OPS - statistical discussion

[Kc61]

Risking a controversial subject here, but I'd like to learn more and see some views on this subject.

IMO, the OPS stat masks the real problem with offenses like the Reds'. A number of important, starting players get a pass because their OPS is acceptable or good. While a good shorthand, OPS can be very misleading and requires deeper analysis.

The slugging half of OPS usually is the dominant half numerically. Slugging percentages generally are higher than OBPs. When a player has a high slugging percentage, his OPS tends to be good. Yet, many such players are low BA and low OBP hitters.

So, on the Reds, Jay Bruce has a .326 OBP and a .481 SLG. Pretty good OPS of .806, but he doesn't get on base much and his BA is .243.

Ryan Ludwick has a .329 OBP and a .535 SLG. Huge differential there. His good OPS of .864 is heavily weighted to slugging, not getting on base.

Frazier is similar, .323/.520. Cozart .294/.407.

Here's my point for discussion. I think that simply saying that guys like Ludwick and Bruce have a good OPS may give them too much credit because a team full of power hitters who don't get singles and don't get on base is often sub-par offensively. (Reds currently have 583 singles, while the Cards have 721.)

Hope the issue I'm presenting for discussion is clear. Fire away.

[westofyou]

Lot's of #5 hitters on this team this year.

This offense needs some BA driven hitters to move the offensive game around more.

[powersackers]

Both are equal parts of the equation. 1 plus 1 equals 2. A balanced lineup should reflect that. Reds just don't have a balanced lineup.

[Kc61]

Both are equal parts of the equation. 1 plus 1 equals 2. A balanced lineup should reflect that. Reds just don't have a balanced lineup.

Actually, I've been reading about this and OBP seems to correlate more to run production than SLG. Yet SLG is more dominant in the OPS formula.

I guess stats like wOBA try to weigh these factors more accurately to give a better reflection of actual performance.

Still, we all tend to use OPS (myself included), it's a helpful shorthand, but I think it can hide a player's real strengths or weaknesses.

[bucksfan2]

Lot's of #5 hitters on this team this year.

This offense needs some BA driven hitters to move the offensive game around more.

Team has too many players that are above average sluggers for their position but below average at getting on base. The problem is they don't really have anyone coming off the bench who does anything different. Both Frazier and Heisey are very similar to their starting counterparts.

[RedsManRick]

You're right, KC. From the perspective of scoring runs, OBP is worth about 70% more than SLG. So OPS puts too much emphasis on SLG and not enough on OBP (to say nothing of the silliness of adding two fractions with different denominators).

OPS is basically no longer used in sabermetric circles. It has largely been replaced by wOBA, weighted On Base Average. Basically, it's OPS with better inputs placed on the OBP scale. There's really no reason NOT to use it.

It's explained clearly, in depth here: http://www.fangraphs.com/library/ind.../offense/woba/

One common complaint is that you can't calculate it yourself, which is basically true. But you can only calculate OPS is you have OBP and SLG already. And if you're on the internet, just use Fangaphs and it's there.

Some might complain that they won't know what "good" looks like, but people had the same problem with OPS until they started using it regularly. And unlike OPS, it's not a new scale -- it's like OBP: .300 is bad, .320 is poor, .340 is solid, etc.

And as others have said, the team is suffering from a lineup full of guys built to be RBI guys without having enough guys to be the runs being driven in. Generally speaking, I'm not a fan of building a team based around a fixed set of runs, but a lot of this team's slugging ability is being wasted due to the lack of baserunners.

The Reds offense is 9th in the NL AVG, 9th in OBP, 5th in SLG -- and 8th in runs scored.

If you could make the effort to start using wOBA instead of SLG (again, it's just as easy to look up as OPS is), I'm sure that would be helpful in setting the trend around here.

I can't strongly enough recommend "The BOOK" by Tom Tango and company.

http://www.insidethebook.com/woba.shtml

Even a read through the chapter previews available on his website can be illuminating.

[Kc61]

If you could make the effort to start using wOBA instead of SLG (again, it's just as easy to look up as OPS is), I'm sure that would be helpful in setting the trend around here.

I can't strongly enough recommend "The BOOK" by Tom Tango and company.

http://www.insidethebook.com/woba.shtml

Even a read through the chapter previews available on his website can be illuminating.

My understanding from the chapter is that wOBA essentially replaces OPS. It uses an OBP-like scale, but considers both on base performance and power in a more sophisticated and accurately weighted manner than OPS. Is that a fair description?

I guess OPS has one advantage - you can easily break it apart into the OBP and the SLG aspects. I assume you can't do that with wOBA.

P.S., thanks for the explanation RMR, it's extremely helpful.

