The word follows some high-profile players around: Streaky. Most often, sabermetrically-inclined analysts poo-poo the idea that streakiness exists. You can’t use a player’s past week to predict their next day, they say. Yes, but. That’s not the whole story.
There are volatile players. These players do go into poor periods. Look at any player’s monthly splits and you’ll see a month or two that look like they belong in the minor leagues. Predicting when those periods happen is nigh impossible, but predicting that a player will continue to be volatile, well, that might not be so impossible.
Bill Petti at the sabremetric research site Beyond the Boxscore has done some excellent work on the subject which culminated in a statistic for volatility. By taking a player’s baseline offensive performance for the year, and comparing them to a rolling ten-day ‘moving average,’ you can see which players spike and trough more violently during the year. For a visual representation of what’s happening here, look at the picture below, which compares Evan Longoria, David Wright, and Scott Rolen this year. The width of the gray band shows how volatile each player is, and you can see how consistent each player has been, visually.
So, when David Wright is called streaky, part of what people are saying is that he is volatile. The peaks and valleys in his performance are more violent. Look, in particular, at his work against Longoria’s. Another way of saying this is: the standard deviations in his ten-day offensive value varies more wildly than Longoria’s. There isn’t a single valley in Longoria’s chart that delves as deep as Wright’s troughs.
How do we use this for daily fantasy baseball? Well, one easy way is to take away the role of K% in a player’s volatility. Petti points out that a player’s strikeout percentage is a prominent part of the formula, so you could try to generally avoid high-strikeout players in your daily selections and you’d benefit slightly.
But there are other factors, as well, including park factors and batting average on balls in play (BABIP). A quick look at the five least volatile players provides a little boost for this group: Joey Bautista, Joey Votto, Adrian Gonzalez, Dustin Pedroia, and Miguel Cabrera. But those players are already pretty good. And the five most volatile (Miguel Olivo, Alex Rios, Alex Gonzalez, Adam Dunn, and Mark Ellis) are already pretty bad. Check back at Beyond the Box Score this Friday for more on the subject from Petti, but it seems that a leaderboard isn’t extremely useful for daily fantasy players.
This might work better in the middle than it does on the margins. What you may want to do is throw a couple players’ stats into the calculator when you’re trying to decide between the two. Like, for example, when you are choosing between Curtis Granderson and Matt Holliday in the outfield. Despite Granderson’s advantages in park, his disadvantages elsewhere make him a much more volatile player (.826) than Holliday (.526). Holliday is therefore the safer player, on average, then Granderson.
One last note. You may want the more volatile player from time to time. In a head-to-head matchup, it makes sense to take the safer guy almost every time. But when joining the larger tournaments like the DFBC, you may want to choose the more volatile player. The thinking there might be, I want the higher peak possibility because I need all of my players to hit their high peaks in order to beat out the thousands of other managers in this tournament.
But, if you’re looking to win more of your daily head-to-head matchups, it looks like this volatility calculator can be another tool in your sabermetric toolbox.
Eno Sarris will have a little fun on Fridays while using sabermetric research to try to help you better your FanDuel choices. He also writes for all three blogs at FanGraphs and recommends them heartily. In his free time, he does his best at Bloomberg Sports, RotoWorld, RotoHardball and the best Mets site out there, AmazinAvenue. Thursday he reviewed an IPA for International IPA day at EnoSarris.com, so there’s that.