The great 28 (part 2 of 2)

Well, here it is.

Every so often here at THT, I ask SG from the Replacement Level Yankee Weblog to do a favor. He has a computer program set up that can run 1,000 simulated seasons using Diamond Mind Baseball with any 28 teams from all baseball history. About once a year he kindly uses it to answer a question of mine on baseball history.

Last week’s column noted the current question, the one I’ve been building toward ever since I’ve been here at THT: what is the best team of all time? A week ago, my column listed the 28 teams I asked him to sim with and explained why I wanted those teams.

This is the week we get to the final, definitive, and all-time unarguable answer—which, of course, is not final, definitive, or inarguable. Part of the fun in the debate on the best team in history is that there is no clear-cut answer.

Okay, enough of it all this. Let’s get to it. With the teams picked in last week’s column, how did it shake out? Who won the title as the best team ever? Here are the results, their record and runs scored/allowed in the sims, and their actual & pythag records:

Team	           W	L	Pct	RS	RA	Actual	Pythag
1927 NYY	95.4	66.6	0.589	915	750	110-44	109-45
1998 NYY	91.7	70.3	0.566	761	676	114-48	108-54
2001 SEA	91.2	70.8	0.563	739	667	116-46	109-53
1939 NYY	90.7	71.3	0.560	828	717	106-45	111-40
1998 ATL	88.9	73.1	0.549	734	681	106-56	106-56
1995 CLE	88.8	73.2	0.548	814	758	100-44	 93-51
1902 PIT	86.8	75.2	0.536	774	710	103-36	103-36
1969 BAL	84.0	78.0	0.519	738	709	109-53	110-52
1906 CHC	82.5	79.5	0.509	693	665	116-36	115-37
1953 BRK	82.3	79.7	0.508	832	819	105-49	 99-55
2004 STL	81.5	80.5	0.503	708	725	105-57	100-62
1896 BAL	81.3	80.7	0.502	761	746	 90-39	 87-42
1942 STL	80.7	81.3	0.498	743	737	106-48	107-47
1910 PHA	79.9	82.1	0.493	727	723	102-48	103-47
1953 NYY	79.2	82.8	0.489	653	665	 99-52	101-50
1954 CLE	79.1	82.9	0.488	685	695	111-43	104-50
1912 BOX	78.9	83.1	0.487	698	711	105-47	102-50
1986 NYM	78.8	83.2	0.486	723	752	108-54	103-59
1984 DET	78.0	84.0	0.481	751	792	104-58	 99-63
1961 NYY	77.2	84.8	0.477	743	787	109-53	103-59
1975 CIN	76.2	85.8	0.470	759	803	108-54	107-55
1929 PHA	76.1	85.9	0.470	719	758	104-46	100-50
2011 PHI	76.0	86.0	0.469	647	706	106-60	103-59
1921 NYG	75.1	86.9	0.464	757	799	 94-59	 95-58
1897 BOB	74.3	87.7	0.459	785	844	 93-39	 91-41
1972 OAK	71.8	90.2	0.443	677	763	 93-62	 97-58
1968 DET	71.6	90.4	0.442	726	817	103-59	103-59
1988 OAK	69.9	92.1	0.431	677	788	104-58	100-62

The 1927 Yankees? Well, that’s a classic pick, now isn’t it?

Growing up, I heard a ton of how the Yankees were the greatest team ever. They were the most famous and successful team, featuring the greatest player of them all, Babe Ruth. Back then, only two teams had ever won more games in a regular season than the Yanks, the 1906 Cubs and 1954 Indians, and both of them arguably were worse than the Yankees.

After all, those two teams each lost their respective World Series—the Indians were swept—while the Yanks won theirs. Also, the 1900s NL and 1950s AL were fairly weak leagues with less talent than the opposing league. You can’t say that about the 1920s AL, and the 1927 Yankees swept their World Series.

Still, all that was back when I was a kid. Since then, two more teams have won more than 110 games in a year, the 1998 Yankees and 2001 Mariners. I figured one of those teams would come on top. They don’t, but they sure come close, finishing second and third.

