How does one get a negative value for OPS+? If it's a ratio, shouldn't it always be a positive number, stopping at zero when you have no offensive value at all?
The negative values all go something like this: In 1955 Sandy Koufax came to bat 12 times, and struck out 12 times. His OPS+ for that year was -100.
Over his career, his OPS+ was -26.
Because the formula for OPS+ is OBP/LGOBP + SLG/LGSLG + -1Originally Posted by Gee Walker
"he probably used some performance enhancing drugs so he could do a better job on his report...i hear they make you gain weight" - Dr. Zizmor
"I thought it was interesting and yes a conversation piece. Next time I post a similar story I will close with the question "So, do you think either of them have used steroids?" so that I can make the topic truly relevant to discussions about today's game." - Eric Davis
http://www.youtube.com/watch?v=gqul1GyK7-g
This is the correct formula:
OPS+ = 100 * (OBP/lgOBP* + SLG/lgSLG* - 1)
It is a ratio of sorts measured against the league average. An OPS+ value of 100 is "average". Multiplying by 100 gives us a whole number to work with.
Ehh...you say TO-MAY-TO I say TO_MAHH-TO
"he probably used some performance enhancing drugs so he could do a better job on his report...i hear they make you gain weight" - Dr. Zizmor
"I thought it was interesting and yes a conversation piece. Next time I post a similar story I will close with the question "So, do you think either of them have used steroids?" so that I can make the topic truly relevant to discussions about today's game." - Eric Davis
http://www.youtube.com/watch?v=gqul1GyK7-g
And that right there is why OPS+ is stupid...it makes no logical sense to add the ratio componants like that...people use it as though a 75 OPS+ bat is 75% as talented as a 100 OPS+ bat and that simply isn't triue.
Registered User
It is true that OPS+ is a misleading name for the stat, because when you see the X+, you are conditioned to expect a straight ratio.
However, granting that, if you figured OPS+ as the straight ratio, it would make even less sense to claim that a 75 batter was "75% as talented" as a 100 batter. This is because OPS+ has approximately a 1:1 correlation with runs per out while OPS does not. Ratios of OPS do not give you any meaningful estimate baseball value, whereas OPS+ can be viewed as estimated runs created per out as a ratio of the league R/out.
Last edited by Patriot; 08-16-2007 at 07:50 AM. Reason: Said "PA" when I meant "out"
Whenever I see a +-1 in a formula it seems contrived. Wouldn't it be more informative to just list a players % or ratio above league in OBP and SPCT individually.
Thanks for the clarification everyone, but now that I know the formula, I have a lot less respect for the stat. I guess it's like finding out how sausage is made.
Since OPS includes SLG, the formula expands to:
OPS+ = 100*[(OBP+SLG)/(LGOBP+LGSLG) + (SLG)/(LGSLG) - 1]
In effect, the formula rewards SLG twice but OBP only once. Is there any good reason for this?
Now I know why a high OBP, low SLG guy like Luke Appling only ends up with a career OPS+ of 112.
No. You've confused an OBP for an OPS. SLG is only included in one term, as is OBP.
I agree with Patriot -- the addition of the two ratio terms is a totally separate issue from misinterpreting the difference between a 100 and 75 OPS+. ERA+ suffers from the second potential mistake, but not the first flaw.
That being said, why *is* OPS+ not just player OPS divided by adjusted league OPS, times 100? The current way gives more credit to players who are extra good in either OBP or SLG versus someone who's more balanced, given the same OPS.
Registered User
The biggest reason is that it doesn't correlate as well with runs scored. The real effect of OPS+ as currently constructed is not to increase the weight on OBA, which is why it correlates better with runs scored. As Tango and others have shown, the best correlation with runs comes from 1.8OBA + SLG, or something in that range. OPS+ is implicitly making it something like 1.2OBA + SLG, since in a typical league you might have a SLG of .400 and an OBA of .330. The league ratio of SLG to OBA is then 1.21, and by treating the two ratios as equals you are essentially comparing 1.2*OBA + SLG to the league.
There are two issues, I think. One is the misleading nature of having "OPS+" not be simply OPS/LgOPS as other "X+" stat are. This is a valid complaint about the naming convention used. But the other issue is which stat tells you more about a player's run contribution, OPS/LgOPS or OPS+. And the answer to that is OPS+.
Why Pete Palmer chose to call this PRO+ (at the time he was calling OPS "PRO", for Production), I can't pretend to know. It's still confusing people 20 years later. But he was right that it is a more telling stat then OPS/LgOPS.
Registered User
Walker: you are reading the formula wrong. It's OBP (not OPS) per league average, plus SLG per league average.
