DraftKings 2013 NFL Season: A Simulation
Editor’s Note: This article was written prior to DraftKings’ announcement of the $3.1 Million Fantasy Millionaire
Hi guys. This article examines DraftKings GPP offerings for this upcoming NFL Season. What I hope to show is the type of profit/loss and variance you can expect if you are a high-bankroll, multi-entry, type of player who is profitable in the long run and if you choose to go hard in these contests. I will introduce the tournaments and strategy, walk through each week, and then provide summary statistics for the 17-week season as a whole. Lastly, I touch on the topic of bankroll management. However, in order to make this model work (i.e. in order to code this) I needed to make a number of assumptions, which I will be totally upfront about and comment and how I think they may affect things. Because of this, I urge you to not take any numbers I present to you as gospel. I hope that you merely find this interesting, perhaps learn something, and figure out how the results of my simulation may differ from the way you play the games. Keep in mind that every person is different.
Overview of Tournaments & Entries
Below is a table of most of the major tournaments DK is offering Week 1. Let’s assume we play all of these contests every week except the million dollar contest which DK decides to run once every four weeks (assumption). In the third column you can see the number of entries we are going to put into each of these contests. This is fairly arbitrary and is just meant to represent a high volume GPP player. There will also probably be overlay from time to time and I therefore attached an overlay estimate to a few contests. While I assumed consistent overlay every week, it is certainly possible that the overlay is higher early in the season and lower as the weeks go on. Lastly I attached a 5% ROI pre-overlay for every contest. Since the lower buyin contests are raked at a higher percentage than the higher buyin ones, this means that we are actually doing better pre-rake in the lower buyin contests, which makes sense.
[Assumption #1: I am not taking qualifiers into account. If you do well in qualifiers and were able to lower your cost of entry, than all of my figures are too pessimistic for you. However this is probably balanced by other assumptions later on, which you shall see. Also any bonus that you are working through will raise your expected profit.]
The Season Simulation
Following the method that I have used in past articles, I ran a Monte Carlo simulation that randomizes our finish in these tournaments, subject to the criteria noted above. I then summed up profit/loss for all of our entries and for each week and simulated nearly 6,000 seasons. The table below shows the average of the results.
[Assumption #2: I assumed complete independence among each entry. This is likely not the case in reality. If you think a certain RB is a particularly good play, for example, and play him on 5 of your 10 entries into a certain tournament, then this builds correlation to your entries. In this case, my variance numbers will be too low for you. There is probably a way to build correlation into the simulation but that frankly tests the limits of my computer programming skills.]
[Assumption #3: I assumed that there is an equal probability to cash in each position. It is certainly possible to build lineups with an asymmetric expected finish, that is, a higher probability of finishing 1st but a lower probability of min-cashing, which would raise both profitability and variance.]
You’ll notice that our entry fees tick up in weeks 1, 5, 9, 13, and 17. These are the weeks we are playing the million dollar contest. Similarly, average profit is highest in these weeks since we have the most money in play. ROI bounces around a bit since the process is very random, but in the end settles to 8.0%, which is a number that makes sense given our 5% ROI assumption + some occasional overlay. The probability of having an unprofitable week is around 64% most weeks and 71% in the million-dollar contest weeks. The standard deviation numbers are high, but pretty hard to make sense of in my opinion. Let’s look at it from another angle.
Bankroll Management
From above we see that we need $1,096 every week and $3,096 every fourth week to play these games. A starting bankroll of $28,632 for the season would cover everything. In this case the only chance we would have of going broke would be losing every single contest we enter – which can basically never happen. Therefore we do not need that much since we pick up expected profit each week along the way. Let’s play around with varying starting bankroll amounts and see what happens.
I simulated the season nearly 6,000 times again but this time counted how often we went broke on each level of starting bankroll. The results are shown here:
What level of bankroll is appropriate is different for everyone but personally, I would not feel comfortable with anything more than a 5% chance of going broke. Therefore, in order to play the tournaments in the manner in which we have laid out, I would want a minimum of about $9,000 to start. Of course, if I do not have this much, then I would scale back the amount of tournament I would play, entering a particular tournament 10 times instead of 20 times, for example.
That is all. Once again, I encourage you to not read too much into any specific number but rather take away general themes. In my opinion those themes are the following:
- Variance in tournaments is driven by the low percentage of people that cash and the top-heavy payout structures. The second thing we can’t do anything about but take advantage of overlay to improve the first.
- Expect higher variance in the Million Dollar contest weeks since that contest is especially top-heavy.
- Figure out how much you want to wager for the season at DK. If you make sure to diversify your lineups a lot and take advantage of overlay, and if you are a profitable player, you can afford to wager up to one-third of your bankroll weeks 1, 3, 5, 9, and 17 for the Million Dollar contests. However you must reduce this to about one-ninth of your bankroll every other week. When in doubt please err on the side of caution. If any of the above assumptions do not hold for you, then you must re-evaluate your strategy.