But the 1927 Yankees don’t just come in first, they win easily. They top the 1998 Yankees by 3.7 games, which means the 1998 Yankees are closer to the sixth-place 1995 Indians than they are to the 1927 Yankees.

The simulations give them by far the best offense. The 1927 Yankees scored 915 runs while the other 27 teams averaged 735 runs per game. That’s an advantage over 1,000 simulated seasons of over a run per game against the finest competition out there. Not bad. With 915 runs scored, they average over a half-run per game more than the second-most explosive offensive, the 1953 Dodgers. Not bad at all.

The award for best pitching is a much closer battle. In fact, two teams tie with 665 runs for the fewest allowed: the 1906 Cubs and 1953 Yankees. The 2001 Mariners are right behind with 667 runs allowed.

It’s interesting, the 1953 World Series is the only one with both its teams represented here. The Dodgers have, according to these sims, the second-best offense while the Yankees tie for best pitching. The Dodgers also have the second-worst pitching (ahead of only the 1897 Braves) while the Yankees have the second-worst hitting. Neat bit of symmetry there.

But in the end, there are the 1927 Yankees, and then there is everyone else. They finished in first place in 376 of the 1,000 sims and came in second place 200 times. They finished worse than 18th place only one time. Every other squad did that at least 10 times.

Eras and results

The 1927 Yankees’ dominance in this exercise is even more interesting when you break it down by era. In general, these sims did, in fact, have some sort of era adjustment.

Almost all of the other teams up top are from recent times, so the 1927 Yankees (and the 1939 Yankees) really stick out near the top of the leaderboard.

A Hardball Times Update
Goodbye for now.

It looks like there were adjustments for era and league strength. That’s the best way to understand what happened to teams like the 1897 Braves and 1954 Indians. The Braves were a great team but sputtered here. With the 1954 Indians, the AL was a far inferior league to the NL in that decade, which hurts their score.

Dividing it up by era

Well, if era matters, let’s divide it up by era. We’ll take the 28 teams and move them into four quadrants based on period. It breaks up as follows: 1896-1921 teams for the first seven, 1927-1954 for the second bunch, then 1961-1986, and finally 1988-2011 for the most recent squads.

Here’s how the oldest teams did:

Team	          W	L	Pct	RS	RA	Actual	Pythag
1902 PIT	86.8	75.2	0.536	774	710	103-36	103-36
1906 CHC	82.5	79.5	0.509	693	665	116-36	115-37
1896 BAL	81.3	80.7	0.502	761	746	 90-39	 87-42
1910 PHA	79.9	82.1	0.493	727	723	102-48	103-47
1912 BOX	78.9	83.1	0.487	698	711	105-47	102-50
1921 NYG	75.1	86.9	0.464	757	799	 94-59	 95-58
1897 BOB	74.3	87.7	0.459	785	844	 93-39	 91-41

In general, these guys did a little below average. They combined for 558.8 wins, which is a .493 winning percentage. That isn’t too extreme, but then again, almost half of the 28 teams in general finished within three games of .500.

It’s a surprise that the 1902 Pirates actually top the 1906 Cubs. That could be a fluke of the computer. Or it could point to a moderate difference in league strength.

Really, they were two tremendous teams. While the Cubs have the best real-life winning percentage here, the Pirates have the second-best. The Cubs had the best actual pythag record, and the Pirates the second-best. The Cubs’ 116-36 record in 1906 looks a lot more gaudy than Pittsburgh’s 103 wins in 1902, but, then again, that was before the 154-game schedule. If the Pirates had 152 decisions like the Cubs, they were on pace to finish 113-39. No, it’s still a bit short, but it’s close.

Let’s look at the next group of seven teams, the 1927-54 gang:

Team	          W	L	Pct	RS	RA	Actual	Pythag
1927 NYY	95.4	66.6	0.589	915	750	110-44	109-45
1939 NYY	90.7	71.3	0.560	828	717	106-45	111-40
1953 BRK	82.3	79.7	0.508	832	819	105-49	 99-55
1942 STL	80.7	81.3	0.498	743	737	106-48	107-47
1953 NYY	79.2	82.8	0.489	653	665	 99-52	101-50
1954 CLE	79.1	82.9	0.488	685	695	111-43	104-50
1929 PHA	76.1	85.9	0.470	719	758	104-46	100-50

Okay, we know the 1927 Yankees finish first, but it’s still impressive when you put them in context. Not only do they have a comfortable lead over all 27 other challengers, but from the 1890s to 1950s, only two teams finish within 10 games of them, the 1902 Pirates and the 1939 Yankees.

