And what about home runs? Do they require double the power? No. The difference between some doubles and home runs are mere feet. Why does a home run hit 405 feet receive twice the points as a 395 foot double? If we’re measuring power, the ratios don’t seem to work out. Yes, a home run requires more power—250 foot line drives can end up in gaps, but they don’t end up over the fence—but does SLG accurately reflect the difference? Again, I think not.

If we’re looking at measuring power, total bases seem to be a poor way to measure it. Sure, it makes some sense, but total bases are dependent on defense, park, etc. in many of the ways that wins are dependent on other things. Hitters are still primarily responsible, but there are other factors that lead to extra bases than a hitter’s power. Instead, it might be more accurate to figure out the average distance a ball is hit—total feet covered by batted ball/number of batted balls. We have all sorts of spray charts, so I figure that, while it will take more time, we do have the technology.

But is SLG completely useless? What about measuring how likely they are to bring in runs? According to The Book website, singles have a run expectancy of 0.47, doubles 0.77, triples 1.09, and home runs 1.39. If we use this model, we don’t have the same problems SLG did with power, namely that doubles and triples are essentially the same batted ball but towards different spots and with different speed runners, but there are still problems. SLG gives doubles double the points of a single, but doubles don’t give twice the run expectancy. Home runs don’t give twice the run expectancy of doubles. It’s a lot closer than simply measuring power, but again, it falls short.

I’m not the first to realize some of these problems. ISO (Isolated power) realizes that all SLG are not created equal. A .220 hitter that has a .420 SLG has more power than a .320 hitter with the same SLG. ISO, then, takes SLG – BA, and the resulting difference gives us a better look at how many more extra base hits one hitter has over another. Unfortunately, it uses SLG. wOBA (and other similar statistics) avoids using SLG, but it’s an overall batting statistic that doesn’t directly reflect power.

SLG is a pretty quick and easy statistic to calculate, but I’m not entirely sure what it *is* calculating. Directly, it calculates the average number of total bases a batter gets per at-bat, but I don’t know that it directly reflects a batter’s power. And if ISO essentially just uses SLG and BA, I’m not sure it does either. Sure, it gives us a general idea, but I don’t know if it’s an *accurate* answer to the question of who the most powerful hitter in baseball is.

This started from an interesting article I read at Wahoo Blues. The idea was that ISO doesn’t accurately reflect power because not all ISOs are created equal. For instance, a .200 ISO isn’t always the same. A .220 hitter with a .420 SLG has more power than someone with a .320 average and .520 SLG because the hitter with a .320 average has quite a few singles boosting his SLG, and thus his ISO. Lewie Pollis has an excellent idea on how to solve it, but I wonder if it’s not based on a flawed premise. If SLG is flawed, then anything even tangentially affiliated with it, no matter how many times removed, is flawed, right?

So what now? Power is about more than just hits. Home runs and doubles are nice, but guys hit really far outs as well. Guys also have really short hits as well. Yet, SLG and other statistics treat all singles the same, all doubles the same, etc. when some obviously demonstrate more power—an infield hit is treated the same way as a line shot that lands just before the center fielder, but they obviously demonstrate differing levels of power. Every batted ball should be counted when considering power because they all demonstrate an aspect of power. A well-struck ball at a left fielder is the same as a ball hit similarly well toward the gap, but they are counted differently. As for how to solve this problem, I have options but not an absolute.

The average batted distance seems to be the simplest way to do it, and I think we have the technology to do it. But how exactly is the distance measured? For instance, a line shot at Jeter will be caught by Jeter a certain distance from the plate, but it obviously could have gone farther. And does a line drive that goes 320 feet require more or less power than a fly ball to the same distance? And where exactly (*I really want to know the answer to this by the way*) does a home run distance come from–where it actually stops or where it would have landed had stands and fans not gotten in the way?

So we could also use the batter’s average batted ball speed. Hit f/x, I think, records such measurements, but I don’t know if that information has been released. But I don’t know if a batter would hit enough of a certain pitch to throw things off. For instance, would a batter who hits and/or faces an inordinate number of fastballs going to have a higher batted ball speed because the pitch speed is faster, and would a batter who faces more breaking pitches see a batted ball speed less than he should? Maybe we could just use batted ball speeds off fastballs. Again, I’m better at giving options.

Those are a couple of my answers, but feel free to leave more in the comments section. Also, if I’m missing something completely obvious that throws all of this under a bus, I’ll take that as well.

Hit FX and batted ball speed are going to be the business. We should finally be able to tell who is making the hardest outs, hitting the ball farthest on average, etc. At least for HRs, we have hittracker.com that lines hitters up by average distance and stuff. Player FX + Hit FX will let us see who has the ball coming off the bat the hardest, and who ends up with the longest average distance (on the fly or line ) for doubles, triples, HRs.

I think you're going in a sound direction, but probably taking it too far. After all, while the difference between a double and a triple/HR might be pretty marginal in batted ball terms, in terms of value the triple or a home run is greater than the value of the double, even if the only difference is the speed of the runner or a few inches of height on the wall.

