DFS at the end of the season is generally a bit more difficult, as the regs simply won’t go away and the pool of extremely casual players is generally reduced. This creates an interesting dynamic in GPP, as we are not dealing with a static environment like in cash games. It becomes difficult to simply find the best value plays of the day and mash them in with abandon, since a larger % of the field will be aware of these plays, thus driving up the overall ownership of the obvious plays.

In the most absurd scenario, you’d end up with a situation where everyone had the same lineup in a 50,000 man GPP and everyone loses the rake. Obviously it hasn’t come to that, but as competition gets tougher it is paramount to rapidly shift gears into going contrarian. I tested some hypotheses by running a simulation using Excel, doing the following:

1) Creating an optimal(ish) median value lineup (one you might use in cash)
2) Creating a 2nd derivative of the best lineup, but using none of the players in the first lineup
3) Creating a 3rd derivative using none of the players in either lineups 1 or 2
4) Comparing how each lineup would do vs. one another in a 5000 trial simulation

Lineup 1 projected score: 289.6
Lineup 2: 278.9
Lineup 3: 273.88


LU1 defeats LU2 55.44% of the time
LU1 defeats LU3 65.94% of the time
LU2 defeats LU3 59.3% of the time

LU1 scores 325+ points 2.76%, 340+ 0.22%
LU2 scores 325+ points 2.44%, 340+ 0.44%
LU3 scores 325+ points 0.6%, 340+ 0.06%

I used 5,000 trials in this sim, yet the weaker LU2 somehow had a larger ceiling than LU1, which goes to show just how much variance there is in GPP. This is almost impossible to be realistic over a much larger sample.

At first glance, it seems logical to take the approach of taking a lineup something like the 2nd one, or just using some combination of the lineup 1 players plus some of the lineup 3 players to be a bit more contrarian. This is a popular strategy and certainly may be a great one, I honestly do not know.

My personal strategy is to simply aim for a score of 340 on DraftKings on most nights, while trying to avoid being too chalky. Normally if you hit 340 on a night when no chalk goes off, you’ll do quite well. In order to do this, I’m looking for my players to score points between 6.5-8x+ their salary using this very rough scale:

$4,500 and under: Ideal target of 8x salary or higher
$4,500-$8,000: 6.7-7x $
$8,000+: 6.5x $

In order to best select the right players to accomplish this task, we have to accurately gauge ownership and create a plan based around this. My personal motto is to generally avoid players who are going to be owned at a percentage much greater than their actual odds of hitting in the desired target range. Let’s say for example you’ve got a $10,000 player projected at 55 points (5.5x salary) with a standard deviation of 25%.

Using some rough math, I calculated that this player has approximately a 22.9% chance of hitting the target of 6.5x (65 points). If you wanted to get a bit more ambitious with your target and make it 7x, his odds of meeting that is 13.62%.

Therefore, if I thought it likely that this player were to be owned much more than 23% in the contest, I would strongly consider a fade. One brief aside: in smaller contests, you might only need a score of 325 or so to win on some nights, and therefore you’d only need something like 6.2x salary to hit a desired target. The odds of this player hitting 6.2x is 30.2%, so according to my guidelines this player is reasonably good if he’s only owned 30%. Remember that the larger the field and more chalky you play, the higher your target ratios should be.

Now let’s consider choosing a similar salaried player who might be $9.500, but is only projected at 46.55 points and the same SD of 25%, for a ratio of 4.9x salary. This player is almost guaranteed to be much lower owned than the latter expensive player due to his relatively poor value, but how likely is it that he will hit 6.5x-7x (61.75-66.5 pts)? According to my simulation, his odds of hitting 6.5x are 8.96% and 7x is 4%.

This is a close situation which often comes up a lot, and it’s important to realize its importance when playing in tougher fields. In a weak field, I might choose player 1, hoping the field is soft and he’s only owned 20-25%.

However, in a tough field, I can expect him to be owned 50-60%, which makes him a very clear fade where I can choose presumably very low owned player 2 as leverage, since no “rational” reg would ever choose to spend heaps of salary on the 4.9x player over the 5.5x one.

A very important distinction to make in DFS is that it’s effectively a sports betting trading market, where the values of players are often quite relative to their ownership percentage and the ownership of similarly priced players.

Let’s imagine player 1 is owned 50% and player 2 is owned 8%. First off, player 2 is a 31% underdog (a bit less than 2:1) to outperform player 1 straight up, but is getting ownership odds of (50%/8% = 6.25 to 1 odds. Needless to say, getting a little bit worse than 6.25:1 on a 2.2:1 proposition is quite profitable.

In terms of odds of hitting a ceiling, the odds of player 1 hitting 6.5x is 22.7%, while the odds of player 2 doing the same is 9.22%. In this case player 1 is about a 2.46:1 favorite to hit ceiling.

This may sound like an impossible task, but the whole process of creating a GPP lineup comes down to weighing ownership vs ceiling upside and creating some system to help pick the best of the two. The ideal player is one who has actual value, but won’t be touched because he’s been playing horribly lately or is just overlooked on the slate.

One player who was very popular at all stakes a few games ago was Justin Anderson, priced at $3,100 on DraftKings. In a major contest he was owned at about 50%, and in some higher stakes contests it was closer to 80%. According to my projections I had him at about a 24.3 value, making him a 7.83x. My target value for him would be about 8x, needing a score of 24.8. His actual odds of scoring that is about 46.8%, meaning he’s an okay play if he’s owned 47% or less, but probably not very good if he’s going to be owned 80%.

To give you a quick idea of how good a player would need to be at 80% owned even if you’re only looking to get 7.5x out of a player like Anderson (perhaps in a smaller field), he would need (at $3100 salary) to be projected at around 31 points or 10x salary. Once a guy starts getting into the 10x+ range, it’s really not that big a mistake to just unload the clip, but always be careful with anyone projected at much less!

Again, these are only my opinions and are not proven based on any in depth analysis, just merely a guideline to help you make solid contrarian decisions on slates where it feels like you’re getting bombarded by tons of value plays. It’s not fun playing the longshot and losing usually, but it is the best way to hit the big payoff.

Previous post

NBA Sneak Peek: March 31, 2017

Next post

NBA Sneak Peek: April 1, 2017