The Angels and the Irrelevant Improbability

Every week, for the Effectively Wild podcast, we solicit listener emails and questions. The questions lean heavily toward the hypothetical and absurd, and an email we got last week asked what it would look like if some given eccentric baseball team owner cared only about winning the most games in spring training. What would the preparation be like? What would the strategy be like? How, in general, would the baseball team look?

You could argue that baseball team would look a lot like the Angels. Last Friday afternoon, against the Brewers, the Angels rallied back from an early 3-0 deficit. They tied the game in the fourth, and after they eventually fell behind again, they rallied again, before walking off in the bottom of the ninth. A player named David Fletcher singled against a player named Tyler Spurlin, and the single scored a player named Matt Williams, but not that Matt Williams. The Angels stretched their spring-training record to 7-0. They stretched their spring-training unbeaten streak to 18.

There’s an important specificity in there. It was an unbeaten streak, and not a winning streak, because last March 28, the Angels tied the Cubs, and last March 29, the Angels tied the Indians. Last spring, roughly 8% or so of all the games ended in ties. Still, a tie is not a loss. On March 22, 2016, the Angels beat the A’s, and on March 3, 2017, the Angels beat the Brewers. Over the duration of the streak in between those bookends, the spring-training Angels went 16-0-2. Such an achievement is mathematically amazing. Here’s the MLB.com headline after game 18:

Shoemaker returns with two innings; Street exits

Williams scored the winning run. He might as well have been looking for dried apricots in an unfamiliar grocery store.

Fletcher drove in the winning run. Teammates surrounded him as if to say, we support you, and are sorry you got stung by a wasp.

The Angels demonstrated a replacement-level emotional investment. It’s not difficult to explain why that is. Individual spring-training games don’t matter, not, at least, as far as the results are concerned. Links between spring-training games are scattered, and especially when you’re recalling games from the previous March. And, of course, in between the 2016 and 2017 spring trainings, there was the 2016 regular season, in which the Angels lost an awful lot. It’s iffy whether you can connect the two springs while ignoring all the other baseball. But spring baseball is different, yes? Specifically because it doesn’t matter. Players perform differently, and they’re used differently, in different numbers. The meaningful-baseball Angels, of late, have been bad. The exhibition-baseball Angels, on the other hand, looked downright unbeatable.

You cannot directly compare an unbeaten streak to a winning streak. Where an average team wins 50% of the time in the regular season, an average team *doesn’t lose* about 54% of the time in spring training. The rough odds of the spring-training Angels going undefeated for 18 games are 0.0016%. That’s basically the same as the odds of an average team putting together a 16-game regular-season winning streak. Since the turn of the millennium, the 2002 A’s own the longest regular-season winning streak, at 20 games. The next-longest streak is 15. We haven’t seen one of those in well more than a decade.

If you’re willing to separate spring ball from the regular kind, the Angels’ streak was something extraordinary. It was also, at the same time, something extraordinarily irrelevant. The Angels finally lost last Saturday. Here’s all the related coverage from MLB.com:

Walks, errors undo Angels in first Cactus loss
[…]
The Angels lost for the first time in Cactus League play after seven straight wins.

So much of our time is spent searching for the exceptional. We’re constantly on the hunt for fun facts and otherwise meaningful indicators, because we’ve become conditioned to embrace the unusual. But said embrace is dependent upon the presence of stakes; there are no stakes in spring training. The Angels did something you probably won’t see again for years, but then, no, you probably won’t see it again in your lifetime, because nobody’s paying attention. It can happen completely under the radar. I only even know about this streak because of a listener email follow-up. 18 games. 18 games! The A’s got a book and a movie. The Angels are getting this blog post.

As seemingly always, it comes down to this: Spring training, results-wise, barely means anything. Here are team spring-training winning percentages and regular-season winning percentages since 2006:

There’s not literally no relationship. This isn’t a picture of complete and total randomness, and the better a team actually is, the better it has tended to perform in March. But this is *basically* randomness, with just enough hint of a signal for various fans to justify crossing their fingers. As anecdotal evidence, just look at the Angels! Last year’s Angels were not good! We’re also coming off a year in which the regular-season results were less correlated with the spring-training results than ever (over the previous decade). Last year’s best Cactus League record belonged to the Diamondbacks. The Angels were in second. The White Sox went 17-13. The Mets were 8-17. The Twins were 19-11. The Cubs were 11-19. It’s nothing. Effectively, it’s nothing. We want so badly for it to be something, because spring-training numbers are the first numbers of the new year, but, overall, it’s nothing. Within that context, the Angels did something truly remarkable.

I don’t have a profound, overarching point to make. About once per 62,428 opportunities, an average team in spring training will pull off what the Angels did through to last Friday. Just in terms of rarity, and how the numbers work out, the Angels’ streak is one of the funnest of modern-baseball fun facts. And yet, it’s not, because nobody cares, because none of it matters. You might not have realized before that fun facts are conditional, but some degree of relevance is central to the fun. My apologies to the Angels, and to their statistical miracle.