Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Primate Studies > Discussion
Primate Studies
— Where BTF's Members Investigate the Grand Old Game

Wednesday, April 11, 2001

The All-OBP Team

Don exams the practical limits of getting on base.

Seeking the Practical Limits of Offense

During the long run of the BBBA series, Brock Hanke and I were blessed   with readers who were both loyal and indulgent?more than that, they were pretty   darned smart. While they were interested in information about the season just   past and the season to come?one aspect of the book’s subject matter?they also   responded to the historical content accompanying the more topical material.   We’d often get interesting notes written on the back of an envelope or an order   form?communications that would send us off on some area of research that we   hadn’t thought of previously.

That tradition was continued in 2001, and one of the examples of this comes   from long-time BBBA reader Duane Thomas, who has bought enough copies   of BBBA and its predecessors over the past dozen years to remember our fixation   with "walkmen"?that special breed of batters who know how to keep   their bats on their shoulders.

Duane wrote in this year with the following question:

"If I had the following batting order for 162 games, how many runs would   my team score?

1.       Eddie Stanky 2b (1945)
  2.       Richie Ashburn cf (1955)
  3.       Eddie Yost 3b (1950)
  4.       Ferris Fain 1b (1951)
  5.       Roy Cullenbine rf (1946)
  6.       Eddie Joost ss (1949)
  7.       Elmer Valo lf (1949)
  8.       Eddie Lake dh (1945)
  9.       Wes Westrum c (1951)

a)       700-800 runs
  b)      800-900 runs
  c)       900-1000 runs
  d)      1000-1100 runs
  e)       1100+ runs
  f)        games go on too long"

Now before I go on to try and answer this question for Duane, I urge all of   you out there reading to stop right where you are, go to the primer "add   a comment" option (one of the things that makes this site unique, I might   add?), and choose between a) and e).

I think we can safely say that f) would also be the case, since we have the   largest collection of true "walkmen" in one lineup here. A "true"   walkman is a batter who draws walks on his batting eye, not on his fearsomeness   at the plate. Babe Ruth, Ted Williams and Frank Thomas are all "walkmen,"   but something close to half their walks occur because pitchers are afraid to   give them a good ball to hit.

That’s not the case with Duane’s team. There is one Hall of Famer in this lineup?Richie   Ashburn. The rest of the batting order features the cream of the post-WWII walk   boom?that temporary phenomenon between 1947 and 1955 when the base on balls   was king (for a little while, at least).

What’s intruiging about this question is that Duane has picked a team that   could probably be assembed on the cheap even today; it certainly could have   been back when theese players were in their heyday. This is a well-traveled   starting nine?only the catcher (Westrum) stayed with the team he came up with   for his entire career.

On-Base Times Slugging (OBP x SLG) and Run Scoring

Let’s try to answer Duane’s question. There is a simple tool, touted for years   by Brock Hanke, which can help us here. While many of us are now intimately   familiar with OPS (on-base plus slugging), it’s far less well-known that multiplying   these offensive components creates a useful runs per at-bat ratio that   can be used to project run scoring.

Let’s walk through it one time just to get the hang of it. Let’s say your team   has an aggregate OBP of .330 (that’s not good in today’s era of inflated offense,   but pretend you’re a long-suffering fan in a "small market" like Philadelphia).   Let’s say that the team is also underpowered (too many Doug Glanvilles), and   has a combined SLG of just .400 (the 2000 NL had an aggregate SLG of .432, just   for comparison).

Multiply .330 by .400 and you get a runs per at-bat ratio of .1320 (not too   far off MLB’s historical average, by the way). We’ve found that by multiplying   this figure by the team at-bats, and then multiplying the result by 1.035, you   get a pretty accurate projection of runs scored.

In this case, the calculation is (.330 x .400) = .1320 x 5500 = 726 x 1.035   = 751 runs.

Whether this is a good or a bad offense depends on the league context, of course.   It’s good in 1968, for example. But it’s not so good in 2000, and probably not   in 2001, either.

We’ve come to view the calculation above as expressing a range for run scoring   (a team with the above OBP x SLG will tend to score between 726 to 751 runs).   Teams fall into this range around 70% of the time, and when they are out of   the range, they’re not far out of it. There are more accurate methods, but none   are as simple and as easy to use as this one.

Armed with this tool, let’s tackle Duane’s question.

First, let’s take a look at some of the key offensive statistics for the players   on Duane’s team in the years he has selected to use. (Like the Legends of   Baseball league that you’ve read about in my weblog, Duane has chosen peak   seasons for his players.)

