In his Baseball Abstracts, Bill James used to identify six “indicators” that would help predict whether a team would improve or decline in the following season. James would often talk about the indicators in his team comments; for example in the 1988 Abstract he mentioned that San Diego had all six indicators positive, while Minnesota and St. Louis (the defending league champs) had all six negative indicators. He also talked about the indicators in his Montreal Expos comment, identifying three significant indicators that pointed to a likely decline in 1988 for them. (As things turned out, the Twins did improve, but both the Cards and ‘Spos got much worse). In 1987, he used the indicators to note in the Pittsburgh Pirates comment that the Pirates had very positive indicators, and that they were (for that and other reasons) predicted to take a big step forward in 1987. Which they duly did, going from 64-98 to 80-82.
The indicators are interesting in themselves, and I think they are well worth taking a look at for 2006. This is in advance of a study that I am going to do on the indicators to see if they’re still as relevant as they once were; at any rate, baseball has changed in the last 20 years but I know it hasn’t changed that much. More free agent talent changes teams now than it did then (in the headiest days of the Collusion Era), but this is still an interesting jumping-off point for preseason predictions. The indicators can help identify the larger-scale statistical trends that we will sometimes miss if we are making a close examination of a team’s talent.
My friend Kent Williams, who is a baseball coach, writer and analyst, points out that the indicators themselves don’t actually tell us much. “I’m much more interested in personnel changes,” he says, “even if Florida and Kansas City were all positive they’d still finish last.” While that’s certainly true, the indicators can help us see things that we miss when we’re obsessing (as we all do) about the personnel changes. It’s easy to forget how unusually young the Braves are for an established team, that the Reds show a more promising trend than the Cubs, or that the Padres don’t really match the profile of a team with a promising near future.
Anyway, let’s get to the data. For each team, I scored each indicator as either “up” or “down”, depending on whether it pointed that a team would be better or worse. The six indicators are:
1. Pythagorean Record. Teams that outperform their Pythagorean record (their predicted won-lost record based on runs scored and allowed) tend to improve the following year. This was once known in sabermetric circles as the “Johnson Effect” after Bryan Johnson, who James credited with discovering the phenomenon. Applying the Johnson Effect in 2006 would identify the Diamondbacks and White Sox as most likely to decline, and the Mets and Blue Jays as most likely to improve (I won’t bother with the rest of the examples for now).
2. The “Plexiglass Principle”. Simply put, teams that improve in one season tend to decline in the following year, and vice versa.
3. The “Law of Competitive Balance”. Baseball teams tend to return towards a .500 record. Teams with a winning record tend to decline, teams with a losing record tend to improve. In these less straightforward times, we usually call this “regression to the mean”.
4. Age. Young teams get better; old teams decline. Because baseball-reference.com lists team ages this way, I split this category into two (age of position players and age of pitchers) and gave half a point for each.
5. AAA performance. Teams with good Triple-A teams tend to improve; teams with bad Triple-A teams tend to decline.
6. Late-season performance. Teams who play better in the second half than the first half, will tend to improve the following season. (And, of course, vice-versa).
Mike Green pointed out to me that combined Double-A and Triple-A performance would probably make a better indicator than Triple-A performance alone, since “many teams often promote the best players directly from Double-A.” That may be, however I’m leaving the indicator as Triple-A alone for two reasons. First, 90-95% of Double-A players are not actually ready to step into a major league lineup; not only do most of them go through Triple-A before reaching the majors, a large number never reach Triple-A. Second, the indicator should measure the ability of a team’s farm system to contribute in the following year; while eventually as many players make it out of Double-A, a much larger number of Triple-A players will see some big-league time in the following season than Double-A players.
I could also have graded each category so that (for example) a record of 82-80 wasn’t treated the same as a record of 100-62, but I’m a lazy, lazy man, so except where a team is exactly .500 (or plus or minus zero, or exactly on a median age line) there are no half-points or other fractions. If you’re looking at your team’s indicators (or its rivals’) don’t be content to stop there; look at how significant the indicator is.
Without further ado, let’s take a quick look at how the indicators stack up.
The team with the most positive indicators is the Kansas City Royals. The Royals underperformed their Pythagorean record by two games. They were the worst team in baseball. They had declined by two games from the previous season. Both hitters and pitchers were quite young (youngest pitchers and fourth youngest hitters). They also performed better after the All-Star break. The only non-positive indicator was their Triple-A affiliate in Omaha, who finished exactly at .500 to earn the Royals half a point. The Royals score 5.5 out of 6.
Two teams, Colorado and Pittsburgh (the NL’s basement franchises) scored five out of six. Pittsburgh declined in the second half, while Colorado’s Triple-A affiliate in Colorado Springs was below .500. Those were the only non-positive indicators for those two teams.
Three other teams, the Detroit Tigers, Cincinnati Reds and Oakland A’s, almost matched the Rockies and Pirates with five. The average age of Tigers and A’s hitters was 28.6 years, in each case exactly straddling the median line in the AL. They each had a quarter point (half of a half-point for hitters’ ages), and totalled 4.75 points each. The Tigers lost their point in late-season performance, the A’s lost theirs for being over .500. The Reds, on the other hand, lost a full point for Triple-A performance and a quarter point for being on the median age line for National League pitchers.
