July 5, 2006
Lies, Damned Lies
More on Elo
These four-day weekends always confuse me. I spent Saturday and Sunday vacationing in Wisconsin, and Monday enjoying fireworks and alcoholic beverages over the Chicago skyline. But yesterday, the actual holiday that created all this ruckus, I spent hunched over my desktop, crunching the numbers in order to determine the highest Elo ratings since 1960.
I decided not to stray at all from the method that I introduced in last week's article. There are arguments for introducing some sort of league-difficulty adjustment for the era before interleague play, and perhaps changing the bonus for margin of victory to coordinate it with the run-scoring environment of the league. But one of the nice things about Elo is its relative simplicity, and in the interest of both time and simplicity, I decided not to tinker with it.
One question that does need to be resolved is just what rating we take to be representative of a team's performance in a particular season. The most obvious alternative is to take a team's rating after the season has been completed, including all post-season play. However, you could also look at a team's average rating over each day of the season, their highest rating on any one given day of the season, or some combination thereof. Sticking with the keeping-things-simple theme, I decided to stick with the year-ending rating. One of the real advantages of Elo is that it is very well-equipped to consider both regular and post-season results, and this is the only method that adequately rewards or punishes a team for its playoff performance. This arguably introduces a slight bias toward teams that played better later in the year than early in the year, since Elo is sensitive to the recentness of results. However, I'm not sure that this is a bias that we should be taking great steps to avoid; many great teams were not fully formed until somewhere in the middle of the season, and a team's performance at the end of the season ultimately determines its place in history.
Since 1960, there have been thirty teams that finished the season with an Eloport record of 100-62 or better, which requires an Elo rating of about 1580. Just missing the list, with a 99-63 Eloport, were the 1977 Royals, 1985 Yankees, 1988 and 1990 A's, 1994 Expos, 2002 Giants, and 2004 and 2005 Cardinals. But this is not a list for also-rans. Without further ado, here are the thirty best baseball teams of the TV generation:
30. 2005 Chicago White Sox (99-63).
As we described last week, the White Sox gained 41 points of ranking during their post-season run, the highest of any team on this list.
29. 1968 Detroit Tigers (103-59).
The 1968 World Series is rightly remembered as one of the best of all time, with Mickey Lolich and the Tigers prevailing over the St. Louis Cardinals in seven games. However, that Cardinal squad was not especially strong, entering the series with a modest 1548 Elo. Still, the Tigers had plenty going for them, winning their league by 12 games.
28. 1979 Baltimore Orioles (102-57).
As you'll see, Elo has something of a fetish for the Earl Weaver Oriole clubs. The 1979 version is not one of the better remembered versions, but Elo is forgiving for their prevailing in a very tough American League East, and losing to an equally tough Pirates club in a seven-game World Series.
27. 1977 New York Yankees (100-62).
Billy's Boys are often overlooked between the dynasties that both preceded and followed them, but represented a credible mini-dynasty of their own.
26. 1969 New York Mets (100-62).
I thought that Elo might not like the Miracle Mets, as the 1968 version finished the season with a poor 1464 rating, some of which was carried over to the new season. But the Mets played better baseball as the season wore on, and picked up 25 points in the postseason with very impressive results against the Braves and Orioles.
25. 1980 Baltimore Orioles (100-62).
Another forgotten-about Earl Weaver team, whose season was undermined by a 19-22 performance before Memorial Day.
24. 1999 Atlanta Braves (103-59).
The '99 Braves were one of several Bobby Cox versions that might have finished in the Top 10 if not for their performance in the post-season. The Braves lost 15 points in their World Series sweep at the hands of the Yankees.
23. 1969 Baltimore Orioles (109-53).
It's actually a bit surprising that the '69 Orioles didn't rate higher, as their 110-52 Pythagorean record is one of the best in the modern era. One factor that harms them slightly is that the Orioles won ten games by nine runs or more, which Elo doesn't give them as much of a bonus for as Pythagorean-based methods.
22. 1984 Detroit Tigers (104-58).
