Archive for Steamer

Last year, I performed this very exercise, in which I compared FANS projections — projections generated by fans — to the Depth Charts projections — a composite of Steamer and ZiPS with playing time allocated by generally informed FanGraphs staff. I intended to highlight the largest discrepancies and offer a quick take on them.

I explain my interest in FANS during the inaugural of this exercise. Said interest pertains largely to anticipated versus most likely outcomes for a player and how those disparities manifest themselves in price distortions on draft day.

This time around, instead of discussing five National League outfielders at length, I’ll focus on the largest differences between FANS and Depth Charts projections in playing time, home runs, stolen bases, wOBA and WAR for a couple of players per category. I’ve set a personal goal for no more than three sentences per player so I don’t spend all day doing this. Because I could, and nobody wants that.

Read the rest of this entry »

A month ago, I compared Depth Chart projections — a composite of the reputed Steamer and ZiPS projection systems — to fan projections, compiled on each player’s page as “FANS.” I proceeded to look at the largest differences between the two systems — loosely considering the aggregation of independent fan projections a “system” — and identify which system better projected that particular player.

For hitters, the Depth Charts won by a landslide, 7-2. Turns out that fans frequently projected better production from prospects than what said prospects actually produced. It’s the cognitive bias I expected to see; I anticipated that fan(tasy owner)s associate hype with swift, robust production. I call it the Mike Trout Syndrome, named after the generational talent who has conditioned fans to expect immediate and immense production from all top prospects.

So how did the fans fare versus the Depth Charts regarding pitchers? Funny you ask. Here are the five starting pitchers whose Depth Charts FIPs varied most dramatically from their FANS FIPs:

Read the rest of this entry »

I have never thought much about, or given much attention to, FanGraphs’ fan projections. It’s primarily a matter of stubbornness: I run my own projections, and I develop my own strategy, so why should I listen to you? At least there’s a trace of rationale behind it: I know how I created my projections (whether or not they’re any good is a topic for another day), but I have no idea how you created yours. Thus, I am more likely to blindly trust a computer-generated projection system such as Steamer instead of random fan projections.

Still, there is a sort of bizarre, secondhand wisdom to fan projections. For every person who is high on a particular player, there could be another person who is equally-and-oppositely down on him. Solicit and aggregate enough fan projections and you could produce a very reasonable prediction of a player’s performance by sheer chance.

Which is why fan projections intrigue me. If, for example, Steamer predicts the most likely outcome from a wide range of possible outcomes for a player, then the fans convey the anticipated outcome for a player. The difference between them, you could say, is what amounts to a market inefficiency (aka a price distortion). The larger the difference, the greater the inefficiency. We see these inefficiencies arise every year — 2014’s most prevalent example is probably Corey Kluber — and they typically manifest because of a lack of information about certain players.

Read the rest of this entry »