If you’re looking for a team likely to decline, the James Indicators say to look no further than the World Champions. The White Sox have all six negative indicators, and are the only MLB team to have 0 points under the system. Close on their heels are the Angels, Padres, and Cardinals, all of whom have only one point. St. Louis and San Diego get theirs for having declined from their 2004 records, while the Angels had a good AAA team in Salt Lake. Another team, the Washington Nationals, didn’t score a full point in any category, but had 1.25 points thanks to a .500 record, young hitters, and pitchers that were right on the median age line.
So there’s six teams picked to improve (Royals, Rockies, Pirates, Tigers, Reds, A’s) and five to decline (White Sox, Angels, Padres, Cardinals, Nationals). The system also picks the D-Backs as a bad bet and the Rangers and Brewers as improvement candidates.
One of the most interesting quirks of the indicators is that the indicators for the Yankees and Red Sox are exactly the same across the board, but to my surprise not negative. Each scores three points, exactly average. More interesting still is the fact that their divisional rival, Toronto, also has three points but each indicator is exactly the opposite of those of the Yanks and Red Sox; when they are up, Toronto is down.
A closer look at the tables (below) raises a couple of obvious but important points. First and foremost, the system tends to award points to bad teams. Not only is one point explicitly awarded to bad teams (thanks to the Law of Competitive Balance) but also bad teams tend to do well in the other categories as well. Bad teams tend to underperform their Pythagorean record, and tend to have declined from the previous year (that’s why they’re bad). They also tend to be young teams.
And indeed, bad teams do tend to get better. This isn’t merely the Law of Competitive Balance writ large; the fact is it’s much easier to improve a bad team than a good one. When the St. Louis Cardinals try to improve their infield, they have to try to find someone better than Mark Grudzielanek, which is hard to do. When the Kansas City Royals try to improve their infield, all they need is Mark Grudzielanek, who is pretty easy to get, and suddenly they have a huge improvement.
The second point to draw from the tables is that the American League, which has been stronger than the National League for several years now, not only widened the gap in talent during the offseason via free agency and trades (a topic I’ll be discussing in a future article and one worth bearing in mind for optimistic fans of AL teams or pessimistic fans of NL ones) but also is likely to be stronger according to the Bill James indicators. AL teams average 3.4 points, but the NL teams average only 3.1 points. 11 of the 16 NL teams had AAA teams under .500, while only four of 14AL teams did. On that basis, I am predicting that the gap between leagues will continue to widen in 2006.
The Bill James Indicators for 2006 (based on 2005 numbers)
American League Pythag Plexiglass Law CB PitchAge HitAge AAA Team 2nd Half TOTAL NYY DOWN UP DOWN DOWN DOWN UP UP 3 BOS DOWN UP DOWN DOWN DOWN UP UP 3 TOR UP DOWN UP UP UP DOWN DOWN 3 BAL EVEN UP UP UP DOWN DOWN DOWN 3 TBD DOWN UP UP UP UP DOWN UP 4 CHW DOWN DOWN DOWN DOWN DOWN DOWN DOWN 0 CLE UP DOWN DOWN DOWN UP UP UP 3.5 MIN UP UP DOWN UP UP UP DOWN 4 DET UP UP UP UP EVEN UP DOWN 4.75 KCR UP UP UP UP UP EVEN UP 5.5 LAA DOWN DOWN DOWN DOWN DOWN UP DOWN 1 OAK UP UP DOWN UP EVEN UP UP 4.75 TEX UP UP UP DOWN UP UP DOWN 4.5 SEA UP DOWN UP DOWN DOWN UP DOWN 3 National League Pythag Plexiglass Law CB PitchAge HitAge AAA Team 2nd Half TOTAL ATL UP UP DOWN UP UP DOWN DOWN 3 PHI UP DOWN DOWN DOWN DOWN DOWN UP 2 FLA DOWN EVEN DOWN DOWN DOWN UP EVEN 2 NYM UP DOWN DOWN DOWN UP UP UP 3.5 WAS DOWN DOWN EVEN EVEN UP DOWN DOWN 1.25 STL DOWN UP DOWN DOWN DOWN DOWN DOWN 1 HOU UP UP DOWN DOWN DOWN UP UP 4 MIL UP DOWN EVEN UP UP UP UP 4.5 CHC UP UP UP UP DOWN DOWN DOWN 3.5 CIN UP UP UP EVEN UP DOWN UP 4.75 PIT UP UP UP UP UP UP DOWN 5 SDP DOWN UP DOWN DOWN DOWN DOWN DOWN 1 ARI DOWN DOWN UP UP DOWN DOWN DOWN 1.5 SFG DOWN UP UP DOWN DOWN DOWN UP 3 LAD UP UP UP UP UP DOWN DOWN 4 COL UP UP UP UP UP DOWN UP 5