It goes without saying that I was hoping for a slightly better result for my favorite baseball team of all time. The Tigers started out of the gate so strongly (35-5 record and 1590 Elo as of May 24th) that they coasted a bit as the season wore on. Also, while their post-season performance was impressive, it was accomplished against one of the weaker playoff fields in memory. The '84 Tigers might have been too good for their own good.
21. 1979 Pittsburgh Pirates (98-64).
Generally regarded as something of a one-hit wonder, the We Are Family Pirates are rewarded for defeating two very good playoff opponents in the Reds and Orioles, and for performing fantastically well down the stretch. Also of note is that the Pirates played an unbalanced schedule during a period in which the NL East was quite a bit stronger than the NL West.
20. 1983 Baltimore Orioles (98-64).
Another team that was helped greatly by its playoff performance. The '83 O's went 7-2 in the post-season, and outscored their opponents by 26 runs in the process.
19. 1986 New York Mets (108-54).
It has to be considered an upset that the '86 Mets didn't place in the Top 10, but they were playing against some fairly watered-down competition in the National League, and their post-season series against the Astros and Red Sox, while memorable, did not score them as many points as a couple of forgettable sweeps might have. The Mets' 103-59 Eloport matches their 103-59 Pythagorean record.
18. 1989 Oakland A's (99-63).
No disrespect meant to my friends in the Bay Area, but it is unfortunate that the one World Championship that the Bash Brothers managed to win was lost in the rubble of the Loma Prieta earthquake. The 1988 and 1990 Athletics clubs each finished with Elo's of 1577.
17. 1995 Atlanta Braves (90-54).
We'll get back to the Braves in a moment.
16. 1978 New York Yankees (100-63).
Accounting for their home field advantage, the '78 Red Sox rated as very slight 6-point Elo favorites over the Yankees in their one-game playoff to end the regular season. Bucky Dent didn't seem to care.
15. 1998 Atlanta Braves (106-56).
14. 1993 Atlanta Braves (104-58).
It's debatable whether the Bobby Cox Braves really deserve to be classified as post-season choke artists. Here are the Elo Ratings for each team in the dynasty, both before and after the playoffs ("End RS" is the score at the end of the Regular Season, "End PS" at the end of the postseason):
Team End RS Playoffs End PS Result ------------------------------------------------------- 1991 1547 +6 1553 Lost WS 1992 1566 -3 1563 Lost WS 1993 1604 -7 1598 Lost LCS 1994 1559 N/A 1559 Strike 1995 1567 +30 1597 Won WS 1996 1550 +14 1564 Lost WS 1997 1573 +3 1576 Lost LCS 1998 1600 -2 1598 Lost LCS 1999 1596 -7 1589 Lost WS 2000 1537 -14 1523 Lost LDS 2001 1526 +1 1527 Lost LCS 2002 1565 -3 1562 Lost LDS 2003 1577 -8 1569 Lost LDS 2004 1571 -13 1558 Lost LDS 2005 1531 -3 1528 Lost LDSOn balance, Cox's Braves have lost only 6 ranking points in the post-season, which suggests that they've played more or less as well as they might have been expected to. Of course, their results have been less impressive in the decade of the 2000s, and Elo gives them credit for playing (but losing) a fair amount of close series, something that only counts in bocce and ice dancing. The real tragedy may be that any of the 1993, 1998 and 1999 squads would have ranked somewhere in the top three on this list if they had managed to win the World Series, with an outside chance of finishing at #1.
13. 2002 Oakland A's (103-59).
The 2002 A's gained 42 points of Elo during their 19-game winning streak, jumping from 1554 to 1596, all while playing in which three of four clubs won at least 93 games. Like the Braves, the 2000-2003 A's may be more fondly remembered in twenty years' time than they are today. But with better results in the post-season, they might have been remembered as one of the best teams of all time.
12. 1971 Baltimore Orioles (101-57).