Player?????? Year??? BB??? BA?? OBP?? SLG
Stanky?????? 1945?? 148? .258? .417? .333
Ashburn????? 1955?? 105? .338? .449? .448
Yost???????? 1950?? 141? .295? .440? .405
Fain???????? 1951? ? 80  .344? .451? .471
Cullenbine?? 1946??  88? .335? .477? .537
Joost??????? 1949?? 149? .263? .429? .453
Valo???????? 1949?? 119? .283? .413? .404
Lake???????? 1945?? 106? .279? .412? .410
Westrum???   1951?? 104? .219? .400? .418
TOTAL????    1040?       .290? .426? .425

The salient feature of this team, of course, is its ability to get on base.   If this team could duplicate the chosen seasons, it would shatter the all-time   record for team OBP (.385, held by the 1950 Boston Red Sox and the 1921 Detroit   Tigers).

That’s reinforced by the walk total. Those 1040 walks were achieved in only   5300 plate apperances (Fain, Cullenbine, and Westrum were not full-time players   in the years chosen). Even so, this figure would also shatter the all-time record   for walks (835, held by the 1949 Boston Red Sox).

Extrapolating that figure out to a full season’s worth of plate appearances   (around 6300) gives us the amazing and probably impossible total of 1209. (Why   is that impossible? I’ll talk about that a bit later.)

Using OBP times SLG, we get a runs/at-bat ratio of .1811. Multiplying that   by 5100 at-bats (we subtract the number of walks from 6300 to obtain an estimate   of at-bats) and adding in the 3.5% boost, we come up with an estimate of 956   runs scored for Duane’s team.

That’s excellent, but it’s not especially close to the all-time record for   runs scored (we’re talking twentieth century): 1067, by the 1931 New York Yankees.

Why doesn’t a team with .040 more OBP than any team since 1901 score more runs?   Lack of power. That .425 SLG is better than average, but much of it comes from   a high batting average (.290). Duane’s walkman team only projects to hit in   the realm of 110-120 homers, which would place them at the bottom of the team   rankings now. Back in the 50s this would be a middle-of-the-pack total.

Another tip-off in terms of this team’s imbalance can be seen in its runs scored   and RBI totals. For the seasons Duane selected, the nine members of the walkman   team scored 813 runs, but drove in only 539. Only Westrum, who batted eighth   most of the time for the Giants due a low batting average, drove in more runs   than he scored. This is a team of table-setters, not table-clearers.

Something would have to change in the hitting dynamic with a team of this type.   One or more hitters would need to swing the bat more in order to drive in runs.   Pitchers would adjust, and throw more strikes, because these players are all   taking advantage of game and lineup contexts in terms of drawing excess walks.

In short, it would be impossible for this team to do as a unit what the members   of the lineup were able to do individually.

Oh, they’d still draw a lot of walks?they’d break the record with some to spare.   But they’d probably lose at least 20% of their margin, and finish with about   950 or so.

Holding all of the other factors even, the likely OBP for this team would be   more like .398 than .426.

Suprisingly, that reduces the estimated runs scored by only about 30-35 runs   (924 instead of 956). Why is that?

Because all that excess OBP, derived from walks, is less valuable in terms   of scoring runs than the other elements of offense. While it is fashionable   to call RBIs "a meaningless stat" and to deride them due to their   "situation dependent" nature, the truth of the the matter is that   offense has two components?getting on base (OBP) and driving runs in (SLG)?that   must work together in some kind of balance in order to be optimized.

The OBP x SLG model gives us a simple and reliable overview for that. More   sophisticated and specialized methods can eliminate distortions that have been   found at the extremes of such projections, but in the ranges that occur for   team performance, you don’t really need anything more complicated.

Career Year Reality Check

One final note about Duane’s team. It’s a fanciful idea, and one that’s fun   to contemplate. But it’s impossible?not just because of the theoretical elements   noted above.

Simply put, teams don’t reach a monolithic peak all at once. Duane’s walkman   team is a collection of players having their best years (or very close to them)   in the same year. This just doesn’t happen, except in simulations. And even   there, simulations such as Diamond   Mind Baseball won’t permit a team whose players combined for a   .440 OBP to produce anything close to that when the simulated season is run.

