Looking at b-ref today, some musings about the ops stat. It's a number that we are so used to seeing now, and all it is is a simple addition of OBP+SLG. So for example, consider these two hitters: Nelson Cruz and Brandon Belt. Cruz: 280 ab, 22 hr, .264/.359/.543/.902 Belt: 281 ab, 13 hr, .302/.394/.505/.900 Virtually identical numbers of at-bats and ops. But how they got to that .900 (or .902) ops is very different. Cruz got there with substantially better power than Belt. Belt got there with substantially better OBP than Cruz. Which is more important? Consider how many bases need to be accumulated to score a single run. The answer seems, of course, to be 4. You need a minimum of 4 bases to score a run, and that's if they all come in the same inning. Naturally, you could have tons of singles hit during a game and never score a run. A home run gets you those 4 bases in one swing. If you are going to do nothing but hit singles, and advance one base for every single, here's how many total bases (when you add in base running) you need to score the one run: Betts singles = 1 total base, 0 outs Benintendi singles, Betts to second = 2 total bases, 0 outs Martinez singles, Betts to third, Benintendi to second = 3 total bases, 0 outs Moreland singles, Betts scores, Benintedi to third, Martinez to second = 4 total bases, 0 outs Next three batters get out So you need a total of 10 individual bases, advancing one base at a time, to score the same run that you'd get if Betts simply led off with a homer, tallying all four of the necessary bases at once. So it's obvious that hitting a bunch of singles is way less efficient than hitting home runs. However, hitting a bunch of singles is much easier than hitting home runs - that's true for even power hitters. Consider Cruz - he has hit 41 singles, 10 doubles, 1 triple, and 22 homers (74 total hits). So he's twice as likely (roughly) to hit a single as he is to hit a home run. Now, the flip side is this. Let's say the two alternatives are: Scenario 1 Betts singles = 1 total base, 0 outs Benintendi singles, Betts to second = 2 total bases, 0 outs Martinez singles, Betts to third, Benintendi to second = 3 total bases, 0 outs Moreland singles, Betts scores, Benintedi to third, Martinez to second = 4 total bases, 0 outs Next three batters get out Team OBP: .571 Team SLG: .571 Team OPS: 1.142 Scenario 2 Betts homers = 4 total bases Benintendi, Martinez, and Moreland all get out Team OBP: .250 Team SLG: 1.000 Team OPS: 1.250 In each scenario, we have 1 run being scored, and eventually 3 outs being made. We highlighted the disadvantage of trying to do this with all singles (and runners advancing only one base). Namely, you need FOUR players to get singles, instead of just one player hitting a home run. But there are other factors. Consider: (1) Pitch count. Scenario 1 should make the pitcher throw many more times than scenario 2. And (2) Outs made. In baseball, outs are currency. You only have 27 of them (regulation 9-inning game). Hitting four singles in a row means that out of those four players, you are accumulating zero outs. Hitting one home run and then getting three outs means that out of those four players, you have accumulated three outs. Since it is much easier to hit singles than home runs, if all you had were either singles, home runs, or outs, you are accumulating outs much faster if you try to hit home runs than if you hit singles. Long story short - which is more important: your on-base percentage or your slugging percentage. - OPS suggests a 1:1 ratio, since the two stats are just added together with no sense of which one is more important. - In the Moneyball era, the conventional wisdom was that there was a 1.5:1 ratio - that is, every point of OBP was worth 1.5 points of slugging. - The Oakland A's view was even more extreme than that - that is, every point of OBP was worth 3 points of slugging. Hence you see in the movie, what does Hatteberg do well? He gets on base. Getting on base and not using up your currency of outs was the driver behind baseball thinking for a long time. But it seems like that's changed considerably. Now it seems that teams would happily trade getting on base for homers. They don't want their players to stop trying to crush homers, even with two strikes (it used to be they'd encourage hitters to choke up, shorten their swing, and put the ball in play). Do you guys know what is the more valuable metric? Slugging or OBP?

OBP is worth about 1.4 times more than slugging percentage and why OPS+ exists. edit: You answered the question yourself.

I thought the coefficient on OBP was higher but I havenâ€™t looked in a long time Edit: bosox answered better

According to Tom Tango the coefficient is closer to 1.7-1.8 http://www.insidethebook.com/ee/index.php/site/comments/why_does_17obpslg_make_sense/

Thanks - but that article RIrooter09, is from 2007. Things have changed in baseball considerably since then. Teams simply do not value OBP as compared to SLG like they once did. So is the math still the same and teams are just "off" in their valuation, or have the relative values actually changed?

Kind of unrelated, but the OPS calculation doesn't make sense to me. First of all, two fractions with different denominators are added and then batting average is double counted. Wouldn't total bases plus walks divided by plate appearances be better.?

Below is a graph showing the team-by-team statistics for the 2017 season. I've shifted the values for SLG by -0.1 so that they more nearly overlap with the values for OPB. The eye can see pretty clearly see that the red points (OBP) are more tightly correlated with runs scored than the blue points (SLG). A quick calculation confirms this: the correlation coefficient between OBP and runs scored is 0.91, while that between SLG and runs scored is 0.87. The difference is small, and the sample size is also small. Draw what conclusions you will.

Cool info. So why is baseball shifting away from OBP and more towards SLG (home runs, mainly) in terms of value? Why are teams and players into the "launch angle revolution", when that drives up power at the expense of getting on base?

