In this series, I often talk about player covariance — or the effect that a player’s performance has on his teammates and opponents — and its importance in building DFS lineups. This week, I’d like to expand on some nuances within that topic by looking at a visualization of this phenomenon.
Maximizing the upside of your lineups is crucial to having success in DFS tournaments. Oft-discussed strategies like stacking and targeting home run hitters are important for upside, but another strategy is to target bad bullpens.
This week over at SaberSim, I released a tool that allows users to view more detailed projected performance for their lineups. Rather than just adding up projected points for each player, this new Lineup Analysis tool allows us to view mean, median, standard deviation, and percentile projections for the lineup as a whole. In other words, rather than combining each player’s distribution separately, SaberSim analyzes the performance of the entire lineup across each simulated game.
So far, this weekly column has largely focused on various general aspects of DFS strategy for the first half of the post, and specific projections for the day in the second half. Today, I’d like to switch gears a bit and discuss my process for building lineups in small (2-5 game) slates, using today’s early 3-game slate as an example.
Last week, I discussed the importance of randomness in DFS, and some strategies one can use to take advantage of the large amount of random variation that occurs in daily fantasy. I’d like to expand further on that topic today by delving deeper into the process of focusing on specific portions of player projection distributions.
The past few months have brought a lot of debate about whether daily fantasy sports are a game of luck or skill. It’s a complex question from a legal standpoint, but from a purely logical and statistical standpoint, it’s fairly clear that both are involved. In fact, much of the skill portion of DFS relates to understanding the luck portion, and utilizing strategies to take full advantage of it.
In these weekly posts over the past couple months, I’ve talked a lot about moving beyond projections when building DFS lineups for large field tournaments, utilizing other strategies to increase upside and the chances of a big win. This idea was hard for me to accept for at first, and may be for some of you as well, but the more I’ve looked into daily fantasy and played it myself, the more I believe that it’s essential to utilize game theory in DFS, and move past using only mean projections in building lineups.
Traditionally, daily fantasy sports projections use average projected points as the primary method of evaluating players. While one can get a sense of a player’s consistency and upside based on their batting profile and game log, it is difficult to accurately and precisely project players’ upside in terms of DFS points and performance relative to each other.
SaberSim Daily Projections
Daily projections require a great deal of context in order to project each specific game. SaberSim daily projections account for lineups, starting pitchers, and bullpens, as well as more nuanced factors like weather, umpires, park effects, home/away, handedness splits, and more. Even within these specific factors, there’s a tremendous amount of detail involved, and constant room for tweaks and improvements. For instance, the park effects are not applied broadly, but rather based on how they affect each individual outcome (BB, K, 1B, 2B, 3B, HR) for left-handed and right-handed batters.
When I first began playing DFS, I approached it with the belief that a good projection system is all that is needed to be successful. I still believe that accurate projections are very important for all forms of fantasy baseball (I run a sports projections site, after all), but over the past year or so that I’ve played DFS, I’ve come to better understand the deeper level of strategy needed to be successful in large-field DFS tournaments (commonly called GPPs).