Best Team Ever Formula / Article

[webmaster]

Is it really possible to create a formula that correctly identifies the best teams ever in Major League history? We here at Baseball Fever certainly wouldn't try and neither would our sister site Baseball Almanac. Research Roger Weber though has taken a shot and his new article is here:

Determining the Best Major League Team Ever Through Logical Statistical Comparison

Read through, download the Excel spreadsheet, and share your opinions with us here in this thread.

Sean

[RuthMayBond]

Similar to my list, maybe he puts more emphasis on winning the World Series

[RuthMayBond]

Is it really possible to create a formula that correctly identifies the best teams ever in Major League history? We here at Baseball Fever certainly wouldn't try and neither would our sister site Baseball Almanac. Research Roger Weber though has taken a shot and his new article is here:

Determining the Best Major League Team Ever Through Logical Statistical Comparison

Read through, download the Excel spreadsheet, and share your opinions with us here in this thread.

Sean

Can anyone tell me why the 1920 Philadelphia Athletics are in the same universe as this list?

[SABR Matt]

Randy (TKD) and I have actually just completed work on a major project that has a number of applications (more stable and accurate park factors, a fair dividsion of the responsibility for changing run scoring contexts between the league itself and the players within the league, the possibility of finding an objective measure for the strength of a league...and determine the greatest teams of all time objectively are among them).

I'll be posting our top fifty list when randy gets online tonight along with a bunch of information about this new team rating system (actually...we've discussed it once before here at BBF but it went rather unnoticed because we weren't quite at the point where we'd fully tested it for errors, and implemented it for the entire history of the game) called the Fiato-Souders Arithmetic Adjustments (FSAA).

I can tell you right now that we have the 1939 Yankees and the 1927 Ynakees as 1 and 2 all time among great teams...that the 2001 Mariners, 1998 Yankees and 1906 Cubs are bang-bang-bang close in overall rating but none of those three approaches the top two (they're close ordinally, but there is a substantial gap between the top two and the next down on the l ist, and that we didn't include the post-season in our analysis, although even if we did, we don't think the results would change too much...most of the winners are crammed to the top of our lists.

[538280]

Looking at the spreadsheet, it is quite clear this method is biased against modern teams. Too bad. It looks like he worked hard on it too. Why is it people always forget the game has become much tougher to dominate in the last 50 years?

[SABR Matt]

A good question...part of the problem is...his formula includes post-season winning percentage...for older teams...there was one post-season series...if they sweep in 4 games...is that the same greatness as say...the Yankees of '98 going 11-0?

[Ubiquitous]

second problem is that in a league of 8 teams playing in an era of poor infrastructure, resource management, and improper/inadequate governing body it is easier for one team to separate from the pack then in newer times.

Teams of the early 20th century were able to rack up the wins because team building and team maintenance overall was so horrible. And everything he is using to measure teams exacerbates the problem. Run differentials are going to be greater when teams can't build and maintain quality teams. Separation from the pack will be greater, and in imbalance in leagues is likely to occur which will favor the team who is built better then the best team in the sister league.

One small quibble is that for some reason he uses teams ERA instead of runs allowed. Not sure why he would do this. This also helps early 20th century teams out because it hides an enormous amount of runs allowed while not masking any of the runs scored. For instance the 1907 Cubs had 264 earned runs, while they actually allowed 390 runs to score. The teams hitters scored 3.7 runs per game. They had a 1.73 ERA, but actually allowed 2.52 runs to score. A huge difference, a difference that dissipates as the years move on and the defenses improve. In fact the more I think about it the more I realize that it isn't a small quibble but another major mistake.

[SABR Matt]

Agreed on both counts Ubi..wow...we agree on something completely...

How odd...

[Ubiquitous]

Not really, I think there is probably a lot we agree on. Agreements don't get talked about, disagreements do.

[Ubiquitous]

I'm actually surprised that they posted that on Baseball Almanac. I remember noticing this thread when Webmaster first posted it and thinking perhaps it was an old article they had found and decided to post on the ALmanac. Which they have done before, but to my surprise its basically just another stat/history thread. One mans opinion backed up by his pet formulas. No real difference between that and say my WARP rankings or I think it was Applings HAA, or SABR Matts PCA.

[SHOELESSJOE3]

Agreed on both counts Ubi..wow...we agree on something completely...

How odd...

If one thinks it over many disagreements on this and any board are not as far apart as they seem. Many are not total disagreements only disagreements in degree.

