The BJOL piece that has fascinated me the most so far is Bill’s “Teams on Paper” from March of last year.  James devised a system to measure the underlying talent of a team’s frontline players, as opposed to measuring their performance in a given year.  The study revealed some teams that had historic talent without a lot of commensurate accomplishments.  Conversely, some teams may have weaker talent bases than we would think based on their titles.  The system encourages us to look at these teams’ successes and failures in a different light, and perhaps even change how we rank them amongst the greats.  What intrigued me right away was that you could use a method like James’s not only to study teams but to study leagues.

League quality is a crucial issue in our understanding of baseball’s past.  Let’s take the question of the greatest team of the 1970s.  The Reds won two championships with enough publicity and self-regard for six.  It has long been my position that they got more than their share of attention at the expense of other 70s dynasties with similarly gaudy accomplishments, including the Orioles, A’s, and Yankees.  Oakland won three straight titles, New York won two straight with 100-win seasons, and Baltimore won 217 games in two years, and in 1970-1971, became the only team other than McCarthy’s Bombers to lead the league in most runs scored and fewest runs allowed in consecutive seasons.  Of course the Reds had more star power in the starting lineup than just about anyone, but I thought that just blinded people to other ways a team could be strong.  A few years ago, I absolutely believed that the 1970 Orioles were a greater team than the 1975 Reds.

Now, I knew that the American League expanded in 1969, and the Orioles got to play some weak teams.  And I knew that the National League had become a stronger league than the American League anyway due to superior racial integration.  I knew in a general way that the NL in 1975 must have been stronger than the AL in 1970, but I couldn’t say by how much.  Was it 3 percent better, 15 percent better, or what?  When we form opinions, we tend to weight the things we know concretely more heavily than the things we know vaguely, for good or ill.  The Orioles won 109, 108 games—that’s concrete.  How could one form a more concrete sense of the quality of their competition?

Like some other people who are interested in this stuff, I tried to frame a league’s quality in terms of its standard deviation of winning percentage.  The tighter the league, the better the quality of play, I reasoned, lopsided leagues often being populated by under-funded, failing franchises.

I would still guess that this is true—that if you took thirty random leagues from major league history, the 15 tighter ones would be better leagues on average than the 15 more lopsided ones.  But it’s only a loose correlation; in specific cases, you would make gross errors betting on standard deviations of winning percentage to measure league strength.  For example, the National League in 1902 had a runaway champion, but the rest of the teams happened to cluster around the middle.  Therefore the 1902 Pirates always look great in standard deviation-based evaluations.  The league wasn’t strong, though; it was conspicuously weak, thanks to raids from the new rival American League.

So we need better tools to evaluate the quality of competition.  Fortunately, we are starting to get them.  Bill has given us one in “Teams on Paper.”  He estimated a team’s frontline talent by assigning a point value to each starting player based essentially on how long he played in the big leagues and how old he was at the time.  The idea was to get away from a player’s stats in a particular season, to wash out the vagaries in a year’s performance, by saying, this team had a 31-year-old shortstop who played 2200 games in his career—how much is that worth?  Discovering that career length didn’t always deliver reasonable values, James added points for MVPs, Hall of Famers, and pitchers who won big.

I decided to compile frontline talent estimates for the teams of 22 major leagues, with an eye to the quality of competition faced by some of the greatest teams.  To get the individual players’ scores, I simplified James’s method.  He used 1% of a player’s games played (or innings pitched) as the basis of the points.  I used 10% of his career win shares, as I believe they’re a better weight for career quality than games played.  I capped this point base at 50, reasoning that past 500 win shares, there’s just so much of your career-long greatness you can bring to bear on a single season.

James augmented his career-based scores with a number of other bonuses to account for short-term players who were genuine stars.  I did that by averaging the career number with win shares per 154 games (or 250 innings for a starter, 125 innings for a relief ace).  I used James’s age curves to adjust the expectations for each player’s season.  The scores generally resemble seasonal win shares (40 points is a true superstar in his prime, 30 points is MVP-ballot player, 20 is a minor star, 15 is average or so, 10 or below is weak for a regular).  For the all position players entered, the average score was 15; for all pitchers, it was 13.  A team made of all average players should score around 185.  A score of 100 indicates an abysmal team.  A score of 300 would mean you had as much frontline talent as any team in history.