Those 1939 Yankees are one of the great overlooked teams. First, due to rainouts and ties, they finished the season with only 151 decisions. So their 106 wins, which is impressive enough as is, actually undersells them. If they finished with a full compliment of 154 decisions like the 1927 Yankees, they’d likely have 108 wins, which is pretty close to Ruth’s Yankees.

Not only that, but the 1939 Yankees actually won fewer games than their pythag record projects, and not just slightly. By runs scored and allowed, the 1939 Yankees should’ve gone 111-40, which is to say they should’ve won more contests than the 1927 Yankees despite having fewer games. Okay, that’s nice. They won the pennant by 17 games. They were a legitimate contender to win this all, and their fourth-place showing in the field of 28 is still damn good.

Thanks to the twin Yankee dynamos of 1927 and 1939, this group of seven won 583.5 games in all, an average of 83.4 per team.

I really can’t explain why the 1929 A’s finished as poorly as they did. Their last-place finish in the group of seven is something of a fluke. Just missing this group of teams is the 1921 Giants, who won even fewer games. Actually, the 1929 A’s wouldn’t finish in last place in any of the other three groupings. In fact, 76.1 wins wouldn’t finish in next-to-last in any of the other bunchings.

Not only does the 1927-54 group have two of the top finishers, but it also lacks any real cellar dweller in the full group of 28.

The same cannot be said for the third group, the seven teams from 1961-86:

Team	          W	L	Pct	RS	RA	Actual	Pythag
1969 BAL	84.0	78.0	0.519	738	709	109-53	110-52
1986 NYM	78.8	83.2	0.486	723	752	108-54	103-59
1984 DET	78.0	84.0	0.481	751	792	104-58	 99-63
1961 NYY	77.2	84.8	0.477	743	787	109-53	103-59
1975 CIN	76.2	85.8	0.470	759	803	108-54	107-55
1972 OAK	71.8	90.2	0.443	677	763	 93-62	 97-58
1968 DET	71.6	90.4	0.442	726	817	103-59	103-59

Huh? In general, there is a trend in this simulations exercise that more recent teams do better. But this bunch missed the memo. Only one squad finished with a winning record and overall they averaged 76.8 wins, easily the worst of any of the four groupings.

You really can’t blame arbitrary endpoints too much here, either. Just missing the group in either direction are the 1954 Indians and 1988 A’s. The Indians did poorly, and the A’s had the worst record among all 28 teams.

So why did these seven teams do so poorly? Let’s go case by case. The 1972 A’s doing poorly isn’t a surprise. They’re in here because they won three straight world titles. They were a gritty and consistent bunch, but they were never historically great in any single season.

Also, some teams are a bit one dimensional. The 1975 Reds had one of the greatest offenses in history, but pitching wasn’t their thing. They can run roughshod over a real league, but when the competition is the best teams ever, their problems are more pronounced. For that matter, the 1961 Yankees are heavily dependent on the longball. Bill James once wrote a critique of them arguing that they weren’t as great as everyone assumes based on their 109-win season because they were too one-dimensional.

The 1984 Tigers and 1986 Mets were both really great teams, but in both cases their real-life record exceeds their projected pythag record by five games. In a computer simulation, pythag records probably mean more than real life marks because the computer is going off the base numbers. The teams in this exercise usually do have a better record than pythag record, but not by five games. The others average a difference of +2. (That said, I think the 1984 Tigers and 1986 Mets are better than what the simulation says of them).

That leaves the 1968 Tigers and 1969 Orioles. Well, there’s an answer here—overall league strength. The 1950s and 1960s were a time when the NL integrated much more extensively than the AL and had a better overall quality of play. There’s a reason why the NL won virtually every All-Star game for quite a while. Put these teams in the NL and see what happens.

