DFS Tournament Strategy: A Success Story

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.

There are two important aspects of this DFS tournament strategy: understanding player covariance, and understanding player ownership. The first, player covariance, boils down to the fact that player performance is not independent. When a player does well, it directly affects the projected performance of other players. When Mike Trout gets a hit, Albert Pujols is more likely to get an RBI, and the rest of the Angels are more likely to get an additional plate appearance. When building DFS lineups with average projected points, the underlying assumption is that each player’s chances of reaching “value” is independent. However, using tools like Conditionals on SaberSim, one can take advantage of player covariance, putting players together that do better when the other does well and vice versa.

The other important aspect of DFS strategy outside of player covariance is ownership, with which game theory plays a huge role. In any particular day, there are millions and millions of potentially lineup combinations, but the biggest MLB DFS tournaments have “only” 100,000 entries. This means that you don’t need to build the “optimal” lineup in order to win one of these tournaments — you just need to build a lineup that scores more than others. Game theory can be utilized in order to better your chances of doing so: if a player is high owned, and they have a poor performance, then not choosing that player in your lineup gives you an edge over a large percentage of the field. In baseball in particular, even the best players have a significant chance of underperforming, making this strategy hugely beneficial. If you can accurately predict who will be high owned, or who will be low owned and has a chance at a big game, you can drastically increase your odds of beating the competition, even if you’re sacrificing your “average score” in doing so.

Yesterday, I was lucky enough to win a big tournament on DraftKings (taking home $10,000 on a $27 entry), in part by utilizing the above strategies.

To start, I used a Conditional on the opposing pitcher for the Cubs allowing >4 runs, which gave me a number of Cubs in the lineup — this is the player covariance piece above. Because the Cubs scored a lot of runs, each player had more plate appearances, and could score points from driving in or being driven in by other players in the lineup.

The other crucial choice I made was to fade — or not play — both Drew Pomeranz and Kris Bryant. I felt that both would be highly owned, especially Pomeranz, and knew that I could gain a big advantage over the field by avoiding them, despite their high projections on SaberSim, and hoping that they would hit a lower part of their projected point distribution. Though Bryant wasn’t as highly owned as I expected and had a big game (though not as big as Donaldson), fading Pomeranz ended up being a fantastic choice. He pitched poorly, and was owned in 72% of lineups, giving me a leg up on a huge proportion of the field by playing Julio Teheran instead.

My big win was of course largely due to good luck, as is any DFS GPP win, but the lesson remains that utilizing strategies and game theory to increase lineup upside and take advantage of ownership can lead to success in the long run if you’re playing in large field tournaments.

DFS Projections: Batters
The early simulation results have many teams projected for high run totals. The simulator projects over five runs per game tonight for the Astros, Pirates, Rockies, Yankees, Blue Jays, Cardinals, Twins, Mets, and Nationals. Furthermore, there is only one team, the Indians, with a projection of less than four runs per game.

Top Offenses
1. Rockies (vs Jeff Locke) – 6.84 r/g
2. Pirates (vs Chad Bettis) – 6.28 r/g
3. Cardinals (vs Brandon Finnegan) – 5.46 r/g
4. Twins (vs Tom Koehler) – 5.27 r/g
5. Blue Jays (vs Tyler Wilson) – 5.20 r/g

Seven of the top ten batters for DFS are playing in Coors field today. Among those, Nolan Arenado ranks first, Andrew McCutchen second, and Starling Marte third. Ryan Raburn has the cheapest price tag on both DraftKings and FanDuel of the highest ranked Coors players. Because so many of the top slots belong to Coors players, an examination of the best options from the rest of the field will be helpful.

Top Batters by Avg Projection (Non-Coors)
1. Bryce Harper 11.48 DK, 15.58 FD
2. Mike Trout 10.65 DK, 14.23 FD
3. George Springer 10.20 DK, 13.47 FD
4. Giancarlo Stanton 10.10 DK, 13.63 FD
5. Jose Bautista 10.03 DK, 13.49 FD
6. Matt Holliday 9.82 DK, 13.26 FD
7. Curtis Granderson 9.74 DK, 12.95 FD
8. Eduardo Nunez 9.69 DK, 12.46 FD
9. Jose Altuve 9.67 DK, 12.38 FD
10. Ryan Braun 9.65 DK, 12.58 FD

Top Batters by 95th Percentile Projection (Non-Coors)
1. Bryce Harper 29.00 DK, 40.40 FD
2. Giancarlo Stanton 29.00 DK, 38.70 FD
3. Mike Trout 28.00 DK, 37.60 FD
4. George Springer 26.00 DK, 34.70 FD
5. Jose Bautista 26.00 DK, 34.90 FD
6. Ryan Braun 26.00 DK, 34.40 FD
7. Yoenis Cespedes 26.00 DK, 34.90 FD
8. Matt Holliday 25.00 DK, 34.40 FD
9. Curtis Granderson 25.00 DK, 34.40 FD
10. Jonathan Villar 25.00 DK, 31.90 FD

DFS Projections: Pitchers

Today’s slate of pitchers presents scarce options as far as big performance potential. The point projections are low and the prices are low as well. The top five pitchers by average projections mostly match the top five pitchers by 95th percentile projections, except for one difference. Jimmy Nelson jumps from #8 to #5, leapfrogging Adam Wainwright, when sorting by 95th percentile projections.

Top Pitchers (Avg Projection)
1. Nathan Karns (vs CLE) – 17.87 DK, 33.77 FD
2. Marcus Stroman (vs BAL) – 16.77 DK, 32.24 FD
3. Gio Gonzalez (@ CWS) – 15.57 DK, 31.10 FD
4. Collin McHugh (@ TEX) – 14.48 DK, 29.00 FD
5. Adam Wainwright (@ CIN) – 13.69 DK, 27.80 FD

Conclusion

There are many ways to create DFS strategies using average projections as well as percentile projections. SaberSim also offers Conditionals for optimization, as well as tools that allow you to include/exclude players, adjust exposure, and much more. To keep you fully informed on all the possibilities, we will continue to explore more strategies in the coming weeks. Also, remember to check back for updated projections throughout the day. As teams release official lineups, SaberSim automatically updates accordingly and reruns simulations in order to stay as current as possible. Best of luck.