The Siren Song of Correlation

Humans are really good at pattern recognition. It’s what makes us who we are as a species. In some ways, we’re too good with patterns. We see them even when they aren’t there. There is an important phrase in statistics that not all of us take to heart – correlation is not causation.

Today, I want to talk about potential false patterns. Put another way, when we pick a player to outperform his draft position, we should have a reason.

In January, I joined an industry dynasty league with fellow RotoGraphist Chad Young. The league is 20 teams with 45 players per roster. We keep 28 and have a 17 round slow draft which began last Saturday. As you can imagine, it’s ugly out there. We took Drew Pomeranz with out first pick and have since selected Andrew Susac, Luis Valbuena, and Justin Smoak. The latter pick is what has me thinking about this topic.

Once upon a time, Smoak was one of the most hyped prospects in baseball. I recall a younger version of myself bandying trade ideas involving Ryan Howard for Smoak and other parts (the younger version of myself was a Phillies fan).

Since debuting, he’s been quite the disappointment. A career .224/.309/.380 slash is pitiful for a first baseman. He can hit for some power, even at spacious Safeco Field, so there is hope for a 20 home run season in Toronto. In a fantasy league of this depth, the possibility of 20 home runs after the 600th player taken is worth investigating.

When we sat down to discuss the pick we used on Smoak, both Chad and I expressed hope that good things could happen this season. He’s a Blue Jay now and thus should benefit from some very hitter friendly parks. He hits slightly better as a lefty, which should be wonderful when visiting Yankee Stadium and Camden Yards.

Unless Danny Valencia and Dioner Navarro steal first base reps (and they could), Smoak appears set for plenty of playing time too. The Blue Jays also have a reputation for salvaging discarded hitters. Jose Bautista and Edwin Encarnacion are the most famous examples.

Here’s the problem, don’t these “pros” feel forced? As a FanGraphs analyst who makes his decisions based on statistical and/or visual evidence, there’s something odd or even hypocritical about this hope. So what if Bautista and Encarnacion broke out as Blue Jays. Both hitters had already demonstrates the raw ingredients for stardom elsewhere. They just needed time to put it all together. I smell correlation, not causation.

Smoak has shown us nothing but low line drive rates and moderate power. Watch him play, and he looks thoroughly mediocre. It’s not just a matter of latent talent, he needs to make a substantial change to his approach. We’re not talking about a monster season either. Said change is necessary for Smoak to remain a major league player.

In retrospect, our hope is starting to feel familiar. Before I ever visited FanGraphs in 2008, I was doing my own brand of fantasy analysis. A lot of it was built on “gut feelings” (i.e. complete BS) developed via pattern recognition. Here, we have a pattern: fringy looking hitters breakout in Toronto. What we’re ignoring is the reason.

I don’t have a concrete reason to expect more from Smoak. Were I managing the Jays, I’d have Valencia in to crush all lefties and Navarro handling the righties. That may be how they end up handling it. Smoak seemingly has a very light grip on any type of job. Did we get caught up wish casting when we could have taken a decent prospect instead?

My lesson from this experience is to have a tangible reason for each pick. There were alternatives in the draft pool. We’re hoping said alternatives drop to our next pick, so I can’t discuss them. I have reasons for liking those players. With Smoak, I think I might have goofed up and fallen for the siren song of correlation.

The good news for us? It’s not like we took Smoak over Michael Brantley. We didn’t ruin our franchise with this pick. Even though our process was questionable, there’s still potential for a good result.