OPS? Over rated stat or not?

[RoseBowlMtg]

So are you saying OPS is worthless for a one year stat and only applies for a career? Looking at an 861 OPS I would always assume a player with a 799 OPS would have had a worse year. If in fact that was not true there must be a problem looking at that stat. So what do you do? Look deeper and see how they did with risp and that will help you tell the story.

[TooLiveBrew]

No - I don't mean that about OPS at all. But the way you're choosing to qualify 'worse/better year' is one that inherently can ignore OPS. Since you value the RBI production & BA w/RISP or runners on, OPS will be punished in many cases. The reason I like OPS is - as stated - that it doesn't vary as much. It's a more accurate way to label or rate a player's production, since the BA w/RISP or runners on stuff fluctuates much more so. Yes, excelling in run-producing scenarios is important, but you don't get yourself in those situations on BBs alone.

 

Players have to get on base in order for a guy like Simmons - to use your example - to have a bunch of chances to bring them in to score. OPS is weighted much more heavily towards OBP, and therefore does a better job of providing a consistent number where you can analyze a guy's production (and don't mistake that last word for only meaning driving in runs or scoring a run yourself, since you rely on teammates in both cases).

 

As with any stat, you can't just use one to have your 'answer.' However, if you forced me to choose between OPS & BA w/RISP or runners on, I'd choose OPS every time, since it tells a larger story than just how you fared in a much smaller sample of plate appearances, whose existence was, essentially, dumb luck for the batter.

[logan82]

BA w/RISP over the course of one year is usually so few at bats that it is almost worthless. Over a career it is much more usefull.

[RoseBowlMtg]

BA w/RISP over the course of one year is usually so few at bats that it is almost worthless. Over a career it is much more usefull.

 

Well I agree it is more important over a career but this does tell you a whole lot imo.

 

BA/risp:

 

Sixto: 19-98 22 RBI's .194 and an RBI every .224 AB

Simmons: 63-169 91 RBI's .373 and an RBI every .538 AB or 240% more efficient.

 

BA without risp:

 

Sixto: 90-298 .302 27 RBI's

Simmons: 125-431 .290 17 RBI's

 

14 solo shots vs 7 really helps the OPS but it's only 7 runs.

 

The entire year:

 

Sixto 109-400 49 RBI's .273 an RBI every .123 AB's

Simba: 185-600 108 RBI's .308 an RBI every .180 AB's or almost 50% better.

 

If you look at the entire year you would have to say a .308 BA is better and knocking in a run .18 is better than .123. Now we know Sixto didn't have as many opportunities so that's why you look at how he did in the opportunities and we have to agree that .194 is worse than .373.

 

If looking at where Sixto got 45% of his RBI's and Simmons got 84% is "worthless" imo you are missing some of the beauty of stats. If a player gets 14 solo HR's to inflate an OPS but only hits .194 with risp you can see a lot.

[TooLiveBrew]

He meant it's worthless in terms of being predictive.

 

I think the most interesting thing you've presented so far is how incredibly higher Lezcano's 'rate' of RBI was without RISP. Doesn't that help you see what others have been saying about this? I don't mean that mean-spiritedly, though I'm sure you know that.

 

http://forum.brewerfan.net/images/smilies/smile.gif

 

I'm sure you could/can find a year in which Lezcano fared much better with RISP. I don't mean this negatively, since you've carefully selected your example & have stuck to it (& it's provoked a great discussion), but since you sort of cherry pick a year in which one guy fared very well w/RISP, and the other pretty much stunk, it's going to work out heavily in Simmons's favor. I know you used this example due to the discrepancy in OPS & BA w/RISP... just sayin'...

[sbrylski]

TooLiveBrew said:

---

I'm sure you could/can find a year in which Lezcano fared much better with RISP.

 

Here is my post from page one of this thread:

 

sbrylski06 said:

---

Step two is addressing the high variance in BA w/RISP. Here's your selected season compared to the one after:

 

Sixto:

1977 - .194 / .339 / .306 / .645

1978 - .229 / .381 / .394 / .775

 

Simba:

1983 - .373 / .423 / .544 / .967

1984 - .269 / .303 / .418 / .721

 

This thread has turned all the way over 1.5 times. Nobody is going to come to a consensus because RoseBowlMtg isn't going to budge off his conclusion - which is actually a bit different than what us counterpointers are arguing. RoseBowl doesn't care about the season after - he's looking for who had the "better" season that year. Nothing wrong with that, but that is why pointing out the high variance in BA w/RISP carries no weight to him. He's looking retrospectively, while everyone else is looking predictively.

 

RoseBowl is correct in his conclusion, I believe. BA w/RISP has been shown to not really be a skill, but that doesn't change the fact that occasionally guys have great years in those situations. A hit with nobody on and two outs is of little worth relative to one with two outs and a runner on third. Saying a .333 hitter will have a 33% chance of a hit in both situations is the predictive argument. If he strikes out twice with the bases empty and gets his hit with a runner one third, that retrospective fact that his hit was worth more. (However, like argued, has virtually zero bearing on when he gets his hit in his next set of three AB's.)