What I mean, an example a Hornsby debate, the home park how much of a factor was this in his career stats. Some believe it meant more than others, some think not as much. There are a good number of debates, disagreements on players, teams, even those who disagree do agree that a factor should be considered, they disagree on how much, of what value was that factor in coming to a conclusion.

At times when a debate gets to go on too long and I know nothing will ever be agreed on 100 percent I usually end it by saying pehaps the answer is in the middle of what I believe and those I disagee with. I'm not 100 percent right and neither are those that I disagree with, I have to yield some and hope they can do the same.

[webmaster]

[QUOTE=Ubiquitous]I'm actually surprised that they posted that on Baseball Almanac. I remember noticing this thread when Webmaster first posted it and thinking perhaps it was an old article they had found and decided to post on the ALmanac.QUOTE]

It was from last year, but I never did post it due to it's size. I asked him to update it for 2005 and went the Excel download angle which has worked well. Since being published yesterday the sheet has already been downloaded 25 times.

As for it going up on Baseball Almanac, I always enjoy adding research. I NEVER wanted to be just another stats site. I still don't. I never want to be a news site - MLB.com has that covered. I wanted to be a site for everything else that we all love which makes the game great. Lists, feats, records, poetry, pictures, baseball cards, history, stories, articles, anecdotes, and so on. I just got sidetracked a bit with stats (I think 145,954 pages counts for being sidetracked) for a couple years and am now getting back to some of the things I love...

Sean

[Ubiquitous]

By the way the #10 greatest team of all time the 1920 A's lost over 100 games. Its a typo it should be the 1929 A's.

Also there is a link and a discussion about this over on Baseball think Factory.

[redbuck]

I'm Roger Weber, creator of the study.

Obviously any study that includes postseason winning percentage gives an advantage to older teams that played fewer postseason games. But it really isn't fair to call a team that finished 111-43 but lost the World Series in 4 games equal to a team that finished 107-47 and won the World Series in 4 straight. Winning a World Series is critical to a team's greatness. Even though there are factors like weather, injuries, etc. that can compromise a great team in the postseason, it seems like the whole point of playing the season is to win a championship.

As for the 1920 Athletics, that is a typo. It's stupposed to be the 1929 Athletics.

There are lots of other factors that could be considered, and the weights I use are obviously debatable. I tried to give reasons for them but obviously something like that will stir up debate, which was in part my intention.

Through all of this, I have pretty much concluded that there is no real perfect way to measure teams from different eras against each other. The league expanded and the postseason was lengthened. That's why it seems so logical to exclude pre-1920 teams.

Maybe a game of strat-o would do it!

[Ubiquitous]

Also why would one exclude 1942? Virtually nobody has left for the WWII and all the regulars are there.

[redbuck]

35 Hall of Fame players and 500 total players left for the war. That doesn't sound like nobody. Warren Spahn, Ted Williams...

[Ubiquitous]

Yes they did but not leave in 1942. Ted Williams, Joe Dimaggio, Stan Musial, Phil Rizzutto, Red Ruffing, Bobby Doerr, Jimmie Foxx, Bob Lemon, Luke Appling, Ted Lyons, Early Wynn, Bill Dickey, Enos Slaughter, Pee Wee Reese, Billy Herman, Johnny Mize, Warren Spahn, and Lloyd Waner all served in WWII were all future HOF'ers and they all played in 1942. 19 in all. On top of that there is at least another 10 future HOF'ers who played that year that never served in the war, plus one quite and went farming.

In 1942 only 71 players left for the war.

Here is a list of the NL players gone from the 1942 rosters, (thanks to steve treder)

Braves: Bill Posedel, John Dudra, Bama Rowell.

Dodgers: Herman Franks, Don Padgett, Cookie Lavagetto, Joe Gallagher, Tommy Tatum.

Cubs: Walt Lanfranconi, Russ Meers, Eddie Waitkus.

Reds: Benny Zientara.

Giants: Jack Aragon, John Davis, Morrie Arnovich.

Phillies: Roy Bruner, Lee Grissom, Bill Harman, Dale Jones, Hugh Mulcahy, Dick Mulligan, Emmett Mueller, Jim Carlin, Joe Marty.

Pirates: Bill Clemensen, Oad Swigert, Vinnie Smith, Billy Cox, Burgess Whitehead.

The Cardinals themselves were missing Johnny Grodzicki and Walter Sessi.