According to Elo, the 1971 Pirates pulled off a bigger upset in the World Series than Miracle Mets in 1969. The Orioles rated as 38-point Elo favorites over the Mets, but 28-point favorites over the Pirates. However, neither is especially close to the biggest World Series upset since 1960:
Year Underdog Elo Favorite Elo Gap --------------------------------------------------------------- 1990 Reds 1527 A's 1596 69 1985 Royals 1534 Cardinals 1580 46 1988 Dodgers 1547 A's 1592 45 1971 Pirates 1568 Orioles 1606 38 1974 A's 1550 Dodgers 1584 34 2000 Yankees 1527 Mets 1559 32 1995 Braves 1588 Indians 1619 31 1963 Dodgers 1561 Yankees 1590 29 1969 Mets 1575 Orioles 1603 28 2003 Marlins 1554 Yankees 1581 27 1987 Twins 1514 Cardinals 1540 26 1964 Cardinals 1553 Yankees 1578 25 1982 Cardinals 1546 Brewers 1566 20The 1990 Reds, of course, deserve extra credit for not only beating the A's, but sweeping them; the teams traded 24 rating points in those four games.
11. 2001 Seattle Mariners (116-46).
The Mariners' Elo peaked at 1622 on October 6th, the next-to-last day of the regular season. They entered the playoffs with 1619 points, which is also a record. Their post-season results are skewed somewhat by losing a 17-2 game to the Indians in the ALDS, a game which resulted in an 8.6-point Elo swing, the largest single-game shift in the database.
10. 2002 Anaheim Angels (99-63).
A very good club that played in a very tough division in a very tough league, and prevailed in the postseason over very tough opponents.
9. 2001 Oakland A's (102-60).
It's a bit of an odd result that the 2001 A's finished with a higher ranking than the 2001 Mariners, considering that the M's outperformed them in the regular season, and neither team did much in the playoffs. This may be one place where Elo's tendency to reward teams that improve in the second half may produce a misleading result. Nevertheless, Elo's goal is ultimately to be a predictive engine, and the A's proved to have quite a bit more staying power than the Mariners.
8. 2004 Boston Red Sox (98-64).
The more imbalanced the leagues and divisions become, the easier it is for a great team like the 2004 Red Sox to accumulate lots of rating points.
7. 1976 Cincinnati Reds (102-60).
The '76 Reds are justly rewarded for having one of the bigger cakewalk seasons of all-time, opening up a 9-game divisional lead by the 1st of August, and going 7-0 in the postseason.
6. 1995 Cleveland Indians (100-44).
The 1995 World Series isn't quite as well remembered as some other recent Classics, but it might have featured the most talent; the Braves entered the series with a 1589 rating, and the Indians, 1619.
5. 1999 New York Yankees (98-64).
Although Joe Torre's Yankees will rightly be remembered as one of the most impressive dynasties of all time, only the 1998 and 1999 editions of the Yanks were really memorable as individual units. The '99 squad gained 34 ranking points in an impressive post-season run.
4. 1975 Cincinnati Reds (108-54).
The 1975 version of the Big Red Machine is generally regarded as the best of the bunch, and Elo does not dissent. If the Red Sox had managed to leverage Carlton Fisk's home run and win the World Series, it would have rated as a 60-point upset; the Reds actually lost a point off their rating while winning the championship.
3. 1961 New York Yankees (109-53).
Another no-brainer entry. The '61 Yanks did everything that Elo likes, including improving as the season wore on.
2. 1970 Baltimore Orioles (108-54).
The 1970 Orioles belong in any discussion of the greatest teams of all time, particularly when one considers their post-season performance. Neither the Harmon Killebrew Twins that the Orioles defeated in the ALCS, nor the early version of the Big Red Machine that they defeated in the World Series, would qualify as pushovers. And yet, the Orioles went 7-1 in the playoffs, and outscored their opposition 60-30.
1. 1998 New York Yankees (114-48).
The '70 Orioles entered the post-season with a 1610 Elo Rating and picked up 23 points; the '98 Yankees entered the post-season with 1611 rating points and gained 24. Just a single Yankee loss to the Padres in the World Series would have shifted the tables and put the Orioles on top. That's as nitty as nitpicking gets, however, and it's no surprise to find the '98 Yanks atop the list.