For each team, I used thirteen players: their starting eight and their five most important pitchers.  That came to about 2,700 players over 22 leagues.  For position players, I went with the guys with the most playing time overall rather than the guys with the most starts at a specific position.  If a team had a clear relief ace, I made him one of the five pitchers.  If not, I didn’t try to force someone into the slot because he had 2 saves; I just used the 5^th most important pitcher.  These decisions were admittedly sometimes arbitrary, but I thought that was better than pretending guys were frontline players when they weren’t, and frankly, when you are looking at a whole league, it doesn’t really change much if you picked the wrong #5 pitcher on the 1967 Astros.

Each team’s score is the sum of the thirteen players’ scores.  For three recent AL leagues, I also entered a designated hitter, and prorated the team score down to 13 players.  So you can get a sense of the differences between James’s “Teams on Paper” method and my “Leagues on Paper” method, let’s look at an example side-by-side:

Scaled down to 13 players, my method calls the 1973 A’s a 235 team.  Bill’s method pegs Hunter and Fingers as transcendent stars; my method seems them as more modest players.  My system expects more out of Bando and Tenace than out of Blue and Holtzman, Bill’s calls it the other way around.  Otherwise, there is a good deal of agreement about where these guys fall on the scale.  When I look at the differences between the scores, I find my own scores consistent with my opinion of those A’s, which is that the pitching was overrated and the hitting underrated due to park effects.  I’d rather have a system that expects “minor star” seasons from Hunter and Fingers rather than “superstar” seasons, as Bill’s does, or just “solid contributor” seasons, as mine does, but I’m fairly satisfied with my scores.  If you prefer Bill’s, that’s okay—you could probably make good money betting on Bill James against me.  But if you will allow that my scores are reasonable enough, we can get on to looking at the leagues.

But first, some caveats.  The number of players in a league with substantial careers does not correlate perfectly with league strength.  First: this estimate is better for comparing contemporary or near-contemporary leagues than for longer-term comparisons, because it doesn’t really tell you anything about changing physical skills over generations or the like.  Second: It could be that some players are staying on in a league, logging win shares, because there isn’t enough pressure for turnover.  Third: It could be that the tendency for leagues to trade internally locks players into playing against inferior or superior competition for their whole careers, making their career win share total itself a polluted product of league imbalance.  Nevertheless, I am willing to bet that having players of substance is far more indicative of strength than of weakness.  Let us proceed then, to the burning question of the strength of field in the 1970s.

Well, on paper, the National League in 1975 was significantly stronger (average team 200) than the American League in 1970 (184), 1973 (188), or 1978 (178).  The 1975 Reds score as the best team in the four leagues.  Even if you remove these titans from the equation though, and just look at the strength of their scheduled opponents (adjusted for unbalanced schedule), they still played the toughest field (193).  The Dodgers, the Phillies, and the Pirates all had first-class frontline talent, and the Expos were the only real stinker in the league.

I always thought that the 1978 Yankees had a long-shot argument to be included in the greatest-ever debate because they had a lot of star players and they were in a really tough division.  Not only did the Red Sox look like an all-star team, but the Orioles, the Tigers, and the Brewers would give you fits too.  Well, it was a tough division, and if you take out the expansion+1 Blue Jays, the top six would rank even with the Red/Dodger-driven NL West in ’75.  But with a weak AL West and the New Jays on the docket, the Yankees arguably played the weakest schedule of the four great teams here.

As the standings make clear, there’s a lot more to a team than its theoretical frontline talent.  We’re rating leagues here, not teams.  Still, some surprises are worth noting.  The 1978 Rangers show up among the league’s elite.  I never think of Texas as having a lot of talent in the late 70s, but the system expects star seasons from Bobby Bonds, Al Oliver, Toby Harrah, and Mike Hargrove, and solid support from Jim Sundberg, Richie Zisk, Doyle Alexander, Jon Matlack, Bert Campaneris, and Doc Medich.  A lot of that didn’t work out, but they did win 87 games and finish 3^rd.  Matlack threw 270 innings with a 2.24 ERA, and they trailed only NY as the best pitching team in the AL.

