Here at FantasyLabs, we have a dizzying array of data at your fingertips. If you so choose, you can simply load one of our Pro Models and build your lineups from there. That’s a solid strategy and one that can be used to turn a profit (coupled with solid lineup construction and game selection, of course).
However, if that’s how you want to play, this article isn’t for you. We’re going to take a look at some potential trends in our MLB Trends tool to consider for MLB DFS and how we can use them to build our own lineups (or even our own models, if you so choose) to gain an edge on our competition.
We covered hitters in last week’s article; now it’s time for the pitchers. While some of the concepts are the same – namely finding metrics that are undervalued by either DFS salaries ownership — there are some differences.
Mainly, the sample size for pitchers is much larger. A typical starter can throw in the neighborhood of 100 pitches a start. That makes them easier to project – and edges harder to find. On any given day, a pitcher’s scoring is going to be a more accurate reflection of their ability and matchup than a hitter’s.
Still, I’ve discovered some trends that seem to have some predictive value and are underutilized by our competition.
Let’s take a look at a few.
Park Factor and Weather Rating
Just like we did with hitters, we can utilize FantasyLabs’ proprietary Park Factor and Weather Rating metrics to aid us with pitcher selection. As with hitters, these scores are back-tested over a large sample size to isolate the best conditions for pitchers – you don’t need to figure out the relative value of temperature compared to wind speed, for example.
After toying around with various combinations of Park Factor and Weather Rating (over the last three seasons), this is the best trend I could find:
Of course, any hunting like this will eventually find something that at least appears predictive. I tried to keep the ranges of scores fairly large so as not to “cheat” with small sample sizes. With this trend, we’re “paying” just under 1% in additional ownership (the baseline is 10.9%). However, we get 0.9 points and an extra percent of upside. (Not pictured in the screenshot, but pitchers in this sample post an “upside” score 13% of the time.)
This is the first example of an important lesson. Given how predictable pitchers are, it’s hard to find truly overlooked players with an above-average shot at a big game. Where we can benefit though, is by targeting players in the second-best tier of various metrics. (This is often the case with the NFL as well. Stacking the second or third highest game by Vegas total is +EV compared to stacking the highest.)
We need pitchers with the combination of talent and situation that they can have a good game. It’s unlikely that pitchers in bad situations – say a hot day at Coors – can truly have a great performance. Players in reasonably strong situations though tend to be under-owned compared to their chances of a slate-winning score.
We’ll revisit this concept later in the article. For now, though, the key takeaway is to target players in above-average – but not ideal – situations.
This was the big surprise of my research into hitters, so it made sense to go back to the well. This time, we’ll look at umpires who produced a Plus/Minus Score of at least 1.5 for pitchers over the past three MLB seasons.
These 11 Umpires (I removed Derryl Cousins, who only worked home plate twice in three seasons) have worked a combined 980 games over three years. That means at least one of them should be behind home plate most days. (Which, of course, gives you two pitchers to choose from.)
Just like with hitters, this was a very strong trend. Pitchers in games umpired by this group actually came in less owned than the field, by 0.4%. Despite that, we’re getting nearly two points of Plus/Minus and 2% better chances of an upside score.
For MLB DFS players, having this data on Umpires is worth (more than) the cost of your FantasyLabs Membership.
Maintaining data on every umpire – while filtering out DFS salary – would be a monumental task. The Plus/Minus “allowed” for each game’s umpire is displayed in our MLB Player Models each day.
Rather than look at raw strikeout rates, I examined how a player ranks relative to their peers on that day, using our Strikeout Percentile trend. This gives us an idea of the relative value of each pitcher that day and if there’s anything the field is missing.
Just like with the Park Factors and Weather Ratings, I toyed with various levels before landing on the final result:
Notably, when you expand this trend to the top 10% of pitcher ownership jumps nearly four percent, but upside actually goes down. Of course, upside is relative to a player’s salary, so it’s likely that total points are higher – but not by as much as salaries would indicate.
This is another example of targeting the next best groupings of players being +EV. Below-average pitchers just don’t have the stuff to miss bats on any given day. Above-average (but not elite) pitchers have the stuff to post the highest score on any given day. It’s just a matter of catching them at the right time.
With GPPs, we only need that to happen once or twice a year to turn a tremendous profit.
My intuition is that by tinkering with various metrics, we’d see a similar pattern across the board. There may be variation in the strength of the effect, but the concept remains the same.
The biggest takeaway is that the most valuable pitching metrics are already accounted for. Sometimes via salary, and sometimes ownership. Our projections are enough to give you an edge, particularly for cash games.
However, if you’re trying to take down a GPP, digging deeper is valuable.
By targeting the “next best” options in a variety of metrics, we can find undervalued pitchers with a chance to post huge scores. Beyond that, most of the field fails to take into account the impact of umpires on fantasy scores.
I’d suspect that eventually changes, but for now, we can use it to our advantage.