Diamondbacks Steamer Projections – 2014 Primer

Over the next couple of weeks, Inside the ‘Zona will be previewing the upcoming season through Steamer projections. Each post will feature a different Diamondbacks unit (ie. bullpen, outfield, etc.) and a player-by-player breakdown within those units. Provided will be each player’s Steamer projection for 2013, his actual performance for 2013, the player’s Steamer projection for 2014 and a capsule noting expected performance. In case you can’t tell, we’re ready for the 2014 season to begin, so let’s start digging in!

If you’re not familiar with Steamer, take a look at the information and links below. I’ll tag this post at the top of each unit for easy access should you need a refresher.

What is Steamer?

Steamer is a projection system devised by Jared Cross, Dash Davidson and Peter Rosenbloom. It seeks to project every player’s performance for the upcoming season by using past performances and different aging curves, plus information such as minor league performance, age, park factors and more. This should reveal a player’s true talent level and take the ‘luck’ out of things. Essentially, it’s all very complicated but widely accepted as one of the best projection systems available. If you’d like to learn more about it, I strongly suggest visiting the Steamer Blog website and taking a look at Dash’s posted PowerPoint on methodology.

What are Steamer’s Limitations?

Like any projection system, Steamer has clear limitations. It cannot predict things like injuries (although it does take past injury history into account) or the way a certain manager decides to utilize playing time (such as platoons or moving a former reliever into the starting rotation). Steamer also has a very difficult time with projecting minor leaguers as it cannot forecast how much playing time the player will get if/when they do get called up. Projections are also more reliable when there is a larger set of past-performance data to pull from and analyze. Because of this, Steamer (and most projection systems) tends to be far more accurate when projecting the performance of veterans than first or second-year players. In other words, it’s easier to project Martin Prado in 2014 than it is to project AJ Pollock.

Improvements vs. Drop-offs

You’ll notice that Steamer projects significant improvements for many Diamondbacks players. I’d argue that it’s because Steamer regresses towards the mean and as you saw in 2013, several Diamondbacks had uncharacteristically poor years. Steamer projects that most of these players move closer to their career averages. But just like down years are projected to be followed up by more average seasons, the same goes for break-outs. Steamer, just as it regressed down years to the mean, does the same for players who had uncharacteristically good years in 2013. Therefore, you’ll see some players are predicted to have a lower performance next year. This isn’t because Steamer hates your favorite player, it’s because Steamer plays the averages based on past performance, age, injury history, park factors, etc. Just as we rejoice over a predicted uptick for a player who disappointed, we must also respect a decrease from a player who came out of nowhere to post career highs.

This projection system is nuanced in ways far more complicated than I’ve just laid out, but hopefully you get the idea. Check back starting next Wednesday to see the first set of projections for the 2014 Diamondbacks. They say the game of baseball starts with pitching and we’ll do the same here, unveiling projections for the rotation to kick things off. See you next week!