[RedsManRick]

My understanding from the chapter is that wOBA essentially replaces OPS. It uses an OBP-like scale, but considers both on base performance and power in a more sophisticated and accurately weighted manner than OPS. Is that a fair description?

I guess OPS has one advantage - you can easily break it apart into the OBP and the SLG aspects. I assume you can't do that with wOBA.

P.S., thanks for the explanation RMR, it's extremely helpful.

Yep; that's a very good summary of wOBA.

You're right in that there's value in understanding where a guy's production is coming from; but OPS doesn't tell you that any more than wOBA does. You can only break OPS in to SLG and OBP if you actually have the guys' OBP and SLG in front of you. At that point, OPS isn't telling you anything new.

If all you have is OBP and SLG, OPS is a quick shorthand for estimating overall production. But if you have OBP and SLG, there's no reason you can't also have wOBA -- you just have to choose to use Fangraphs as your go-to data source. For people used to using B-R or ESPN, that's a change.

[AtomicDumpling]

OPS is a mongrelized statistics that doesn't make a whole lot of sense at first, but it does correlate very well with run scoring at almost exactly the same level as wOBA does. OPS is a Frankenstein stat that is built using bits and pieces of other stats just like Frankenstein was built using bits and pieces of different bodies. OPS is not pretty but it gets the job done very well. It is accurate despite being ugly and sloppy.

wOBA is the best stat to measure offensive production, but OPS will tell you the same thing. Both of them are light years better than Batting Average, which often tricks people into thinking some hitters are better than others when they are actually far inferior.

There are two huge flaws in Batting Average. All hits count the same despite it being very obvious that some hits (home runs, triples, doubles) are far more likely to produce runs than other hits (singles). A good stat would weight those hits proportionately. The other huge flaw in Batting Average is that it totally ignores Walks and Hit By Pitches, which contribute to run scoring almost as much as singles.

On-Base Percentage fixes one of those flaws by including walks and HBPs into the equation, but it still doesn't account for the differences between walks, HBPs, singles, doubles, triples and home runs. OBP correlates much better with run scoring than Batting Average does.

Slugging Percentage suffers from one of the same flaws as BA by ignoring walks and HBPs, but Slugging Percentage does weight the hit types. So a home run is 4x as valuable as a single while a triple is 3x as valuable and double 2x as valuable as a single. The problem with that is those hit types are not actually worth 4x, 3x and 2x as much as a single in the real world of run scoring. So Slugging Percentage actually overvalues extra-base hits while failing to value walks and HBPs at all. Still, SLG correlates much better with run scoring than Batting Average does and is very close to as accurate as OBP is.

It turns out if you add OBP and SLG together it makes a thing of beauty -- OPS. The flaws in each base stat (OBP + SLG) are canceled out by the overvalues in the opposite stat. For example, since SLG overweights the value of extra-base hits it negates the fact that OBP ignores extra-base hits completely. OPS correlates even better with real-world run scoring better than OBP or SLG do individually.

wOBA is a linear weights stat where each individual event (single, double, triple, home run, walk, HBP etc) is factored into the equation at its proper contribution to run scoring. That is called using "Linear Weights" to establish the value of each event and that is the origin of the "w" in wOBA or Weighted On-Base Average. wOBA also includes Stolen Bases and Caught Stealings, so it is not a pure hitting stat but rather an offensive stat. Intentional Walks do not count, although I think it is debatable that they should count.

Here is the wOBA formula:
wOBA = (0.69×uBB + 0.72×HBP + 0.89×1B + 1.26×2B + 1.60×3B + 2.08×HR + 0.25×SB -0.50×CS) / PA

Those linear weights can change each season based on the run scoring results in the MLB games for that season.

Some versions of wOBA do not include Stolen Bases, while other versions even include Reached-Base-On-Error (RBOE) stats. The FanGraph stats use the formula above that includes SBs, CSs and does not include RBOEs.


Here are the 2012 stats for the Reds hitters and their team ranks for each stat.

Name			AVG	AVGrank	OBP	OBPrank	SLG	SLGrank	OPS	OPSrank	wOBA	wOBArank
Joey Votto		0.342	1	0.465	1	0.604	1	1.069	1	0.444	1
Ryan Ludwick		0.266	6	0.335	3	0.549	2	0.885	2	0.370	2
Todd Frazier		0.273	4	0.331	5	0.527	3	0.858	3	0.360	3
Jay Bruce		0.243	8	0.326	6	0.481	4	0.806	4	0.341	4
Brandon Phillips	0.295	2	0.334	4	0.453	5	0.787	5	0.341	4
Drew Stubbs		0.231	10	0.302	9	0.391	8	0.694	10	0.313	6
Scott Rolen		0.240	9	0.320	7	0.388	9	0.708	6	0.309	7
Chris Heisey		0.268	5	0.309	8	0.397	7	0.706	7	0.305	8
Zack Cozart		0.246	7	0.292	11	0.406	6	0.698	9	0.304	9
Ryan Hanigan		0.274	3	0.354	2	0.350	11	0.704	8	0.302	10
Devin Mesoraco		0.218	11	0.297	10	0.367	10	0.664	11	0.281	11
Wilson Valdez		0.216	12	0.246	12	0.241	12	0.487	12	0.218	12
Miguel Cairo		0.154	13	0.176	13	0.212	13	0.387	13	0.175	13