The Dodgers had great talent in the 70s.  I recently watched the 1977 World Series in its entirety, and one impression I got was that they were much more like a modern team of wall-to-wall thick-forearmed sluggers, than the Yankees, who were more like a proper 70s outfit of speedsters and little guys surrounding the big boppers.  Anyway, the myth of the Reds is that they were destroyed by free agency.  Actually, they were simply dethroned by the Dodgers (and there’s never any acknowledgement that the Reds lost in 1977 despite acquiring the best player in New York City).  Of course the 70s Dodgers didn’t get it done in the World Series, and the Reds did.  Talent is one aspect of greatness; accomplishments are another.  The very greatest teams have to have both.  So I give up: the 1975 Reds were the greatest team of the 1970s.

You may ask why, if I am going to adjust the schedule strength score for the unbalanced schedule, I don’t just go all the way and give Cincinnati 242 points for each time they beat LA, 169 points for every time they beat San Francisco, and so on.  First off, I didn’t do that because I don’t want to project a false sense of precision onto these estimates.  More importantly, though, I would discourage the notion that a team has any sort of “true quality” essence that it applies to each of 162 games.  In reality, what a team brings to the field each day varies wildly depending on who is in the lineup and what kind of shape they’re in.  The assumption that teams have a static “true-quality” winning percentage has led sabermetrics down some dubious paths.

But that’s a subject for another day.  Let’s see how some leagues stack up from the parity decade:

Well, the parity is quite evident in the NL in ’86, less so in the 1989 AL.  The 1986 Mets were much less reliant on frontline talent than other great teams.  They had a very good fifth starter, they a second good relief ace, they platooned at two up-the-middle positions, they had a third baseman and a left fielder on the bench who were better players than the guys who were starting, and Rafael Santana was the starting shortstop.  Their top stars also partied themselves out of some better career totals.  They did rock at the time though, and they played a fairly tough schedule.  The 1984 Tigers, by the way, scored about the same as the Mets (211).  As Bill pointed out back in the 1986 Baseball Abstract, the ’84 team was highly dependent on an outstanding performance by its bench players.

I’m surprised by how the 1989 AL profiles.  I’m an admirer of the 1989 A’s.  For a couple of years, it just seemed like they were playing the game on a higher plane of intelligence than anyone else.  But I didn’t figure that they’d have the second-best frontline talent in the six division-era leagues in the sample.  And given that the AL played tight around this time, I thought the league’s distribution of talent would appear more even.  The system expects an MVP season from Rickey Henderson (39), an MVP-ballot season from Mark McGwire (31), star seasons from Dennis Eckersley (20), Jose Canseco (24), Tony Phillips (24), and Carney Lansford (22) and double-digit support from everyone else.  In real life, Canseco got hurt, Rickey’s MVP was a year away, and Phillips didn’t come on until he changed teams.  But in calling Rickey, the Bash Brothers, and the Eck big stars circa 1989, I think the system is doing what I asked it to do.  Outstanding frontline talent, deep bullpen, solid bench: the 1989-1990 A’s are an underrated great.  However, their opponents don’t come out looking too good here, and without a robust schedule as a mitigating factor, the A’s didn’t win enough to be considered the greatest team of all time.

This season, 1989, is about as far as we can go.  There were five guys in this league who got regular playing time and were still playing twenty years later: Gary Sheffield, Tom Gordon, and three Mariners: Omar Vizquel, Randy Johnson and Ken Griffey Jr.  The Mariners had some awesome talent in the 90s.  With Unit now deactivated, there are only four men left, which is not a big an issue for us.  Once you get into the 1990s, though, there are an increasing number of players whose active careers could change the scores.  Let’s go back to the 1960s:

I chose the two expansion years because I think the performance of each league’s expansion teams is instructive.  The Angels and Senators sort of fit right into the AL, while Mets and Astros got clobbered in the NL.  Sure enough, the talent on the Angels and Senators appears no stronger than that of the Mets and Astros, but they were entering a conspicuously weaker league.  Examining the leagues in this way definitely bears out Bill’s long-standing skepticism toward the 1961 Yankees—they played the weakest schedule of our 22 champions.  Of the 22 leagues I tested, the 1962 NL “felt” the strongest as I was entering the data.  It seemed like nearly every team in the league had the guts of a pennant-winner.  That makes sense, as the Pirates and Cardinals were just two years off a title, and the Dodgers and Reds just one year off a title.  The Braves still had their core superstars.  The Giants themselves were loaded enough to have won more than they did, and the Phillies were two years away from their near miss as well.  Without the two expansion clubs, the eight NL holdover teams averaged a score of 203.