A better solution than getting rid of the stat altogether is probably changing the way we talk about it. Instead of saying it's a measure of power, which IS kind of a crude description, maybe we could start referring to it as a total base average or something more precise.

Maybe I'm off on this one, but I have never interpreted SLG PCT to be a measure of how far a batter hit the ball. It is a measure of "power" hitting as it applies to extra bases. How many bases per at bat does a guy produce? How far a ball is hit is incidental. A hitter may fly out to the wall in one at bat, then hit a gapper that just clears the infield. A guy who hits a large number of doubles and fewer homers, may slug with a guy who hits more homers, but has fewer doubles – who cares how far it went (or how hard it was hit), the measure is on bases per at bat. As such it is a good stat, although I often think the stat should be modified, utilizing plate appearences and incorporating walks would seem to be more accurate. A more direct measure than OPS.

Hit FX and BBS as mentioned in comment #1 would also seem to be very useful, and may indirectly correlate to SLG, but I think ultimately the two measure different things.

If the point is that slugging doesn't really purely measure power, then absolutely, I'm on board. But if you correlate slash statistics to runs and RBIs, SLG is actually a stronger indicator than OBP, OPS, and AVG. At least it was when I did this stuff back in college, which seems like a long time ago, these days.

I was under the impression that OBP was most closely correlated with runs scored. Either way, there's a fairly strong correlation between SLG and runs scored.

I don't think the idea of slugging percentage was to ever identify power. It was created to identify the number of bases a batter got per at bat, and does exactly what it was designed for. It was probably mis-named as slugging percentage, and would have better described as a base percentage instead.

The only way it was ever really used to describe 'power' was due to the deadball era, in which a 'slugger' would hit more doubles, triples and homeruns (mostly inside the park) than the Punch and Judy hitters who only hit singles. The closer you got to home in your own at bat, the closer you were to scoring on any type of hit.

I think the only flaw in it is that it doesn't count stolen bases, as they are earned as surely as any hit or walk earns a base.

I think you answered your own question– All offensive stats try to measure the same value; how many runs does this player create? No one values power (or total bases) as a stand alone trait– it is the assumption that greater power equals greater run production.

Rather than use the archaic 1.0 for a single, 2.0 for a double….why not use the values presented above in the same manner (singles 0.47, doubles 0.77, triples 1.09, and home runs 1.39) to measure the amount of runs created per AB. THAT would be a more useful metric to measure a players offensive value.

It seems that in terms of measuring power:

future hit f/x metric > something like power factor > iso > SLG calculated using wOBA coefficients (also known as wOBAcon – this is already in use) > SLG

Stuff like ISO and SLG have issues because of the weighting problem and because they are dependent on batting average. Perhaps only a caveat is necessary – to only use SLG/iso/wOBAcon when batting averages are similar.

I don't think SLG measures power. It measures total bases per at bat. This post is silly. This blog has become way too stat obsessed. I'll stop complaining now and while I'm at it, this post is the tipping point. I will no longer read this blog. Thanks, it was fun while it lasted.

Good post and many good points being made, but nowhere has anyone talked about the horizontal component coming into play other than the difference in between a double/triple or single/double. A measure that uses vertical trajectory and velocity will have value, but you will introduce another error as well.

Each horizontal avenue through the field will also have a value associated with it, as the corners are usually much closer to the plate than center field, the slugger who pulls the ball a ton will receive a higher ISO and SLG for hitting the ball with the same vertical trajectory and velocity as the slugger who hits mostly to center field. Looking at Jeter's homeruns in 2009, if he hit most of them to center instead of right field, they would have been long outs, lowering SLG and ISO while keeping the trajectory and velocity the same. I would think it is safe to assume a player would also consistently hit the ball to the same field over his career. Seems like SLG and ISO would be the better tools for predicting future success hitting HR's as well, unless horizontal angles are part of the equation and that even more complicated.

Having said that, someone post the data and I'll be happy to nerd out with Matlab for a while :)

That's hit f/x data stuff, which unfortunately isn't publicly available at this time, or else I would join you in nerding out. However, there is one month of data that is publicly available that you can get if you sign up here: http://baseball.sportvision.com/summit

I just signed up as non-attending. I'm not sure if this will grant me access to the one month of hit f/x but it's supposed to based on this: http://www.insidethebook.com/ee/index.php/site/co…

That's exactly why I wanted to take a physics-based look at the batted balls and figure out their final resting positions instead of simply where they bounce. If we ever get the hit f/x data, count me in for a coding session!

Awesome comments above. I'll try to clarify a few themes.

1) SLG is used as a measure of power. When you look at slash rates, most people think hitting ability/ability to get on base/ability to hit the ball hard. Yes, technically SLG is the number of total bases per at bat, but how many times do you really see it used that way? Now, like Brien said, am I messing around in the margins? Probably, but my point is to ask if there's something better. SLG seems to have some problems.

