Saturday, December 28, 2013

Did all the GM's get lost in Brad Pitt's eyes?

Don't get lost in his eyes - Brad Pitt has important things to say!

This past weekend I was enjoying an artisinal craft lager (1) with my current friend and former couchmate John Lynn when we began discussing my most recent blog post. Although John cares roughly as much for sports as I do for canker sores, he is so devoted to being an active listener that he feigns interest in my rants and even provides valuable constructive criticism. After hearing me complain for entirely too long (2) about the consistent follies of NBA decision makers and how amazing it is that they continue to give away wins by employing horribly unproductive players, John posed the following question: "Is this misguided choice to play unproductive players unique to basketball or pervasive across other sports?" 

As John wandered off to have his first cig (of the last 12 minutes) I pondered his question and decided I'd look at baseball. Baseball has long been at the forefront of the sports analytics movement, a revolution that has only gained steam since the explosion of Moneyball (3). Consequently, I presumed that this massive headstart in emphasizing data driven analysis would result in a crop of savvy MLB decision makers who are better able to evaluate production than their NBA peers and subsequently to avoid giving away wins by not playing unproductive players. Then again, one of MLB's mascots is a crack-addict so maybe I shouldn't have given the league too much credit. 

In order to test the extent to which MLB decision makers understand the value of not employing unproductive players, I turned to Fangraphs' Wins Above Replacement (WAR) statistic. WAR is a great metric because of its conceptual simplicity: it calculates the number of wins that a given player earned above a "replacement" level player (i.e. a player who could be found in the minor leagues or acquired for a choco taco). Specifically, I charted the aggregate negative WAR for the entire MLB from the years of 1984-2013 in two year increments. For fun, I looked at both all players who appeared during those years as well as strictly players who tallied enough appearances to qualify for batting/pitching titles. What I found was, well, sigh:

According to the data, MLB decision makers still give away tens of wins each year by allocating appearances to players who aren't simply below average, but are so unproductive that they don't produce at the level of a "AAAA" replacement player! Moreover, decision makers are giving away more wins now in the Sabrmetrics era than they did in the 80's and 90's!

When I saw this chart, I was convinced that something must be wrong. How could MLB decision makers, who must have seen Brad Pitt leverage easily available data to work wonders for the imaginary A's, still manage to employ such unproductive players? In order to dig a tad deeper, I counted the number of negative WAR players employed throughout this same time period(4). What I found was equally damning:

Yup, MLB decision makers are employing a greater number of sub-replacement level players now than they were in the 80's. 

So what's going on here? There are a number of explanations that likely contributed to this increase including an increase in injuries, changes to free agency/rising salaries, the steroid era increasing the performance discrepancy between the most talented juicers and least talented non-juicers as well as the expansion of the league - during their infancy, the Rockies and Marlins [added in 1993] and Diamondbacks and Rays [1998] all played some pretty horrible players (here's looking at you, Greg Dobbs!). That said, the increasing pool of talent coming from Latin America, Japan, etc... as well as the greatly increased understanding of what makes a productive player should have been more than enough to offset this change, right? 

All of which makes me wonder - did all the GM's get lost in Brad Pitt's eyes during Moneyball?

Let me know your thoughts, questions and critiques in the comments!


1. You're right - it was a Coors Lite. And that's only because they didn't serve Keystones...

2. Entirely too long = 2 Coors or 3 missed opportunities to speak to girls, depending on how you choose to measure time.

3.This trailblazing can be primarily attributed to logistical factors more than anything else:

-Baseball has been around forever (i.e. since the US of A was founded and time technically started)
-The game is (painfully) slow and full of deliberate actions, making it easy to track and quantify events
-Although there are clearly "team plays" that require teammate interaction (e.g. double-plays, sacrifice squeeze, etc...), baseball can be more easily evaluated as a series one-on-one battles between pitcher and batter as opposed to other sports (looking at you, football) in which it can be difficult to isolate an individual's contributions within a team performance.

4. In this case I strictly looked at players who had enough appearances to qualify for batting/pitching awards.

Monday, December 9, 2013

The Secret Ingredient to Consistently Winning in the NBA

To consistently win games in the NBA, team decision makers must follow a seemingly simple recipe: acquire and employ productive players. However, for a variety of reasons, this recipe can be hard to execute. For one, most NBA teams do not properly evaluate talent, leading to the misallocation of valuable, limited resources (draft picks, cap space, minutes, etc…). Additionally, many productive players never even make it to the open market since league rules not only stipulate that new players must enter the draft, but also incentivize players to re-sign with their current teams via contractual favoritism. Moreover, NBA history is pockmarked with examples of teams who tanked in order to acquire a sure thing prospect, only to receive a picks that is worse than what they’ve “earned” or to find that the “can’t-miss” prospect they drafted did indeed miss. Throw in the unpredictability of injuries, and it’s clear that acquiring and employing productive players is more difficult and less controllable than teams would like.

But what if there was another way to win more regular season games that, while not as impactful as employing stars, was far easier, cheaper and more controllable? Surely NBA decision makers, being the rational actors they are, would recognize this strategy and use it to their benefit…right? If you’re nodding “yes” right now, you’re probably reading the wrong blog.

So what is this oft-overlooked, easy, cheap and controllable tactic that could enable teams to easily add regular season wins to their totals? Simply put, it’s to avoid giving any meaningful minutes to really bad players, particularly to veterans who aren’t likely to improve.

To accomplish this goal, a team simply needs to reallocate all the minutes it’s currently giving to unproductive players. In theory, this reallocation may sound feasible but tricky, given that every other rational team should be competing for the same limited group of productive players. In reality, however, this trade would be remarkably simple, since NBA talent evaluators inaccurately measure productivity, leaving a ripe field of underplayed, attainable players just waiting to perform average or better.

Furthermore, teams don’t even need to uncover hidden gems to win more regular season games: simply reallocating their wasted minutes to average or even slightly below average players would lead to an increase in win totals! And yet, a quick scan of recent NBA history indicates that even in the information age, almost every team is guilty of employing players who should only see the floor in case of emergency.

As a quick and dirty way of capturing the extent to which each team has allowed unproductive players to sabotage regular season success, I looked at data from the 13 seasons spanning 1999-2013 and summed each team’s negative wins produced over that time. This data certainly doesn’t serve as a complete reflection of a team’s decision making competence and there is the occasional (rare) justification for playing unproductive players (e.g. rookies, trade bait, garbage time, etc…). However, this data serves as a decent proxy for revealing just how many wins are given away by decision makers whose job security is closely tied to their ability to win games:

For even more fun, I generated a scatter plot comparing each team’s negative wins produced and their overall win percentage. What you see is a strong correlation that, although is to be expected for a number of reasons, nevertheless accentuates how the most consistent winners (San Antonio, Dallas, Houston, etc…) not only acquire good players, but understand the value of not giving away wins via the employment of bad players:

To summarize, it’s important to recognize just how much a team loses by choosing to employ very bad players, particularly veterans who aren’t like to improve. These teams waste valuable, limited resources in the form of minutes and roster spots that at a bare minimum should be given to unproductive young players with potential. Moreover, since we know that fan attendance is tied to wins, teams that give away wins are also choosing to give away money.

Nevertheless, there are clear limits to the gains a team could achieve by redistributing minutes given to poor players. Realistically, teams need good players to win, particularly in the playoffs when wins can be largely attributed to your top 5 players. However, in a competitive league where playoff seeding (and home-court advantage) can come down to one or two games, it’s amazing to see NBA decision makers consistently give away valuable wins by choosing to play horrible players.