What is the Value of Our Advanced Data?
Whether you’re one of our Pro subscribers or a free user who follows us through social media and our content, one of the first things you’ll notice about FantasyLabs’ 2016 MLB initiative is an emphasis on advanced data. You will hear from us several times per day that Player X is hitting the ball Y feet farther than normal over the past 15 days or Player Z’s fastball velocity is down by three miles per hour. I’m sure that many of you have been wondering what these types of things actually mean in terms of fantasy production.
The truth is that even though I work for FantasyLabs and can experiment with the data using our free Trends tool, I have been wondering too. Our Plus/Minus stat (accessible via our free Ratings tool) is probably the best performance measurement tool in the DFS space, but what is the true value of our advanced data?
For example, we can create a simple trend that tells us that pitchers whose recent fastball velocities have dropped by at lease two MPHs have lost 1.83 fantasy points per start:
But is that a value we can apply uniformly across all types of pitchers? Or are there variables that we need to be aware of?
Although there are few things as satisfying as watching your $12,000 DFS ace mow down overmatched hitters with 98 MPH fastballs for seven or eight innings, there is more than one way to get opposing batters out. And there is certainly more than one way to try to get opposing batters out. While some pitchers rely on the fastball almost exclusively, others use it very sparingly.
Fastballs and Blanket Statements
You know that guy who won the National League Cy Young last year? Fastball has been Jake Arrieta’s third most frequently thrown pitch since the start of the 2015 season, behind his sinker and slider. In fact, you could make the argument that one of the reasons he became such a great pitcher after several years in the big leagues is that he stopped relying so heavily on his fastball and instead learned to mix it up. To mediocre results, Arrieta used the fastball as his go-to pitch in each of his first three seasons in the big leagues.
The point is that in DFS there are very few blanket statements that you can make. Some DFS players make the mistake of having tunnel-vision: They place too much importance on a single stat and fail to consider a player’s matchup as a whole. A pitcher’s velocity is merely a stat. It’s an important one, but it still needs to be placed in the proper context.
That’s what this study is about.
I’ve run dozens of trends and looked at countless FanGraphs Player Profile pages in an attempt to place velocity into some kind of context.
Please note that, because this type of advanced data is so new, the sample sizes on many of the following trends are rather small. Because of that, I think it would be a mistake to draw any hard conclusions here. Instead, consider all of this as a series of observations about what has happened so far and what might be true moving forward. We’ll need to monitor the results as more data becomes available.
This study will be presented in two parts. In Part I, we will break pitchers down into groups based on their overall average fastball velocities.
Sorting by Average Velocity
In the last section, I questioned whether we could apply velocity data uniformly across the board. In order to determine whether we can, we first need to break pitchers into groups. In the following sections, I group pitchers by total velocity, batted-ball profile, and pitch type. Most of the time, I use FanGraphs to help me identify which pitchers should be in which groups. I then create trends using the “Player Name” filter and adding the pitchers individually.
Low Velocity Pitchers
First, we’ll look at pitchers with the lowest average velocities since the start of the 2015 season. With some of the other subsets, I will look at the group as a whole, but the truth is that many low-velocity pitchers are just bad. Players like Jered Weaver and John Danks are very DFS-relevant . . . as stack targets, but this article is about the pitchers themselves.
In order to have a low-velocity cohort that isn’t needlessly weighed down by the dregs of the league, I’ve handpicked the pitchers in the subset. Each of these pitchers has an average fastball velocity under 91 MPH dating back to last season:
• Aaron Nola, Alex Wood, Bartolo Colon, Chris Young, Collin McHugh, Dallas Keuchel, Drew Smyly, Hector Santiago, Hisashi Iwakuma, Jaime Garcia, James Shields, Kyle Hendricks, Marco Estrada, Mike Fiers, Mike Leake, Ubaldo Jimenez, Yovani Gallardo
As you can see, this group’s baseline Plus/Minus before applying any filters comes out to +0.60 over 1,000 results.
Now, let’s see how they have done when their recent velocities are up by at least one MPH:
This is a scenario that just doesn’t happen too frequently. Of the 1,062 total games, only 24 matched the new criteria. It’s worth noting that the results have been very good, though rare. It’s really hard to draw any conclusions from such a small sample, but you can make a mental note that when a low-velocity pitcher experiences a sudden increase in velocity, that is probably a good thing.
