Deep, But Playable: Blind to the Present

For several years now, I have attended a screening of the Oscar-nominated Animated Shorts. This year, the offerings were quite good on the whole―headlined by Pixar’s award-winning Piper―but there was one particularly disappointing nominee, Blind Vaysha. The premise is this: the main character is a girl who sees the past out one eye and the future out the other, but is blind to the present. The film is not subtle, and the ending narration smacks you over the head with the theme while drawing all the conclusions for you.

Heavy-handed film-making or not, Vaysha did get me thinking―just not in the manner the creators likely intended. Rather, I pondered this question: how accurately could we surmise a baseball team’s season by knowing only of the previous and subsequent seasons that sandwiched it?

I asked resident benevolent wizard Rob McQuown to generate a simple model to test whether having only the team record in the year prior (Year 1) and year following (Year 3) a given year, would allow you to approximate that team’s record in that given year (Year 2). The results are plotted on the chart below and the correlation was a very nice .69.

It stands to reason that knowing the future gives some insight into the present. Knowing the events of the season prior and the following season allow us to gauge where a team might fall in the meantime. The 2013 Royals were solid and narrowly missed a playoff spot, and the 2015 Royals won the World Series. It is then not surprising that one might hazard that the 2014 Royals were likely good, probably made the playoffs, and perhaps even had a deep run. Baseball teams tend to move slowly toward their destinations, especially uphill, and usually take some time turning things around once they’ve hit bottom.

But what about the ones that don’t?

The aforementioned Mr. McQuown helped me find a bunch of seasons in which the “present” season—the one that we’re looking at despite knowing the win-loss records of the year prior and the following year—deviates substantially from the win-loss records of the years around it. In fact, the top 10 seasons with the largest delta are as follows:

[Extremely Dr. Frasier Crane voice] Hello Seattle!

They might not have the top spot, but locking down the next three is impressive (and erratic) as hell. It’s worth noting that the 2001 Mariners and their 116 wins are sandwiched by a 91-win season and a 93-win season, so while they fit our parameters, they’re still something of an outlier among outliers. So what exactly happened to these other teams? I investigated four randomly selected seasons below.

So what does all of this teach us? Not much, exactly. But, small sample size warning, seven out of the top 10 outlier seasons above featured teams vastly outperforming their surrounding seasons. Perhaps it’s easier to be strangely good than it is strangely bad. Probably too simplistic an observation to make given that we’re dealing with 10 outlier seasons, and there are complicating factors (a weirdly bad season could result in a sell off, thus subsequent bad seasons which wouldn’t get caught by our search terms). Still, something to ruminate on or investigate another day.