The main appeal of OPS is that, while it's a little dumb, it's vastly superior to any of the traditional baseball card stats and requires only adding 2 of those numbers together. You can usually calculate it mentally when a batter pops up at the plate. If you're going to do anything more complicated, just use a linear weights system like wRC+ or Batting Runs.

You'd have to look at the relative impacts, no? If the launch angle revolution drives up slugging by 2 points for every 1 of OBP it sacrifices, it's probably worth it even if OBP is worth 1.7 times SLG. Also the impact on OBP is varied; launch angle may hurt OBP by making somewhat less contact but also help OBP by turning some fly balls into HR. It's tough to tell the overall impact. FWIW (very little), MLB's aggregate OBP is .318 this year compared to .317 in 2015 (a year picked randomly from the dropdown on the stats page I was on).

Are they going towards slugging or are they going towards ISO? Does ISO translate to runs scored better than just pure slugging percentage?

Good question. I definitely think they're going for home runs. Fiddling around more with B-ref... I added up Total Total Bases, which for my calculation includes all the following information available on B-ref: Total Bases (4xHR, 3x3b, etc.) + BB + HBP + ROE + SB + bases taken. I then factored out DP, CS, Pickoffs, and OOB. So TTB = all bases taken minus all outs made. It doesn't cover literally everything, as it doesn't tell us how many ROE (reach on error) are two-base errors or three-base errors, etc. But it's pretty thorough. I then wondered what the ratio is between TTB and R. Over the past four years, here's the major league TTB per R scored: 2018: 5.09 2017: 4.97 2016: 5.07 2015: 5.17 Here's Boston's: 2018: 4.75 2017: 4.88 2016: 4.74 2015: 5.00 So compare 2016, 2017, and 2018. Year: MLB, Bos 2018: 5.09, 4.75 (-0.34) 2017: 4.97, 4.88 (-0.11) 2016: 5.07, 4.74 (-0.33) Interestingly, the Sox in 2017 had a very poor offense, with very little power. So what this very small sample of 2.5 seasons might be saying is that as the Sox have gone with more power (by design or result is immaterial), they've required fewer TTB to score a run. In other words, more efficient. Maybe that's the upside of more power.

that's why things like wOBA and wRC were created. OPS is a good starter sabermetric stat, but more is needed to evaluate a player. The data points for OPS per each PA result I believe are the following: -out/reach on error: 0.000 -walk/HBP: 1.000 -single: 2.000 -double: 3.000 -triple: 4.000 -HR: 5.000 it's a bunch of round numbers essentially averaged up. Sacrifices I think get accounted for as well

This is similar to my feelings on WHIP. I get that it's quick and simple and the components are always easily available, but any stat that treats allowing a home run the same as allowing a single seems pretty pointless to me. It certainly treats the Brad Radke and Josh Tomlin type pitchers more favorably than they deserve (the guy who pitches with the mantra, "As long as I throw strikes, the batter will get himself out more often than not"). Replacing hits allowed with total bases allowed, making it TBWIP, would tell us much more. *

The obvious approach there is to have a 'wRC+ allowed" or "wOBA allowed" stat for pitchers to measure the exact same thing as for hitters. I'm still not sure why I don't see that very often. (I don't think that stat even exists at all on a place like Fangraphs, though it's possible that I just missed it.)

The wOBA equivalent for pitchers most commonly seen is FIP (which both Fangraphs and Baseball Reference list). The linear weights link I posted upthread discusses both wOBA and FIP.

The problem with both OPS and WHIP is that both are really useful at the extremes (we know an OPS of .550 is bad regardless of how it is derived and we know a WHIP of 0.801 is good regardless of how it is derived)... but the middle portion (which includes most players) is just squishy. We need more information. But I've never been a big fan of one stat fits all anyway. I'll take BA/OBP/SLG every day of the week and have a really good idea of what a hitter is.

FIP is fine, but it only looks at the three true outcomes. That may be good predictive measure, but sometimes you want to be more at the descriptive end of the continuum.

As long as you factor in the parks they play in, I still think it's the best way to measure a players skill with the bat. People want everything in 1 convenient stat tho. I hate the new trend that has people doing OBP/Slugging/OPS. It tells us less information and assume we can't add.

There is a problem with On Base Percentage, smaller to bigger depending on the year looked at, as the rule has undergone several changes/removals/restorations. 1889-1893 -- Players were credited with advancing baserunners on bunts, ground outs, and fly balls, but they were not exempted from an official team at bat when credited with a sacrifice. 1894 -- sacrifices were limited to bunts and the batter not charged with time at bat. 1908-1925 -- sacrifice fly credited if runner scored after catch; batter not charged with time at bat; however, sacrifice bunts and sacrifice flies were lumped together in official stats. 1926-1930 -- any fly ball that advanced runners to second and/or third was counted as a sacrifice fly with no time at bat charged. 1931-1938 -- sacrifice fly rule eliminated. 1939 -- sacrifice fly rule reinstated. 1940-1953 -- sacrifice fly rule eliminated. 1954-- sacrifice fly rule reinstated

xFIP bakes in a certain factor of luck on homeruns. In effect, fly ball pitchers would look a bit better here.

Since OBP is a function of plate appearances, not at bats, those particular changes affect some stats, but not OBP.

Isn't there still the problem in the early days when sacrifice bunts and sacrifice flies were lumped together?