2) Maybe SLG shouldn't go, but the headline was maybe just to grab attention. If slugging doesn't measure power, then what does? ISO and Power Factor are better, but are they really measuring power. The idea of extra base hits is nice, but batted ball speeds and distance seem more of an indication of a guy's power. Extra base hits are only a small percentage of what could be a wealth of information detailing a guy's power.

I don't understand the preoccupation with 'power'? It doesn't really matter how far a guy hits the ball. A 302' home run counts the same as a 500' home run. Frank Taveras, he of the two career home runs, once hit a grand-slam that was a bouncing ball down the third base line that got into the corner and rolled around. Counted for 4 runs, just the same as any ball that crossed the fence.

The infatuation with power is just a leftover of the steroid era, when guys like Adam Kennedy were hitting three home runs in one playoff game. Power, for power's sake, is overrated. Some guys hit the ball 400 feet on a regular basis, and it's a fly out to center field. Others hit 60 feet on a regular basis and end up on 1st. I'll take the single over the out any day.

It doesn't matter how hard or how far you hit the ball, it matters how many bases you get out of it.

While a 302' homer is worth the same number of runs as a 500' homer, we're talking about probabilities. If one can hit the ball 500' then he is more likely to hit the ball 302' than a hitter who can just barely hit the ball 302'. While I agree that the ultimate goal is to maximize wRC, it's useful to gauge a hitter's likelihood of parking the ball.

Sorry, but that doesn't mean anything to me. I'm not against sabremetrics. I think they're are a good thing for the game.

But I don't deal with probabilities. I deal with reality and what actually happens, not what might happen

Probability says any batter might hit a home run or strike out.

Reality is Adam Dunn will do a lot of both, while Ichiro won't do much of either.

Reality is that Adam Dunn hits the ball a lot farther than Ichiro, but reality is Ichiro was on base more times, and had more total bases (if you count stolen bases, which you should).

Power, for the sake of power, is over rated. It's good to hit the ball hard and far, but what are you doing with it? That's the reality.

Dunn has a career .381 OBP to Ichiro's .376. He averages a .521 slugging percentage to Ichiro's .431. Ichiro actually averages more total bases, but his slugging percentage is dragged down by the fact that he doesn't walk very much, which means he averages a lot more at bats (if anything, this is probably the most obvious problem with SLG%).

Anyway, your final paragraph seems pretty ridiculous. The harder a ball is hit, the less likely it is to result in an out. That seems pretty obvious.

It's okay if you don't understand what I'm talking about. Some people are caught up in numbers and don't really know anything about the game as it's played on the field.

If you want me to clarify something, just ask. No need to be rude and insulting about it.

"It's okay if you don't understand what I'm talking about. Some people are caught up in numbers and don't really know anything about the game as it's played on the field.

[…]

No need to be rude and insulting about it. "

I believe there is a saying about pots and kettles and the color black that may be appropriate here.

OK, let's play it your way.

While it's true that any hitter might homer or whiff, that's not what probability is about. Probability isn't about whether an outcome IS possible, it's about how LIKELY the outcome is. If the reality is (and it is) that Dunn will do either more than Ichiro, that means that he is more likely to do so. Probabilistically, based on past outcomes, we can say that any given Adam Dunn at-bat is more likely to result in a home run or strikeout than any given Ichiro Suzuki at-bat.

The point is that if you are MORE LIKELY to hit the ball very far, that means you are MORE LIKELY to hit a home run (which is one of the three deterministic outcomes, the others being walking and striking out). If a hitter has a higher probability of hitting a home run, he will hit more home runs in reality.

And just as an aside, I disagree that SB should be counted in TB. The TB stat is designed to measure how many bases a hitter gains from at-bats, which is an interaction between pitcher and hitter. In my opinion, it would be more prudent to add BB into TB than SB because BB involves a pitcher-hitter interaction while SB is another best entirely. Another statistic, say "Bases Gained", would be useful for measuring a player's total on-base output… assuming you could account for the fielding statistics that would compound the equation.

"According to The Book website, singles have a run expectancy of 0.47, doubles 0.77, triples 1.09, and home runs 1.39."

I majored in English, so could someone please explain why homers aren't a base one in this system? The only thing I can think of is that it's accounting for additional runners, which would mean it's count RBI.

You should check out this: http://sports.yahoo.com/mlb/blog/big_league_stew/…

"For example, home runs have a run value well over one because runners can be on base, and singles have a value slightly higher than walks, because singles frequently move baserunners from first to third or from second to home. The exact coefficients are determined by an analysis of game data. They are, therefore, relatively precise averages."

Although some people might equate SLG with power, I agree with commenters above that most of us realize that it reflects total bases. The insight that each base is non-linearly associated with run production is an important one, and so weighted slugging seems like an obvious improvement.

I disagree that we need some sort of measure of power. Fans have been fascinated with it for ages, probably since Ruth, but it's not important to outcomes. Number of bases, sure. That's why OPS is useful. In the end, as my man the Idiot points out, size doesn't matter. Only results matter, and hit distance is probably only loosely correlated with outcomes over and above the bases-based stats.

quite enjoyable…fuckin' lovely