But NL 1962 wasn’t the strongest league in the sample: NL 1967 was.  With a league strength of 216, it’s not even close.  I frankly don’t know what to make of that.  The expansion teams had improved, but it’s like the whole league is featuring quality players up-and-down.  By 1967, the major talent engine of the AL, the Yankees, had given up the ghost, and the NL’s advantage in competitiveness and racial integration were in full bloom.  Even still, I’m pretty surprised.  I chose 1967 because James claimed that the Cardinals were a truly underrated team.  This analysis does not contradict the claim.  Their average opponent scored a 213.  I haven’t up to now, but I think I have to start taking the 1967 Cardinals seriously as a great team.  Let’s go back to the 1950s and see if we can see where the AL loses it:

I chose 1953 because I think it was possibly the greatest World Series match of all time, with two long-time rivals each having arguably the peak season of their respective dynasties.  I still believe in Stengel’s Yankees, despite the weakness of their leagues.  They don’t do well in these sorts of analyses focusing on the starting lineup.  His rotation system and military service played hell with the Yankees’ career totals.  The 1950 Yankees scored at 231 and the 1951 Yankees scored at 229.  If Ford hadn’t been in the army, the 1951 team would score about 239, and if Newcombe hadn’t been in the service, the 1953 Dodgers would have scored around 245.  A full-on project to “restore” seasons lost to Selective Service might show that the league quality in seasons like 1953 weren’t as weak as they appear, but the major conclusions here would likely hold: the Yankees played in the lesser league, and as Bill suggested in “Teams on Paper,” the Dodgers didn’t have such explosive talent that you’d overlook their relative lack of accomplishments.

I chose 1958 because it is an interesting year from the league-quality standpoint.  In a rating based on standard deviations, AL 1958 looks great because all the teams clustered around the center.  What really happened though is that the Yankees surged out to a huge early lead and the AL went kaput.  Then NY stopped pedaling and drifted back toward the pack.

In the NL, Milwaukee was one of those super-talented teams that came away with less hardware then they might have.  I gave the Braves Joe Adcock (bigger contributor) rather than Frank Torre (more at bats) at first base, because I wanted to guard against shorting a group about whose great-team credentials I have been skeptical.  Although they did not win enough games or titles to be considered the greatest team ever, their frontline talent, with Adcock, rings in at 258.  The season before, it was 265, the highest score I found between the Depression-era Yankees and the 1970s.  The issue of league quality was part of the trash talk in the World Series.  After the 1957 Series, the champion Braves said the Yankees would finish in the second division in the NL.  After the 1958 Series, Casey Stengel speculated that perhaps the Yankees could play in the National League after all.

Well, the system regards the Yankees as the second-most talented team of 1958, behind the Braves, but the Braves surely faced a much tougher field.  The separation of the leagues is already apparent in 1953, and really dramatic by 1958.  In The Hardball Times, Steve Treder’s studies of talent production by franchise place the cross-over point at which the NL was responsible for more of the talent in the big leagues than the AL, in 1953.  To pull even in ’53, the NL must have already been out-producing the AL for several years.  The NL’s advantage is generally attributed to its being more progressive on signing (and playing) black players, but James has an interesting argument, which is that the aggressive integration was only a subset of a more competitive environment in the NL when Walter O’Malley hired Branch Rickey away from St. Louis.  Rickey had the chance to build two great clubs and the rest of the league emulated O’Malley’s determination to get better, kind of like the Yankees and Red Sox driving the AL to greater heights today.  Meanwhile in the NL, there was no O’Malley to challenge Yankee supremacy, and the league stagnated under their tyranny.  The research of Sean Smith, of baseballprojections.com, on players who changed leagues also supports James’s theory.  Smith discovers the NL advantage dates to the 1940s, when integration was still in its infancy.  Let’s back up before the war:

The 1936-1937 Yankees score as having more frontline talent that the vaunted 1939 club.  If you’ve ever seen one of those tournaments where they run the great teams through a simulated season, the 1936 Yankees will do better than the team that many think is the greatest of all time, because computer never knows what to do with the 39ers’  pitching depth.  The 1938 Yankees scored at 279 because they had the most perfect alignment of the bunch—Joe Gordon replaced an aging Tony Lazzeri, Tommy Henrich and George Selkirk got regular time in the outfield.  Eight good players on the field: that was the one year in the run when there was no Jake Powell or Babe Dahlgren to pull down the score.  If you were to give the ’37 team its two most valuable outfielders, Henrich and Selkirk, instead of the two with more playing time, Powell and Myril Hoag (Adcock ‘em, if you will), it would score a 280.

The NL in 1942 was still near full strength (sort of like the Korean War’s impact on the early 50s clubs).  The war cuts into the career totals of the players on the 1939 and 1942 teams, obviously for more ’42 players than’39 players.  Nevertheless, I think the lower scores of 1942 also reflect the AL’s superiority up to that point, and I don’t think the 1942 Cardinals were as talented as the near-contemporary Yankees.  If you were to fill in the Cardinals’ missing war years with win share estimates based on their established level of performance, they’d have some big gainers, like Enos Slaughter and Terry Moore, and the ’42 Cardinals would look better on paper.  But if you did the same for the 1939 Yankees, they’d have even more big gainers: Gordon, Keller, Henrich, and DiMaggio.

You can’t know what would have become of guys like Johnny Beazley, whose careers never got back on track after the military stint, but again, the other teams also had guys who couldn’t break back in in the postwar years.  No matter how many generous assumptions you make, I don’t think you can argue the 1942 Cardinals past the 1939 Yankees.  OK, now let’s see how those 30’s Yanks stack up against their AL predecessors:

In our 22-league sample, the team with the most frontline talent is the 1932 Yankees.  In James’s “Teams on Paper,” he determined that the 1931 Yankees had the most of all time.  Joe Sewell argued that his ’32 club was better than the ’27 club, because among other things, “we beat a better Athletics team.”  I don’t really buy it, but Sewell was right about the A’s, and I would add, they beat a better Senators team too.

We are at the point in baseball history where in terms of the roster, there really isn’t much more to a team than the frontline talent.  However, the frontline talent still doesn’t bear a 1:1 relationship to greatness.  The Yankees were the most talented team in the league between their 1928 and 1932 titles.  But Philadelphia beat them—not once, but three straight times.  Rob Neyer writes the A’s off as lucky, and James doesn’t seem too impressed with them either.  But it seems clear to me that while the A’s had less total talent from 1929-1931 than the Yankees, what they had was better balanced and better directed.  Connie Mack had his team playing smarter and harder while the Yankees were busy screwing up.  I guess, depending on how you feel about these things, to say the Yankees were more talented but Philadelphia just played better is either to explain why the A’s weren’t as great as they looked, or to explain why in fact they were.  To me, it’s more the latter.

The Yanks had to bury a Hall of Fame manager and they were sort of a mess for a few years.  When McCarthy got both feet planted, Philadelphia’s advantages in management and motivation evaporated and talent reigned supreme.  OK, two more:

Those famous Cubs beat the hell out of almost everyone they could reach, but they dominated a weak league.  The 1906 Cardinals and Braves are the two worst teams in the 22-league sample, and Brooklyn is not far ahead.

Intuitively, everyone knows the 1906 weren’t the greatest team of all time.  But it’s hard to kick them out of the argument, because 116 wins are a damn impressive concrete credential, especially when you have no number to hang on their quality of competition.  Well, the point of doing this analysis was to make more concrete this vague but no less important aspect of the debate.  The 1911 A’s played a schedule (181) that was, at least by one measure, about 10% stronger than the 1906 Cubs did (165).  In 1958, the NL’s frontline talent was about 13% stronger than the AL’s.  In the 1970s, the Swingin’ A’s were playing 180 teams, while the Big Red Machine was facing 200 squads.  You have to guard against being seduced by the illusion of precision when you hang a number on something, but it is an absolutely essential step to understanding the scale of the problem.