When pitchers from this group have lost at least one MPH on their recent velocities, the results are negatively comparable:
Again, it’s hard to draw conclusions from the sample, but it may be worth noting that both of the worst performers in this category rely primarily on the fastball:
So losing velocity, even for pitchers who don’t rely solely on high-speed pitches, is probably a big red flag. And it’s probably even more important for pitchers whose primary pitch is the fastball.
Before I move on, I should mention that it is possible for a pitcher to be successful with a low-velocity fastball. Drew Smyly has recently been great with a fastball that averages only 90.7 MPH.
High Velocity Pitchers
I came into this study expecting high velocity hurlers to be the most affected by velocity swings, but the last group produced some pretty significant numbers and will be hard to top. Here is the list I compiled by sorting average fastball velocity on FanGraphs over the past season and a half:
As you would probably expect, the baseline Plus/Minus in this group is much higher than that of the previous cohort. Before the application of any additional filters, high-velocity pitchers outscore low-velocity pitchers by +0.7 Plus/Minus.
Now, here’s what happens when this type of pitcher loses at least one MPH in recent fastball velocity:
Again, we are dealing with a small sample. Interestingly, both groups saw a loss of around 2.7 points.
What’s surprising is that when this group has a recent velocity that is up at least one MPH, the Plus/Minus is actually 0.3 points BELOW the baseline of +1.38.
There are two pitchers from this group in particular who stick out. Michael Wacha (+1.8 vs. -5.4) and Stephen Strasburg (+11.6 vs. -8.7) have both posted positive Plus/Minus scores when their recent velocities are down and negative Plus/Minus scores when recent velocities are up. So what is unique about these two pitchers?
Wacha and Strasburg: A Consideration
For both pitchers, the “up” matches came last fall and the “down” matches came this spring. They are throwing softer in 2016 than they did in 2015, at least so far. But why have the results been so positive so far? As it turns out, each pitcher has changed his repertoire a bit.
In 2015, Strasburg threw only 26 sliders all year. Through the first month of 2016, he has already thrown that pitch 111 times. This fact suggests that we probably shouldn’t think about a player’s fastball in a vacuum.
Even if a pitcher has a tremendous fastball, Major League hitters are going to figure out how to hit it if it is thrown 90 percent of the time. By mixing in secondary pitches more often, Strasburg has been able to increase his Swinging Strike Percentage on his fastball, even though its velocity is down! Similarly, Wacha is mixing in his cutter much more frequently in 2016.
For both pitchers, it’s less about what was taken away from the fastball and more about what was added elsewhere. In these two pitchers, we have some evidence supporting the idea that, even if a pitcher declines physically, he can still be a competent MLB contributor if he leverages his experience to become more skillful.
There’s a Season for Everything
It’s notable that Wacha and Strasburg’s performances have a seasonal split: The end of 2015 versus the beginning of 2016. Over the offseason, both pitchers very likely made conscious and strategic decisions to change the way they pitch. A lot of thought and practice probably took place behind the scenes. In other words, the changes these pitchers made are the kind that can most likely occur ONLY during the offseason.
My assumption here is that a velocity drop starting in April is less devastating than one that happens mid-season. Of course, there are exceptions (notably Chris Archer this season).
I just crafted quite the narrative from an extremely small sample, so I would encourage you to setup your own trends and monitor performance as more results come in. But I do think there is probably something here.
By Month
Actually, we don’t have to take my word for it that an early-season velocity drop is less notable than one later in the season. We have our free Trends tool for that. (Make sure to note the Count column here before going overboard with the results.)
Pitchers who saw their Average Velocity drop last fall suffered disastrous results, as some of these Plus/Minus numbers are off the charts. Conversely, the pitchers who saw early-season declines didn’t suffer too much.
This dynamic will absolutely be something that I watch closely through the remainder of the 2016 season. If you’d like to follow along on your own, here is the trend I just set up:
Intermission
There’s more to the story, and in Part II we’re going to take a more under-the-hood approach, breaking up pitchers by batted-ball profile and pitch type. But even based on the more general approach that we took here in Part I, we can begin to see that it would be a mistake to weigh all velocity changes equally. How heavily a pitcher relies on his velocity and at what point during the season velocity change occurs are two variables that substantially affect Plus/Minus and subsequently DFS viability.
