The Church of Betting

Winning at DFS & How to Optimize Your Lineup

DFS trophy

My journey in the DFS universe continues with my next article from the series. In my introductory article I briefly covered calculating expecting points and the different DFS providers available. Today I will talk about which tournaments are best to compete in, my biggest win yet, how to optimize your line up and some other lessons I learned while making my first steps in the football fantasy world.

Tournaments

As already pointed out in the previous article, I am currently playing DFS on the FanTeam website where I focus on football tournaments. The biggest tournament by far on FanTeam (currently ~800 weekly players) is the weekly Premier League (PL) challenge, which is also the one I regularly participate in. Since preparing the weekly Premier League line up takes quite a lot of effort (it takes some time to analyze 20 teams worth of players) I was drawn to some of the smaller tournaments during the week with considerably less games. The Champions League (CL) tournament had only 7 games this week and strong participation (about a quarter of the PL tournament participation) and there were also batches of 1-4 international games per day the week before during the World Cup qualifiers.

My thinking was the following: fewer games per tournament mean less time spent doing player analysis. Meanwhile my model should still be able to identify the optimal team, thus giving me some positive return. In fact, it should be even easier to identify the optimal team since a lower total number of games means less variance and a smaller pool of players to choose from. Therefore, even though the prize pool is smaller I will need to invest much less time so my return per hour spent might actually end up being higher.

The 1-game tournament

I decided to take that theory to the extreme and joined a bunch of 1-game-tournaments that were available during the week. Calculating the expected points of the 22 players took no time at all and assembling the optimal team was an easy task. I was already excited about how much faster and easier that was compared to picking my weakly PL team. There was only one problem.

Picking a team from a pool of 22 players turned out to be exactly as easy for my opponents as it was for me. In one instance the tournament running on the game England-Lithuania had 15 participants and almost all of us seemed to have picked more or less the same team. At the end a few impact bonuses were given away to the players, which have left me 1 point behind the first place. And 1 point behind the first place in this case translated to 9th place out of 15 and leaving the tournament with empty hands!

I have learned my lesson. In DFS you should avoid the low-hanging fruit. If it seems too easy it probably is, meaning everyone can do it and whatever advantage you had quickly disappears. From now on I will avoid the 1-game tournaments and I’d advise you to do the same.

PL and CL tournaments

The Premier League and Champions League tournaments are a different story altogether. The 7 games from the last Champions League round already required some time investment to pick the team, plus all games started at the same time meaning no guess work for the starting elevens was needed. I have finished my only CL tournament so far at place 39 from 183 players, which was a few points short of the prize pool, but still showed some profit-potential.
And what about the Premier League challenge? At the end of the day for me it is the one offering the best value-for-the-money from all tournaments. Yes, here you would need to invest the most time in player analysis. Yes, you will also get the highest variance meaning you could (and often would) end up with nothing to show for it.

However, you will also get the highest number of participants, meaning the average skill of your opponents will be the lowest. In poker speak, you will find the most fish in this pond. This will increase your average rate of return and in the long term that is all that counts. After all, remember you need to recover your 10% rake and earn something on top of it, so 12% return-before-rake is two times 11%, and with 9% you are still in the red! Therefore, in DFS you should never sacrifice expected return for easier lineup preparation, lower variance or whatever. Your opponents will get to use the exact same benefits and you will probably fail to recover your rake. The only viable options are to go big or go home.

My biggest win

This brings me to my favourite part of today’s article – the one where I get to brag about my biggest win yet in DFS! Last week I’ve finished 20th from 784 players in the weekly PL challenge. This has earned me a prize of €100.08, beating my previous record of €58.44 from an international round. Here are the guys who brought home the sweet prize:

Fantasy Lineup

I was lucky to have Raheem Sterling playing a great game for Man City who smashed Stoke 7-2 with 1 goal and 2 assists from my captain. Overall I had variance on my side this day which lead to this pleasing result. Yet you would notice I wasted quite some budget on Danilo who didn’t start at all, which again underlines the importance of guessing the starting eleven. This has cost me dearly this week when two players of my twelve picks did not play, which has effectively thrown me out of the competition despite all the others earning at least as much points as my model expected them to. I make another note here: guessing the starting eleven is a big part of winning at DFS and I have to work further on improving this part of my model.

Optimizing your line-up

However, there is one thing my model does really well and it is optimizing my lineup conditional upon the expected points I’ve calculated. I have written in the introductory article that this would be something complicated to do as it includes optimizing a linear system of equations under a number of conditions. I was partly right – the optimization indeed required solving a linear system – only doing it turned out much easier than I expected! Searching for a solution of my optimization problem, I have first stumbled upon some articles featuring pages of complicated R and Python codes, until finally arriving at a Youtube Video about the Solver Excel Add-in which, apparently, solves linear systems just fine. Watch the video and thank me (or rather its author) later.

Conclusion

Alright, that was my latest fantasy football report, the second of what I hope to be a long series of articles on DFS. Let me know of your experience with DFS or any questions you might have in the comments. You might have also noticed that I have slightly upgraded the design of the page. I hope you like it.

Thanks for reading and till the next time!

6 thoughts on “Winning at DFS & How to Optimize Your Lineup

  1. Thanks, that’s interesting.

    I play on a local site and I’ve struggled with whether 1 game slates are beatable (i.e can skill overcome the inherent variance). I feel like I’m making good decisions, but the salary is such a non-constraint that anyone can build a good team…

    • That’s similar to my experience with 1-game tournaments. At the end your team ends up similar to the others and there is just not much you can do to stand out. Which site do you play on if you don’t mind me asking?

      • I play on moneyball.com.au because it has the local Australian league (A-league). I spent many hours building a database with all the players and stats etc., but then when the season started I realised the only volume was on the 1-game slates.

        There’s no late swap feature, so playing all-of-round slates is high variance. I usually bias the first game so I know the slate, but maybe pick a keeper and a defender who is the captain (guaranteed starter almost) from a later favoured game. I’ve had good success in all-of-round, but the latest tournament only had 17 entries ($2 buyin…)

        • Congrats on your database with player stats, I am looking forward to building one myself.

          I had a look at the site, it seems to be more focused on basketball. 17 entries with $2 buyin really is a small ticket. But you seem to be on the right track. The site also seems to be relatively new (started on February 2015) so there is still a lot of room to grow. Once the player base for Australian football starts growing you will reap the benefits of your effort. Good luck!

          • Thanks

            While the player base is so small, I’m looking into whether I can use my player database for sportsbetting. General idea is to check announced lineups for games when released (1 hour prior), and quickly check if they’re missing any key players (injury or bad manager choices). If those are unexpected, they might not be in the betting prices and might have a few minutes to take advantage.

            I have a 9-day free rotowire subscription so I used their projections and solver in excel to build a team in yesterday’s NBA tournament on moneyball. My team was good. I placed 49th out of 1050 but only won 2x my entry fee :). That’s the thing about when tournaments get massive: even if you’ve got a great team, you’ve still got to get lucky. Also, other people are entering in up to 50 teams (not common, but super common to put in 20), and I’m not willing to do that in the absence of better proof I’m profitable.

Leave a Comment