Actually, the 1969 Orioles are a legitimately great team by any definition. They do, after all, finish with the best record in this bunch. Actually, that undersells them. They finish the simulations with the best record of any team between the 1939 Yankees and 1995 Indians. For a half-century, they were the best team in all baseball.

That just leaves the most recent period, teams from 1988-onward:

Team	          W	L	Pct	RS	RA	Actual	Pythag
1998 NYY	91.7	70.3	0.566	761	676	114-48	108-54
2001 SEA	91.2	70.8	0.563	739	667	116-46	109-53
1998 ATL	88.9	73.1	0.549	734	681	106-56	106-56
1995 CLE	88.8	73.2	0.548	814	758	100-44	 93-51
2004 STL	81.5	80.5	0.503	708	725	105-57	100-62
2011 PHI	76.0	86.0	0.469	647	706	106-60	103-59
1988 OAK	69.9	92.1	0.431	677	788	104-58	100-62

This is the best bunch of teams, averaging 84.0 wins per squad. And they do that despite featuring the last-place team among the entire field of 28, too. The six most recent teams average 86.4 wins each. Four of the top six teams ever were from this period.

And the squads do well despite a remarkable difference between their real-life and pythag records. I just noted a little bit ago that a pythag record should mean more than a real record in a computer simulation. Well, the first 21 teams’ pythag marks all exceeded their actual records by a combined 29 games. These seven most recent teams exceeded their actual mark by 32 games.

So how come they did so well here? Well, first off, if you’re really that good, it doesn’t matter that much if you exceed your pythag projection. For example, the 2001 Mariners have a pythag mark of “only” 109-53. Yeah, it’s seven games below their actual record, but it’s still one of the best pythag records of the last half-century. It’s slightly worse than the 1969 Orioles’ 110-52 pythag record, but the 2001 AL was better than the 1969 AL.

The most recent years of baseball are also the best in overall league quality, so when a team fights its way through a season with a historically great record and nearly as great a pythag mark, they’ll score among the best in the 1,000 sims, which is what happened.

This makes it all the more surprising that none of these squads tops the 1927 Yankees for best record ever. I suppose the pythag/actual record deviation did them in. The 1927 Yankees had a pythag record of 109-45, which works out to 115-47 over 162 games. The 1998 Yankees have a pythag record of 108-54.

None of this is definitive. As noted up top, the great thing about debates like, “Who is the best team ever?” is that you’ll never have a perfect answer. But un-answerable questions are the most interesting to answer and the most fun ones to spend time researching.

References & Resources
This article would not be possible without SG of the Replacement Level Yankee Weblog and his computer program that simulates 1,000 seasons.


13 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Todd
11 years ago

The 2004 Cards finished the sim with a slightly negative run differential, but a record slightly over .500. The 1942 Cards are the opposite. What’s going on there? It’s one thing for that to happen in a single real season, but over the course of so many sims, surely that should average out?

Hizouse
11 years ago

Thanks for doing this, Chris; interesting.

I assume the seasons featured 6 games each against the other 27 teams.

I’d be interested if you could add column(s) to the first table for times coming in first, best record, and worst record.

Do you know how DMB handled the league and era adjustments?

Chris Jaffe
11 years ago

Ken – Don’t know how the DH works.

Todd – I don’t know how it works, but in both cases the runs & wins are right around .500.

Hizouse,

Yeah – 27×6.  I don’t know about era adjustments. 

Here’s best/worst wins for all 28 teams in the 1,000 sims:

Season   MIN   MAX
BRO53   62   105
CLE54   58   97
NYA39   70   108
NYA53   60   98
PHA10   63   102
PHA29   54   95
SLN42   61   105
BLN96   64   99
BOS12   64   95
BSN97   53   96
CHN06   64   103
NY121   54   94
NYA27   73   113
PIT02   66   111
ATL98   70   106
CLE95   72   107
NYA98   71   112
NYN86   59   101
PHI11   53   96
SEA01   68   112
SLN04   54   105
BAL69   64   100
CIN75   58   96
DET68   51   90
DET84   58   99
NYA61   57   97
OAK72   51   94
OAK88   46   89

SG
11 years ago

Was the DH used for modern A.L. Teams? 

