I think there’s a problem with: “)PS says B performs about 21.4% better than A. But OTS says B performs 50% better. Quite a difference! (And one that, while extreme, is well within the range of plausible real-world values.) So adding instead of multiplying causes the true scale of differences to be compressed, sometimes badly.” and the underlying math.

Comparing .850 to .700 doesn’t make sense because the scale of ability to get and keep a major league job doesn’t start at .000 in terms of OPS. For a position player, it starts at replacement level. WAGging that RL OPS is around. 600, the difference is between .250 and .100 or 150%.

Almost every problem with OPS led to using linear weights and wOBA. I’d just recommend using this.

Short answer: DOESN’T FACTOR IN GAMER-TUDE.

What’s OPS?

Signed Gianta Management

Making the assumption that this posting’s focus upon OPS is meant to be:

A. Evidence of the Giant’s failings…
B. More specifically, evidence of Sabean’s failings…

My opinion would be that team OPS is a more important than individual OPS. Clearly the Giants were one of the worst run scoring teams in MLB, and at my last glance, the worst team OPS in MLB. So small sample size supports that, and I will trust that Bill James, Sandy Alderson, Paul DePodesta, and Billy Bean did the homework properly and say that this is within expected trends. And since any discussion about runs scored, and OPS, and team building goes back to Moneyball and Sabermetrics, it would be incorrect to use PA instead of AB in statistical analysis since any out is considered to be bad. So your use of PA is bad Moneyball/Sabermetrics.

Based purely upon (runs scored = success) the Giants are an anomaly. The record speaks for itself. The Giants were the 9th best record in MLB this year out of 30 teams. However, I would say that more correctly they were 5th in the NL this year out of 16 teams since these were the teams they played mostly head to head. So, actually I don’t think (runs scored = success) is true. How about:

Team (RS – RA)
____________ = Win%
MLB (RS – RA)

This takes into account runs allowed as being an important ingredient. Since, in a baseball game, the team that scores more runs than the other wins the game, I think this is more accurate. Plus, a positive delta runs will be directly proportional to win%. Since we are now also considering pitching and fielding it is more accurately measures a successful team.

I think the Giants have a huge amount of room for improvement offensively and I am not very encouraged by Bochy’s commitment to more small ball, nor Sabean’s thinking that the addition of a patient hitting role model will help the younger players morph themselves into OBA guys. However, this team is much better than simple offensive analysis tends to say.

If you were simply trying to wonk out on OPS, then fuck it, have a good time. Your post is way too complicated for me.

Math hurts my brain.

wOBA

He’s a bum!

It’s popular because it’s stable, easy to calculate and correlates with runs scored on a team level very well. There’s lots of ways to improve on it, but I’d wager that the differences between the various methods aren’t huge.

Weighted On Base Average or wOBA

On-base percentage is a great statistic because it tells you something important, and in a clear language: at what rate did this player reach base? It doesn’t tell you how far he reached base (second base? third? home?), but only whether he did or did not.

Slugging percentage is another great statistic because it tells you something important, and in a clear language: how many bases did the batter gain for himself per at-bat? It doesn’t consider walks as either a positive or negative event (it simply strips them away as if they don’t exist). It also tries to establish the importance of the single and HR by weighting the HR four times as much as the single.

We have one statistic that is deficient in one area, and another one that is deficient in another. Why not simply combine them as: OBP plus SLG, and call it this new-age statistic named OPS? Might this statistic allow the deficiencies in OBP and SLG to cancel each other out? Let’s see.

From the preceding section, we know the run values of each event. For example, we know that the run value of the HR is 1.4 runs above average, and 1.7 runs above the run value of the out. In rate measures, like OBP, the value of the out in the numerator is zero. If we recast the run values of the most common events relative to the out (rather than relative to the result of an average plate appearance), we get the following:

HR 1.70, 3B 1.37, 2B 1.08, 1B 0.77, NIBB 0.62.

Those numbers are the values of each of our events (again, relative to an out, which now has a value of zero). If we apply these weights to the statistics of a league-average hitter, and divide by plate appearances, we end up with a rate of almost 0.300. This is a fairly convenient number for an average, but we can do better. Since we like OBP as a measure of a batter’s effectiveness, let’s scale our new statistic so that the resulting values are similar to OBP values. It turns out that, if we add 15% to this 0.300 figure, we get the league-average OBP. Therefore, we will add 15% to the weights of each event and define our new statistic as follows:

(0.72xNIBB + 0.75xHBP + 0.90×1B + 0.92xRBOE + 1.24×2B + 1.56×3B + 1.95xHR) / PA

Note: Depending on the specific analysis, the PA term (plate appearances) may exclude bunts, IBB, and a few of the more obscure plays.