 

I don't think there is historical WPA* (definition) numbers for these two players and the selected seasons, but I'd predict based on the near extremes of their "clutch" numbers that despite having the higher OPS, Sixto would have added less win probability than Simmons - a real-time statistic supporting RoseBowl's argument.

 

Using data that is available to illustrate the point, from last season:

 PLAYER PA OPS RISP WPA LI
David Wright 711 .963 .310 4.09 0.95
Aramis Ramirez 558 .915 .338 4.55 1.08 

Just the first two players I picked. Wright had the higher OPS, but Ramirez contributed more win probability for his team despite 153 less plate appearances. He did so because he performed better in important situations, and had more of them overall, as shown by his LI (http://www.hardballtimes.com/main/statpages/glossary/#li'>definition).

 

So just as Ramirez did have a more valuable season compared to Wright, Simmon's 1983 was also more valuable than Sixto's 1977. (Even ignoring their defensive positions.) However, RoseBowl, the way Ramirez had been blasted by the Chicago fans and media the two seasons prior for only hitting when it didn't matter is one example of why digging into situational batting statistics, such as BA w/RISP or WPA, don't hold water for predicting future worth.

 

(Another note: If you want to be all-encompassing in your retrospective offensive evaluation, we should also take into account Simmons' awful baserunning - but that's for a different thread...)

 

I don't think I said anything a lot of us don't already know, have discussed, or said in this thread already. But this thread is going nowhere, so I hope this post can serve as a type of conclusion. I'm sure it'll get picked apart though...

[TooLiveBrew]

If Riske can seemingly hold his LOB% high, I'd bet there are hitters, too, who hold their BA w/RISP at a good level. Couple that with consistency in OPS & you have an elite player.

[sbrylski]

If Riske can seemingly hold his LOB% high, I'd bet there are hitters, too, who hold their BA w/RISP at a good level.

Very, very few. Simmons just happens to be one of them - an exception to the rule.

 

[TooLiveBrew]

Would you describe the Riske occurrance one that 'very, very few' sustain as well? Intuitively I'd guess so.

[RoseBowlMtg]

(Another note: If you want to be all-encompassing in your retrospective offensive evaluation, we should also take into account Simmons' awful baserunning - but that's for a different thread...)

Great take once again that I appreciate digging into baseball when it's cold out. Side note of note about Simmons bad baserunning was that he had 4 SB's with 2 CS vs Sixto's 6 SB's with 5 CS. A catcher outperforming a RF just in that year has to be a surprise.

 

My point on this thread was not to talk about history since I know it fluctuates quite a bit from year to year but to see why an 861 OPS was worse than a 799 OPS. Not history and variance but only that year. 83 wasn't Simmons best OPS with risp in his career but his best in Milwaukee. A 967 OPS with risp is interesting imo since his year end was 799. 13 HR's with 108 RBI's is what is special since I believe knocking in 95 teammates is second best in Brewer history behind only Coop's incredible 1980 season with 97 teammates knocked in. I guess I am an anti-all HR type of fan and appreciate ABC baseball and the knack to do it. FYI from 71-83 Simmons had 6 yrs with an obp of 409 or better. A career .299 with risp, .299 with men on and .272 with no one on to end up with a career .285 ave. Sixto's had a .254 with risp, .272 with men on, and a .270 with no one on for his career .271 ave.

 

So a .285 career and a .271 isn't much of a difference until you see that .285 hit .299 with risp and .271 hit .254 with risp and those are big differences imo.

 

I am excited to see how his influence rubs off on our young studs this year.

[bjkrautk]

I don't have anything approaching substantive numbers to throw out about this discussion, but I do feel it's emblematic of my underlying skepticism with statistics: they only measure what they're intended to measure.

 

That probably sounds either profound or stupid to most of you (probably the latter), but hear me out. Ever since the onset of the box score and batting average, hitting has been a binary measurement: base hit or out. With the OBP revolution of SABRmetrics, the category expanded, but is still essentially binary: on base (hit, get hit, or walk) or out. This begs the question -- is the point of baseball to not make outs, or to score runs?

 

There are situations wherein one comes at the cost of the other: the intentional walk, the sacrifice fly, the routine grounder to first base which brings around a runner from first, etc.

 

There are a number of stats out there which can (arguably) start to dig away at that middle ground: RBI, "Productive Outs," etc....but they're generally regarded by the SABR-types as less predictive than OXS or OPS, and rightfully so. Clouding the matter even further is the realization that players skilled enough to engage in this sort of "situational hitting" tend to also be successful in SABR measurements. Carlos Lee is the example which keeps coming to mind: he's usually among the league leaders for RBI's in part because of his place in the batting order, in part because he does a good job of situational hitting, and in part because he's just a good OPS hitter.

