A Brief Glance at Variance

Much has been made over the past couple years as teams seem to, more than ever, embrace one of two attitudes leading into any particular season.  The first attitude is of those teams who clearly aim to compete.  They may have “won the offseason”; they might employ an incumbent core of players capable of making another postseason push; they just might be bearing the fruits of a rebuild and plan to compete accordingly.  The second attitude is that of the non-contender.  Here, teams delved not into free agency’s upper echelons, might have only added to their 25-man roster by way of subtraction, and in general turned their sights toward to future predominantly.  

This perceived trend has far reaching implications.  It affects players in free agency.  It makes for case studies in rookie service time cases. It influences the makeup of whole divisions.  The list goes on.

How many of the Red Sox 108 wins last year came against the 47-win Orioles, for instance? Is it good for baseball that three teams averaged over 103 wins last season and three other teams simultaneously averaged under 56? Has the fork in the road to success become especially distinct given data-driven rebuilding and delicately calibrated windows of contention?

Given last weeks data, I figured that it might be an interesting task to check whether, in a pointedly un-rigorous fashion, this dichotomy in teams’ approaches has led to greater disparity in the win column as of late.

As such, the following graphics will address two key questions.  First, have team records become increasingly spread out in the past 19 years? Second, more in the vein of triviality, which teams have performed most and least consistently (for better or worse) since 2000?

To quantify what has been referred to as “spread out” in question one and “consistently” in question two, standard deviation and variance are called upon.  

Standard deviation, which captures the average dispersion of a set of numbers means, is about as fundamental a measure as there is in statistics, most generally in populations of data that follow a normal distribution.  Variance, the second moment of a distribution and the squared value of standard deviation, plays an additionally central role in statistics.   Common uses of variance include hypothesis testing, quantifying the efficiency of a sampling distribution, and probability testing; in fact, an entire group of statistical models (ANOVA) are named for the analysis of variance. 

Here though, we pretend that 2000-2018 is our entire population so extrapolating using various measures even won’t be necessary.  Instead they will be used simply as benchmarks to be taken at face value.

Visualization!

2018 Red Sox and Orioles be damned, dispersion of win totals is trending downward. 2002 and 2018 are clear outliers here, with 2002 actually representing the widest gap among win totals league-wide this century. Still, it appears as though rumors of parity’s demise in baseball have been exaggerated. Is this an airtight conclusion? Absolutely not. Is it of interest? I, personally, think so.

Of course, with a population of just 19 data points, all baseball needs is another year like 2018 to tilt this trend upward. The Rangers matched their 21st century low in wins last year; they don’t figure to be too much better. The Orioles and Royals do not appear primed to turn the corner, so to speak. On the other hand, Boston, New York, and Los Angeles don’t appear to be approaching a cliff anytime soon. Repeating a 2018 type season is absolutely conceivable.

Visualization, again!

There does not appear to be any eye-catching correlation between success of teams and consistency (given by variance), although the Dodgers, Yankees, and Cardinals appearing in the bottom four is of note.  After all, if you’re going to be consistently anything, why not be consistently good?

“Tank-it til you make it” poster children Cubs and Astros additionally occupy top-5 spots in variance on the other end, highlighting their extended valleys that, only eventually, have given rise to obvious and considerable success.

These results weren’t exactly expected, but in cases like this, that just means an outcome that is all the more interesting. In any event, an exercise worth doing during a slow offseason.