Jonathan Bernstein returns this week with the second part of his two-part series on Cactus League baseball. This week, he looks at the reliability of using spring training records and stat lines to predict success during the regular season.
In the old days, teams heading north following a poor Spring Training record could console themselves that losing in March was no ticket to the second division. Of course, teams leaving Florida with winning records probably thought they were destined for greatness.
Well, teams these days are as likely to head west as north, and the second division, after spending twenty five years on its deathbed, finally went the way of the Federal League and the legal spitball after 1993. But teams return from Florida and Arizona are still finding optimism in wins and ignoring losses. Journalists, as usual, are no help: either they repeat cliches, or they find one or two data points to support whatever they want to say. Who is right?
It's the losers. Spring Training results make great lining for birdcages, and you if you fold them just right they might float, but I wouldn't recommend taking them to Vegas.
Granted, this is from a brief study. I looked at the Padres and Giants Cactus League results over the period 1969 to 1993. It is possible that a larger study could show different results; it may even be true that attitudes about spring training vary systematically with different teams, so that it's worth paying attention to what the Yankees do but not the Padres and Giants. It's possible, but on the other hand the period under this study includes at least five ownership groups and over a dozen general managers and managers.
There is a small correlation between spring training wins and regular season wins, but that's not really the place to look: what fans want to know about their team on or about March 30 is whether improvement is in store for them this year. And, for that, spring training falls woefully short as an indicator. A variety of statistical tests failed to find any relationship between spring training results and regular season change.
I'm not going to give tables of regression coefficients; the failure of spring training as an indicator of regular season change is easy enough to see just from looking at the extreme points. The worst spring training turned in by either of these teams was the Pads' 6-16 in 1973. An indicator? The Brown-and-Yellow that year finished all of two games worse than in 1972 (yes, Enzo Hernandez was the leadoff hitter for that team, both years, thank you very much). The Giants are responsible for the next two entries on the worst spring list; San Francisco couldn't quite manage to play .400 ball in Arizona in 1982 or 1990. In the latter year, the Giants regressed seven games after winning the pennant: no surprises there. But in 1982, the Giants almost shocked everyone by staying in the race until the last week, lousy play in Arizona notwithstanding.
The same thing happens if you look at good spring trainings. The Padres destroyed (12-2) the Cactus League in 1990, the same year the Giants stunk. San Diego had finished in second, three games back of the Giants, in 1989. But it was the red-hot Reds who surged ahead and stayed there in April 1990, not Tony Gwynn, Roberto Alomar, and Benito Santiago.
The biggest season-to-season surge among these teams over the period was by the Barry Bonds Giants, up 31 wins in 1993 — after a 14-17 spring. The Padres did have a good spring in 1982, the year they began the push to the 1984 pennant, but didn't in 1984. The Giants ended fifteen years of incompetence in 1986, the first year of the Rosen-Craig regime and the rookie year for Robby Thompson and Will Clark, but you wouldn't know it from the spring record: 15-12, a little better than the spring preceding the 100 loss disaster of 1985 (Joel Youngblood at 2B, anyone?) and a little worse than the fine spring of 1984, which failed to signal the collapse of the formerly reasonable Haller-Robinson clubs that summer.
Frankly, I was surprised. Playing time in the spring isn't as different from playing time in the season as some people think. For example, the starting eight for the 1989 Giants (13-19 Cactus League, 92-70 National League) used 52% of the spring ABs, compared with 70% of the regular season ABs. Smaller, but still substantial. The difference in pitching is a little more dramatic: the top five Giants pitchers in innings pitched during the regular season completed only a third of spring IP, but almost 60% of regular season IP. If you had asked me going in, I would have said that the awesome performance turned in by the Fish this March was a positive indication for regular-season success. I guess I'd still rather see a team go 20-10 than 10-20, but you can't prove it by the real-world results of the Padres and Giants.
At least for these two teams, in these fifty seasons, there's not much there. If you really want to know what the White Sox are going to do this season, you're better off basing your bet on whether the Hawk reports an even or odd number of strokes over par than trusting the news from Sarasota.