***
As Patriot points out, the (OBP+) plus (SLG+) minus 1 does "work", in that it has a roughly 1 to 1 correspondance to runs.
For example:
http://www.tangotiger.net/markov.html
Click CALCULATE, and you get:
4.957 : Runs Scored per Game
AVG / OBP / SLG: 0.270 / 0.341 / 0.405
Click the BACK button, and enter "35" for AB, and you get:
5.624 : Runs Scored per Game
AVG / OBP / SLG: 0.286 / 0.359 / 0.429
5.624/4.957 = 113
And: 359/341 + 429/405 - 1, all times 100 = 111
At the extremes, it won't work quite as well.
On a technical level, it is silly, since you can get a situation where a player with a higher OPS will have a lower OPS+ than someone else. For example, a .380/.380 player and a .280/.480 player (both 760 OPS in a league of 760 OPS, .340/.420), would come in at 102 and 97, respectively. On the plus side to that is that is that it gives more weight to the OBP, which it should. So, it "works" in spite of itself!
OPS+ has proven it self over the years to be a valid comparative tool. Yes, it is ragged around the edges, but it still correlates well to actual run production. The correlation is what's important not the logic or arithmetic behind the number.
In a common time frame it is no more informative than OPS, but when comparing players from different time frames you need this tool to indicate how the players performed against their peers in a relative sense. Otherwise you would conclude that a guy with a .775 OPS during the Dead Ball Era was not as good as a guy with a .800 OPS during the 1990's.
Buck O'Neil: The Monarch of Baseball
I get where you're coming from with this, but I disagree. You need to start with logic, then apply more logic until you're within an acceptable error range. OPS/lgOPS is less accurate than OPS+, but it has a better foundation. If you take it a step further and do (1.8*OBP + SLG)/(1.8*lgOBP + SLG), I'm sure you get something better than both OPS/lgOPS and OPS+.
When you get close by accident, the method actually does have an underlying logic you can discover, or else it's a fluke and not helpful in going any further.
Registered User
Correlation is not causation. Furthermore, a correlation only gives you the relationship of the existing data. It doesn't necessarily let you extend that relationship to data outside your sample, nor does it necessarily hold for data at the extremes of the sample.
Logic however is logic.
Correlation is a starting point, but one that is trumped by other more logical processes.
Registered User
I don't want to belabor an unimportant poimt, but I don't agree that OPS/lgOPS has a better foundation then OPS+. I agree that the construction is more appropriate for the name, but OPS itself is used for the same reason OPS+ is--"because it works". There is no more solid theoretical foundation for adding OBA and SLG then there is for adding relative OBA and relative SLG.
Loyal Indians Fan
Wow. I'd always assumed there was more to OPS+ than that...what a dumb stat. I no longer respect it.
CLEVELAND INDIANS Central Division Champions
1920 1948 1954 1995 1996 1997 1998 1999 2001 2007
So how do we interpret OPS+ if it's inaccurate to say that if 100=lgAVG then 115 = 15% above lgAVG and 75% = 25% below lgAVG?
Registered User
You can still say (*) that a guy with an OPS+ of 115 is 15% better than average; just recognize that the 15% is on a scale of runs created per out, not OPS.
(*) Recognizing of course that OPS+ is not an ideal model of offense, and accepting all of the caveats that go along with its use.
No kiddin'. And there are some who use OPS+ as the backbone for comparing players...adding and subtracting points here and there, for this or that. Silly.
"Baseball brains are not put into everyone’s head. Babe Ruth…had baseball brains…" - Eddie Collins
"Ruth was great too, but he was different. Totally different – easygoing, friendly. There was only one Babe Ruth. He went on the ball field like he was playing in a cow pasture, with cows for an audience. He never knew what fear or nervousness was. He played by instinct, sheer instinct." - Rube Bressler
"In the matter of runs, Cobb was a retailer, Ruth a wholesaler." - Fred Lieb
Sultan,
Re-read Patriot's comment above yours. Why does that change your view of the stat so much? Two different scales.
THE REVOLUTION WILL NOT COME WITH A SCORECARD
In the avy: AZ - Doe or Die
Registered User
Unfortunately, the small print is ignored. OPS+ goes from a quick and dirty measure that has riders attached to it to the ideal backbone to evaluate all kinds of players, however extreme.
People don't use the riders. And, it is very easy to improve on OPS+ by simply doing away with it, and have a RC+ (with LWTS or BsR being the backbone of RC).
As it is, people want Sean Forman to do all the work for them, and will quote OPS+ because he calculated it for them.
Posting Permissions
View Profile
View Forum Posts
View Blog Entries
View Articles
Reply With Quote
Visit Homepage