The NL lost 31 players the AL lost 40 that year. Now can you seriously state that the NL of 1942 was a weakened league? The 31 guys they lost amount to nothing really.

[SABR Matt]

Agreed there Ubi...I can't find any serious study of league strength that concludes that 42 was a weak league (either league)...the drop in quality starts in 43.

[redbuck]

It is true that most of those players were still playing in 1942, and your point is well taken.

Still, many of those players did enlist during the year and toward the end of the season many left for the war.

Of articles I've read, the feeling during 1942 was more of patriotism than a competitive spirit to win. New regulations about night games, all star games were put in to prepare for wartime baseball.

The league may not have been weakened in the names of the players that played, but still because so many players did leave or were preparing to leave, grading that season didn't seem logical.

The Cardinals, 106-48, though, are arguably one of the great teams but probably were not as good as the 1931 Athletics.

[RuthMayBond]

Randy (TKD) and I have actually just completed work on a major project that has a number of applications (more stable and accurate park factors, a fair dividsion of the responsibility for changing run scoring contexts between the league itself and the players within the league, the possibility of finding an objective measure for the strength of a league...and determine the greatest teams of all time objectively are among them).

I'll be posting our top fifty list when randy gets online tonight along with a bunch of information about this new team rating system

Did I miss something?

[BillyF29]

Like many have said, in 1942 baseball was not really as competitive as in the previous years leading up to it, but the big drop off's were in 1943 and even larger in 1944 with 1945 being about on par with 1942. As expected with the color barrier being broken, a huge leap in competative balance was made in 1947.

[Ubiquitous]

It is true that most of those players were still playing in 1942, and your point is well taken.

Still, many of those players did enlist during the year and toward the end of the season many left for the war.

Of articles I've read, the feeling during 1942 was more of patriotism than a competitive spirit to win. New regulations about night games, all star games were put in to prepare for wartime baseball.

The league may not have been weakened in the names of the players that played, but still because so many players did leave or were preparing to leave, grading that season didn't seem logical.

The Cardinals, 106-48, though, are arguably one of the great teams but probably were not as good as the 1931 Athletics.

So how come 1918 or 1919 gets rated? How come Federal League years get rated? What about the years in which the Mexico league were trying to make a push? How about when they were drafting players for the Korean war? Or minor leaguers during the Vietnam war?

The FL years took many many more players away from the majors then 1942 yet you have 1915 ranked 21st all time, and 1914 ranked 32nd.

[redbuck]

Very good points Ubi

Unfortunately, it isn't really even fair to compare pre-1947 or pre-1961 teams with present day teams. I don't remember if the write-up I did for this study had lists of only teams from after those eras, but I did make those lists. Of course, the 27 Yanks, 39 Yanks and others are left out then, but there was a huge change when blacks started to play, when the leagues expanded and even more recently with issues like ballparks and steroids affecting the game so much.

I see your points and I guess you can take out whatever teams you want. It's impossible, though, to accurately measure how much different circumstances affected the game and it definitely is impossible to effectively measure teams against teams 100 years older.

[SABR Matt]

I don't believe it's "impossible" to measure the strength of a league and b so doing measure the differences between great teams of all eras. Obviously there will always be some amount of uncertainty but I think we can get a very good idea.

I believe the way to measure the strength of a league as accurately as possible is to do something very much like what is done when we want to see how strong AAA or the Nippon League is. For those not familiar with this process, the way they calculate MLEs (Major League Equivalencies) for Japan League imports and Major Leaguers is by looking at players who crossed contexts (players who played in both AAA and the majors) and seeing how their performance changed when they did this. It's observed that when sluggers from Japan come to America they lose about half of their home runs. This estimate will get better as more Japan League players cross the barrier into the majors.

I believe once I have player ratings I'm happy with, the way to rate strength of league will be to examine win scoring rates and see how they change from season to season for each player during his active lifespan. If we do this for just one player, you have the age problem obscuring things, but the neat thing about the major leagues is that there are always players coming and going...league turnover is on the order of 8% every year with some years featuring a lot more turnover. If we combine all of these rating changes together for every season-to-season pair and look for seasons where there's a LOT of rating mobility relative to normal, those will be seasons we can consider "weak". We can use rating mobility as a pretty good estimate of the strength of each league except the first one (1871 NA) because there's no "before".

[SABR Matt]

Did I miss something?

Sorry...we found a couple of minor problems involving forfeitted games that needed to be fixed and I'm now waiting for the corrected data...the ratings won't change much but better safe than sorry.