 

That's the problem with RBI: it's not as consistent from person-to-person, year-to-year, and number 1 hitter to number 8-9 hitter. If RBI's were the only productivity stat we had, I am sure we'd all know which batters bring in hitters at a rate above league average (or even replacement level), but we have a measure that - while not covering everything, does a pretty good job of measuring offense on an independent, individual level.

 

In the end, I do use OPS (OXS when I can get away with it, especially when players have a significant disparity in OBP). I just think it also would be useful to know which players knock in more runs than their situations & stats would have predicted (if anyone was actually measuring for it), and especially if the same sets of names tended to come up year after year.

 

 

(NOTE: If there are statisical models to determine whether a player knocks in more runs than situation & stats would predict....then chances are it's in that 90% of SABRmetrics that I know and understand even less than xFIP.)

[TooLiveBrew]

First off, great post, bjkk.

 

statistics: they only measure what they're intended to measure.

Exactly. The reason OPS is used a lot is that it tells you something about a hitter over which he has a relatively high amount of control, as opposed to ERA for a pitcher.

 

This begs the question -- is the point of baseball to not make outs, or to score runs?

I'd argue the clearly superior strategy of the two is the first - the second comes as a byproduct of the first. But don't tell that to Ozzie Guillen.

 

NOTE: If there are statisical models to determine whether a player knocks in more runs than situation & stats would predict....then chances are it's in that 90% of SABRmetrics that I know and understand even less than xFIP.

The ironic part to me about this is that xFIP is relatively easy to understand - heck, I 'get' it! A pitcher's expected Fielding-Independent Pitching (meaning, measuring even moreso just what he 'controls'). While I'll never pretend to fully understand the number-crunching that goes into something like this, I know what the appx. range of 'good' to 'bad' is.

 

Once you'd get into predictive stuff for counting stats like RBI & R, I think I'd be lost too. There'd almost have to be a new formula for each player for each season.

[Phenomenal Smith]

I would like to see a Runners Inherited and Runners Driven In stat. Of course, driving in a guy from third with less than 2 outs is easier than doing it from 1 with two outs, but for simplicity sake I would still prefer a simple summation - kinda like Runners Inherited and Runners Scored for relief pitchers.

[logan82]

I would like to see a Runners Inherited and Runners Driven In stat. Of course, driving in a guy from third with less than 2 outs is easier than doing it from 1 with two outs, but for simplicity sake I would still prefer a simple summation - kinda like Runners Inherited and Runners Scored for relief pitchers.

 

It would be interesting. I think you can find that stuff in the splits at ESPN.com. That kind of stuff wouldn't tell you much over the course of a season though. I wouldn't be very predective or be able to tell you much about a hitter until you get a few years worth of stats. Much like looking at a hitter's batting stats over the course of one month. One or 2 hits either way can cause a huge fluxuation.

[sbrylski]

I would like to see a Runners Inherited and Runners Driven In stat. Of course, driving in a guy from third with less than 2 outs is easier than doing it from 1 with two outs, but for simplicity sake I would still prefer a simple summation - kinda like Runners Inherited and Runners Scored for relief pitchers.

It would be interesting. I think you can find that stuff in the splits at ESPN.com. That kind of stuff wouldn't tell you much over the course of a season though. I wouldn't be very predective or be able to tell you much about a hitter until you get a few years worth of stats. Much like looking at a hitter's batting stats over the course of one month. One or 2 hits either way can cause a huge fluxuation.

And it will likely correlate very highly with BA and probably even moreso with SLG.

[Ennder]

Even if you are a big fan of RBI as a statistic to judge how many runs a player produced it is completely meaningless without including R's as well. Run production is a combination of getting on base and driving in runners.

 

That being said OPS is overrated because it doesn't count OBP enough so it doesn't tell the whole story.

[TooLiveBrew]

Is EQA a better measure? I know there are people who swear by that stat. Otherwise, I suppose one could just adjust OPS with the 1.8x weight on OBP, right? Heck, that probably already exists under a different name.

[sbrylski]

Otherwise, I suppose one could just adjust OPS with the 1.8x weight on OBP, right? Heck, that probably already exists under a different name.

 

GPA (Gross Production Average) = (OBP * 1.8 + SLG) / 4

 

OPS correlates something like 96% to runs scored. The best, most complicated statistics correlate about 98%. I'm content with OPS as judgement of overall hitting worth. GPA's not hard to calculate either, or OXS, but they really are only marginally better.

[mothershipconnection]

Why is it divided by 4? Simply to have a smaller number, or is it then used in a more elaborate equation that makes smaller numbers easier to work with?

[brewers19]

Probably so that the scale relatively matches that of batting average for reference or something.

[sbrylski]

Why is it divided by 4? Simply to have a smaller number, or is it then used in a more elaborate equation that makes smaller numbers easier to work with?

Like b19 said, it scales it very similar to batting average. 300+ is very good, 260- or so is below average.