DH was used based on home team rules, so all AL teams that had a DH got to use it at home and against other AL DH opponents on the road.

Do you know how DMB handled the league and era adjustments?

As far as the era adjustment, Diamond Mind has a complicated system of event tables that are created for each player based on the park and era they played in, and then they also have a neutral era which is designed to allow you to do these cross-era type match-ups.

Ken
11 years ago

Thanks SG for the response, that’s exactly how I would’ve handled the DH rule. Love DMB, though I haven’t played it in a few years. Great for doing these kind of hypothetical matchups. Thanks guys!

David
11 years ago

How about relief pitching?  The modern teams have relief specialists that the older teams do not.  Just like the DH, this gives them an advantage in roster construction.

Also, how about playing small ball?  Sabermetrics tells us this is not the best strategy, but the older teams relied on it and modern teams do not.  How did the sims handle this strategy?

Ken
11 years ago

Was the DH used for modern A.L. Teams?  To me it’s a must that they should be as the teams were constructed around the rule. Great study!

Graham
11 years ago

Spreads between min/max:

Biggest -> SLN04 = 51 win spread

Average = 40.25 win spread

Smallest -> BOS12 = 31 win spread

sleepyirv
11 years ago

Am I the only one having problems trying to get on the rlyw site?

Philip
11 years ago

Chris, when you wrote part 1, I urged to replace those juice-injected 1988 A’s with the 1974 Dodgers. You had expressed doubts about Oakland. How fitting the A’s finished last.

Do you know how the lineups were handled? Was it Combs, Koenig, Ruth, Gehrig, Meusel, Lazzeri, Dugan batting in that order c. the 61 times they did in real-life? Or was DMB allowed to select its own batting order?

How did the head-to-head breakdown? Was Murderers’ Row able to play .700 ball in the 6,000 times they played the Juice Brothers?

Not surprised the 1929 A’s didn’t fare well. Am surprised their offense didn’t put more runs across.

Not surprised Miller Huggins’ crew took it. I would have thought the 1906 Cubs would have fared a little better, but benches were weak in those days, roster sizes being smaller. Plus, many of the teams they faced have some pretty good catchers, which no doubt hurt the Cubs’ running game. Would be interesting to see how those Cubs fared solely against their contemporaries in head-to-head competition. My hunch is they had the best record in those 36,000 games.

As a Red Sox fan, not too disturbed the 27 Yankees won it. Certainly much better than the 98 Yankees winning it. Plus, the 1912 Red Sox finished higher than the 1961 Yankees!

Great study!

Moeball
11 years ago

Have been fascinated by many things about ‘27 NY. Dominance not just offensively but in pitching, too.

How many teams in history have had top 3 in league in runs scored and HRs? (No others that I can think of)

5 different players were among league leaders in various offensive categories (in addition to the Big 3 of Ruth, Gehrig, Combs – Lazzeri and Meusel also had excellent years).

Actually had very good hitting pitching staff – position players on A’s had better team average than Yankees but NY pitchers outhit PHI pitchers by a good-sized margin.

Held team record for SLG until Red Sox in what, 2003?

Top 3 pitchers in league in ERA; matched by ‘54 Indians, I think, but not sure if anyone else matched that.

And in a really freaky display of dominance – yes, we know Ruth outhomered all other AL teams that year.  But the team Home/Road splits are truly bizarre. On the road – NY hit 75 HRs, allowed only 12.  Yes, that is a ratio of more than 6 to 1 and I don’t think any other team in history has done even half that.

Chris J.
11 years ago

Moeball – the 1927 Yankees also have the greatest home run differential of any team ever: they hit 158 while allowing just 42, +116 in homers.

2nd most is …. the 2010 Jays?  Huh: +107 homrers (257 to 150)

Patrick M. Premo
9 years ago

Chris,

I recently heard about your baseball simulations and have read them all to date — great ideas with very interesting results. Your write-ups are first notch.
Have you ever considered doing simulations with other than the major leagues? That is, Negro Leagues, Federal League, etc.? (Perhaps the sim engine does not include these teams, but it would be fun it it did.)

Keep up the great work — drop me a line if you can.

Pat
Patrick Premo
Allegany, NY