You can see that OPS and wOBA give you almost the exact same order for each player. The notable exception is Drew Stubbs, whose performance rates much better via wOBA than OPS. Much of this difference is likely the Stolen Bases that are included in the wOBA formula. Another interesting tidbit is Ryan Hanigan, who excels in AVG and OBP but lags badly in SLG, OPS and wOBA.

The takeaway is this, while wOBA is considered a more pure, mathematically sound statistic to sabermetricians you can still use the more easily calculated OPS and essentially get the same picture. You can also see how a reliance on AVG would give you a seriously deformed perception of reality.

Valdez and Cairo are far below replacement level according to any statistical measure.

Why not use that list above as the daily batting order? Is it too crazy to put that .465 OBP at leadoff? Should we make sure our most productive hitters get the most opportunities?

Another interesting aspect of linear weights is the strikeout. Linear weights can be applied to outs as well as hits. It turns out that a strikeout is no more harmful than other types of outs. We often hear that striking out is worse than hitting into an out because when you strike out you can't advance any runners. But it turns out that striking out also helps you avoid fielder's choice plays where a lead runner gets retired and double plays (not only grounding into double plays but also hitting line drives that cause a runner to get doubled off, or hitting "sac flies" that result in a runner getting thrown out trying to advance). The negative and positive effects of strikeouts cancel each other out and the end result is that strikeouts are just another out.

[Wonderful Monds]

It's for this reason that I'm not the biggest fan of WAR. Defense and base running are definitely important, but I don't think that they should be weighted exactly the same as offensive contribution. That's where we end up with silly things happening like Brett Gardener being proclaimed a top 5 player in the league.

[nate]

I posted this earlier in the year but here's a great infographic on how OPS and wOBA weighs events differently.



Article here.

[Raisor]

So how different is the top ten list of each stat this year?

[nate]

So how different is the top ten list of each stat this year?

Excellent question!

By OPS:

Name			OPS	wOBA
Joey Votto		1.069	0.444
Andrew McCutchen	1.040	0.434
David Ortiz		1.023	0.421
Mike Trout		1.010	0.441
Ryan Braun		0.972	0.410
Miguel Cabrera		0.965	0.405
Carlos Ruiz		0.958	0.407
David Wright		0.952	0.399
Carlos Gonzalez		0.951	0.403
Edwin Encarnacion	0.940	0.400

By wOBA

Name			OPS	wOBA
Joey Votto		1.069	0.444
Mike Trout		1.010	0.441
Andrew McCutchen	1.040	0.434
David Ortiz		1.023	0.421
Ryan Braun		0.972	0.410
Carlos Ruiz		0.958	0.407
Miguel Cabrera		0.965	0.405
Carlos Gonzalez		0.951	0.403
Edwin Encarnacion	0.940	0.400
David Wright		0.952	0.399

[AtomicDumpling]

Excellent question!

By OPS:

Name			OPS	wOBA
Joey Votto		1.069	0.444
Andrew McCutchen	1.040	0.434
David Ortiz		1.023	0.421
Mike Trout		1.010	0.441
Ryan Braun		0.972	0.410
Miguel Cabrera		0.965	0.405
Carlos Ruiz		0.958	0.407
David Wright		0.952	0.399
Carlos Gonzalez		0.951	0.403
Edwin Encarnacion	0.940	0.400

By wOBA

Name			OPS	wOBA
Joey Votto		1.069	0.444
Mike Trout		1.010	0.441
Andrew McCutchen	1.040	0.434
David Ortiz		1.023	0.421
Ryan Braun		0.972	0.410
Carlos Ruiz		0.958	0.407
Miguel Cabrera		0.965	0.405
Carlos Gonzalez		0.951	0.403
Edwin Encarnacion	0.940	0.400
David Wright		0.952	0.399

So both lists consist of the same 10 players. It goes to show that OPS and wOBA both reach the same conclusions via different methods.

[Raisor]

The only difference I saw was Easy Edwin Encarnacion on thw wOBA list instead of David Wright. Couple of guys were flip flopped in order also.