to me is "he's weighting a save 20 times!". I didn't realize he wasn't weighting innings pitched at all.

Zagi, again, I suggest taking out saves all together. Sometimes the best relievers aren't closers. Just because a reliever finishes a game with a three run lead or less doesn't mean they should get any special reward for it.

[Zagi-CRO]

... I didn't even realize that, but you're right! The only thing that registered to me is "he's weighting a save 20 times!". I didn't realize he wasn't weighting innings pitched at all.

Zagi, again, I suggest taking out saves all together. Sometimes the best relievers aren't closers. Just because a reliever finishes a game with a three run lead or less doesn't mean they should get any special reward for it.

SBF, I have two formulas, one for the relievers and one for the closers.
Why? The best closers are those which have the most saves, is it right?

Again, only for the closers---> This is my math for the beginning

MVCl=(20*SV + SO+IP - RA - 60*(WHIP +ERA))/162

** Final result -MVCl must be in the range 0.00 - 6.00 /the worst to the best/

My questions:

SV - weighting a save 10 times - Ok
SO - weight /1 time - Ok
IP - weight / 1 time - Ok
RunsAllowed - weighting how much times ??
(WHIP+ERA) - medium value about 3.00 - weighting how much times ??

Please, say your opinion! Txs.

[SamtheBravesFan]

That's just it. It's the fact that you're even considering saves at all. That's not a good idea. Even if you have seperate formulas, that just means that they finished the game. Sometimes, closers finish the game and they don't get saves because of the score. Why are you going to weigh that 20 times against anything else?

[Zagi-CRO]

That's just it. It's the fact that you're even considering saves at all. That's not a good idea. Even if you have seperate formulas, that just means that they finished the game. Sometimes, closers finish the game and they don't get saves because of the score. Why are you going to weigh that 20 times against anything else?

I was thinking about one math for the relievers, ok.
So, I accepted your suggestion, without SV.

It could be only:

1st math
MVR= (SO+IP - 15*ERA)/IP*4.5

So, the final result should be at range 0-10 and minimal required innings= 30 inn for the season.

When I applied the math for 2007 season:

                  MVR    ERA    SO      IP
Carlos Marmol	9.35	1.43	96	69.1
Jon Papelbon	8.86	1.85	84	58.1
Takashi Saito	8.50	1.40	78	64.1
Heath Bell	7.96	2.02	102	93.2
Raf Betancourt	7.80	1.47	80	79.1
Fran Rodriguez	7.71	2.81	90	67.1
Jon Broxton	7.59	2.85	99	82
Joe Nathan	7.58	1.88	77	71.2
Juan Cruz	7.49	3.10	87	61
Fran Cordero	7.45	2.98	86	63.1

Now, the only question is which weight give for SO, IP and ERA?

Ex. for some reliever with a very good stat:
SO = 90, IP=70, ERA=2.00
SO weight=90
IP weight=70
ERA weight=15*2= 30

This indicates that SO has 3 times higher weight then ERA!?

MVR=(90+70-15*2.00)/70*4.5 = 130/70*4.5 = 8.36

or with double weight for an ERA --> ERA weight= 30*2.00= 60
It's better, I think.

Now we have:

MVR=(SO+IP - 30*ERA)/IP*4.5
MVR=(90+70 - 60)/70*4.5
MVR= 100/70*4.5 = 6.42
So, it's low mark if we expected mark about 8->10.

----------------------------------------------------
Correction:
2nd math
MVR=(SO+IP - 30*ERA)/IP*5

So, with this math applying on 2007 stats:

Carlos Marmol	         8.84
Takashi Saito	         7.81
Jonath Papelbon          7.45
Raf Betancourt           7.27
Heath Bell	         7.22
Joe Nathan	         6.45
Jon Broxton              5.82
Rafael Perez	         5.71
Peter Moylan	         5.50
Francis Rodriguez        5.42

SBF, what do you think which math /the first one or the second one/ is better?

[SamtheBravesFan]

I think I like the first forumla the most. It takes into account who goes out there and pitches a lot and who produces. I like the ratio on the ERA too. This is much better, I think.

[Zagi-CRO]

Apr/29 Reliever Power Rating by Zagi

AL
Santi Casilla	10.7
Joakim Soria	9.5
Troy Percival	9.0
Mariano Rivera	9.0
Andrew Brown	7.2
Dennys Reyes	6.8
NL
Brad Lidge	9.4
Billy Wagner	9.0
Chad Qualls	8.9
John Grabow	8.6
J.C. Romero	8.6
Jorge Campillo	4.2

MVR= (SO+IP - 15*ERA)/IP*4.5