Do we really need another statistic? Yes, we do. Instead of trying to take two statistics (OBP, SLG) and combine and correct their flaws in the hopes of getting one number, we prefer to start from scratch. Furthermore, by recasting the number onto the OBP scale, it makes it much easier for the reader to get a grasp on the number. wOBA is weighted on-base average (we call it an average rather than a percentage). When you look at wOBA numbers throughout the book, just think OBP, and you’ll be fine. In other words, an average hitter is around 0.340 or so, a great hitter is 0.400 or higher, and a poor hitter would be under 0.300.

If you are a little more experienced with run values, you might have figured out the following:

Run value per PA above average = (wOBA for player – wOBA for league) / 1.15

So the run value chart, which we presented in the previous section, and the wOBA statistic defined in this section are directly related.

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

They rely on OOPS.

Any measure whatever is just so much tavern-talk fodder unless it can provide absolute, not relative, results. By that I mean that the end product of the metric on a team basis had better be actual runs scored or runs yielded, and that individual men’s performances, so measured, can be combined to yield a team total that—again—equals actual runs scored or yielded. If you cannot readily extract actual runs values from a measure, it is just so much vaporing.

Such measures exist. Some are better than others, though how that is reckoned depends on who is defending which measure, but the essence is that all such absolute metrics give results pretty similar to one another (they have to, else they are scarcely valid, since the output is a number, runs, that can be and is measured against real, knowable results).

OPS is a relative, not absolute, measure. Of the common set of such measures, it is the best, but the point here—which was not specifically aimed at the Giants’ current situation, just as general background—is that “better” is not “good”, and that basing all reckoning, especially of individual men, on OPS can lead to significant mis-evaluations, especially when comparing men.

The runs equation I speak of takes a collection of stats and cranks out the number of seasonal runs that “should” be scored by a team producing at those rates. The actual number of games over which the raw stats are collected is immaterial, because everything is converted to a rate before being used. (Except, of course, for the usual caveat about sample sizes.)

Because the result is always a seasonal run total regardless of quantities of the raw data, the formula can be applied to any coherent collection: a team’s stats for one game, the entirety of a league at any point in a season—or over many years—or, finally, to the numbers for a particular player, whether seasonal or career or whatever. It says that a team playing at that quality of play is thus-and-so good for scoring runs.

(That also has the advantage that the result is intuitively meaningful, because anyone who follows the game understands that, say, 570 runs is a piss-poor season, while 925 runs is pretty good.)

If one uses the formula on the batting roster of a team on a man-by-man basis, then weights those individual “seasonal runs” numbers by percentage of playing time, the net result for the team—as assembled that way, not by using the team stats per se—will be a bit high, usually by something very close to 1.5%. Of course, if one instead weights each man’s OBA and TBA and combines those data for a team OBA and TBA, the result will be correct, since we are back to actual team data. The reason for the discrepancy is that the product of the sums is not, in general, equal to the sum of the products.

To clarify that possibly cryptic remark, consider these two sets of data:

M = A x B
N = C x D

If we average A and C, then average B and D, and multiply those two averages, we will not—except by chance—get the same result as if we average M and N.

  6 = 2 × 3
20 = 4 × 5

Average 2 and 4 and you get 3; average 3 and 5 and you get 4; multiply 3 by 4 and you get 12. But average 6 and 20 and you get 13.

To keep it very simple conceptually, assume that each batter who appeared at all got the same number of plate appearances as all the others, so that no weighting is involved: we can just straight average all the individual player numbers produced by the formula. Because each such number is—to oversimplify—the product of two base data, OBA and TBA, such averaging will produce a result different from what you’d get from first separately averaging those two components, then multiplying those two averages (which yields the correct team result).

The crux, though, is that the differential between simple averaging of by-man results and of proper, full calculation is only a very small one, strongly suggesting that the measure is quite satisfactory for rating individual players, even in a “team context”.

I hope that clarified, rather than further muddying . . . .

I don’t understand numbers at all. Can I just keep rating players in who I would personally like to be friends with?

NERDS!

/ General Manager Brian Sabean

I intensely dislike the way Mr. Sabean has assembled his teams, and the way Mr. Bochy has managed them. That does not mean that I think either is some Dark Lord of Baseball, lusting only to do evil. Each is a hard-working man, doing the best he can by his lights. I would like little more than a chance to sit down with either, or both, for a couple of hours of talk; it is my belief—perhaps naive—that any man can be made to see reason if that reason is explained clearly and logically from first principles. Consider ex-player Billy Beane . . . .