The reason for this is what Bill James called The Law of Competitive Balance.   This law operates in a different manner on teams than it does on individuals.   One is more complex than the other, and while we can do a reasonably good job   of projecting hitters’ future performance, the variables that exist there become   harder to manage when we’re asked to project an entire lineup or an entire roster.

So while we can get an accurate measure of run scoring after the fact from   the OBP x SLG model, we have no truly reliable way to project what those values   will be before the fact at the team level.

For a team today, we’d simply add up the individual projections based on age   and performance-to-date. For a team like Duane’s, where the players are all   long retired, we could try one of two possible approaches?career averages   or career-to-date including the "peak" year.

The hope is that such an exercise in modeling will "even out the errors"   in the individual performances and produce a reasonably accurate result.

Cutting Duane’s lineup down to the career averages of its hitters takes a lot   of starch of its run-producing ability. Interestingly, it takes more away from   the SLG component than the OBP component, which probably hurts its offense more   than if it were the other way around. Here are the values:

Type?????? OBP??? SLG? Runs
Peak??? .426?? .425?? 956
Career??? .384?? .375?? 820
Diff?? -.042? -.050? -130

What’s interesting here, I think, is that we get a range for how much run scoring   fluctuation is possible from a team composed of the same players from year to   year. It’s a crude figure, to be sure, but it’s probably a reasonable rule of   thumb.

Since teams don’t usually remain static year after year, of course, such a   range doesn’t express the total potential for variability. You might think of   this as representing the "inherent variability" in run scoring.

Will anyone ever assemble a team with a .400 OBP? It’s happened three times   in baseball history?all in the same year. This feat, which occurred in 1894?baseball’s   most offensive year ever?was not due to a proliferation of "walkman"   teams, however. A hard rain of singles, doubles and triples was the cause, as   three teams?the Baltimore Orioles, Boston Beaneaters, and Philaldelphia Phillies?all   hit .330 or higher. It took that level of batting average to push OBP to .400   and over, and it’s something that we’re likely never to see again.

The only place we’re likely to assemble a .400 OBP team, alas, is in our dreams.   But that’s no reason why Duane?and the rest of us?shouldn’t keep on dreaming.

 

Don Malcolm Posted: April 11, 2001 at 06:00 AM | 8 comment(s) Login to Bookmark
  Related News:

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

   1. Scruff Posted: April 12, 2001 at 12:03 AM (#603647)
I guess C.
   2. Scruff Posted: April 12, 2001 at 12:03 AM (#603651)
This is the first time in awhile that I've heard that the SLG is more important than the OBP. Isn't the weight usually 1.4:1 for OBP, 1.25:1 if you want to be conservative?

James posed the question in one of his Abstracts, when he asked: What team would score more runs, a team of Vince Coleman's or a team of Kevin McReynolds'? He said that while McReynolds was a better player in the context of a normal team, the Coleman team would score more runs because there would always be people on base to keep the cycle going. If anything this would be enhanced by a team of OBP guys, not diminished. Throw in the extra pitches and how deep they'd be into bullpens. They would hit into a ton of DPs though, so that would lower it some.

Also, if the pitchers decided to throw more strikes (which I don't think they would), the hitters become better hitters, not worse.

As for 9000 plate appearances, there are 4374 outs in a major league season. Figure a team like this would hit into 200 DPs give or take a few. A team w/a .425 OBP would have about 7250 PA's, you'd have to be at about a .535 OBP to get 9000 plate appearances.
   3. scruff Posted: April 12, 2001 at 12:03 AM (#603654)
Sean, when a pitcher throws more strikes because he has more command it makes him a better pitcher. But when he starts throwing strikes because he's afraid of walking people usually gets hammered. That's one of the things that makes Thomas and his type so great. He's got such a great eye that to get him out pitchers have to risk putting the ball right where he wants it. The same thing would happen if pitchers tried to pitch these guys differently. I think the pitchers did the best they could to get them out as it was. I don't think that would have changed if they had all played for the same team.
   4. Scott Lange Posted: April 15, 2001 at 12:04 AM (#603667)
Can someone offer a bit of explanation about the "Law of Competitive Balance"? As I have understood it, it is the idea that players or teams who have "career years" will tend to move towards their career averages the next season. As applied to the All-OBP team listed above, it would mean that since they all had years that were better than their usual performance, they would probably not do as well next year. I don't understand why the article seems to say that they would have done worse if they had been on the same team. Similarly, I don't understand the comment that a sim like Diamond Mind wouldn't let a team of .440 OBP guys have a .440 team OBP. Why not? If you were projecting their results for the next year, then of course you would expect them to move toward their career averages, but if you are just seeing how a team with all of them on it would have done, then shouldn't they hit just as well as they actually did?
   5. Tangotiger Posted: April 17, 2001 at 12:04 AM (#603670)
This is another good example of the limitations of OBAxSLG or any other shortcut. If I've seen it once, I've seen it one hundred times. Just because OPS or OTS works in "normal" conditions doesn't mean it will work in extreme conditions like this one. I trust a properly constructed simulator more than such a simple formula which UNDERWEIGHTS WALKS to begin with. I understand the rationale of the argument, but it doesn't hold up to testing. On top of which, David Smyth used a more "fuller" formula and he came up with the same number as the simulator.
   6. Mike Emeigh Posted: April 18, 2001 at 12:04 AM (#603673)
I had a chance to look at Michael Bodell's simulator code (thanks,
Michael). Most of Michael's simplifying assumptions (no SBs, CS,
sacrifice bunts, etc.) likely have little or no net effect on run
scoring, especially for a team of this ilk. However, he does make
one simplification - no DPs - that probably does have a major
impact on scoring. I would expect a .290/.425/.425 team with normal
GB/FB ratios to ground into about 180 double plays. That would cost
such a team about 90 runs, pushing the estimate down to around
1100 runs.

The 2001 Mariners, as a quick example of what this type of team might
be able to accomplish, lead the AL in walks so far, have a decent BA
(11 points above league average) but below-average SLG. Their OTS
estimate is 73 runs, and they've actually scored 77, about 5% more
than the estimate. If the 5% OTS underestimate for high-walk/low-SLG
is real (and that's only one data point, of course, so I have no way
of knowing whether or not it's representative), the hypothetical team
of Don's example would be pushed over 1000 runs.

Regardless, I think this would be an interesting point for future
analysis.

-- MWE
   7. Mike Emeigh Posted: April 18, 2001 at 12:04 AM (#603678)
Following up on my own comment:

Using Sean Lahman's database and a couple of imaginative queries,
I looked at the OTS run estimates for two groups of teams since 1946:

-- teams which were in the upper 10% of the group in on-base
rate (BB+HBP/PA) and in the bottom 10% in isolated power
((TB-H)/AB)

-- teams in the bottom 10% of the group in OB rate and in the upper
10% in IP.

This didn't tell me much, since there were only five teams in
group A and two teams in group B. However, for the high OB/low SLG
teams, the OTS estimate was too low for four of the five teams
(BA/OBP/SLG in parens):

BRO(N) 1946: 686 est/701 act (.260/.348/.361)
PHI(A) 1948: 684 est/729 act (.260/.353/.362)
CHI(A) 1949: 650 est/648 act (.257/.347/.347)
WAS(A) 1950: 679 est/690 act (.260/.347/.360)
MON(N) 1974: 650 est/662 act (.254/.335/.350)

The two low OB/high SLG teams both played in Milwaukee, albeit in
different leagues, and the results here were quite different:

MIL(N) 1965: 739 est/708 act (.256/.310/.416)
MIL(A) 1980: 863 est/811 act (.275/.329/.448)

I ran one more test, this time using IP/OB as calculated above. I
selected 48 "low SLG relative to OB" teams (IP/OB<1) and 54 "high
SLG relative to OB" teams (IP/OB>=1.87) For the 54 "low SLG" teams,
the estimates split right down the middle; 27 were too low, 27 were
too high. For the 48 "high SLG" teams, on the other hand, 46
estimates were too high, and just two were too low.

From this, I would hypothesize that the "accelerative effect" of OBP
is correctly stated in the OTS formula, but the "accelerative effect"
of isolated power is overstated. If you want to interpret this as BB's
being undervalued, I suppose you can do that.

-- MWE
   8. Kevin Harlow Posted: April 19, 2001 at 12:04 AM (#603689)
Assuming 8.75 Outs/Game over a 162 game season, that is 4253 outs. A .426 OBP requires getting on base 3156 times, for a total of 7409 PA. In order to satisfy the simultaneous constraints of a .290 BA and a .426 OBP, there must be 1420 BB and 1736 Hits, for a total of 5989 AB. This gives a total of 5989*.426*.425*1.035 = 1122 runs per 162 games, or 1067 runs per 154 games.

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
The Ghost's Tryin' to Reason with Hurricane Season
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Syndicate

Page rendered in 0.4325 seconds
47 querie(s) executed