Winning Consistently at NBA DFS: The Intelligent Guide to Growth

I'm going to let you all in on a little secret. You may already know this, but maybe you don't. NBA basketball offers you the best chance at actually being profitable on DFS sites. Why? Simple. Basketball is the most consistent and predictable sport. It is very low variance compared to football, baseball, hockey, soccer, etc.

So if basketball is easier to project, why doesn't everyone win? Because the only thing that really matters when assessing DFS profitablity is the quality of the opposition. Most DFS pros will tell you this: "Basketball is very rewarding to those that get it". So what is "it" and how we get it?

First, let's explore what we know about the NBA. It is what we call a star-driven league. The NBA in general is skewed in terms of production to the star players. The true superstars of the game in general have a very large projected median average and will consistently stay within the standard deviation of that average. However, they also have the potential to produce incredibly high numbers on any given night. In DFS, we refer to that potential as ceiling.

What this means is that variables like matchup ratings and Vegas lines don't matter quite as much as they do in the NFL for instance. The art as always, is in the lineup construction.

I've been able to be profitable in NBA matchups since I started seriously competing this season by following a very simple strategy: Pay up for the studs, pay down for the scrubs.

My goal is to build a high floor and a high ceiling, which actually enables me to enter my lineups in both cash games and GPPs on some nights. You have a very simple roster construction on DraftKings: PG, SG, SF, PF, C, G, F, UTIL. The 8 roster positions allow you to play up to 3 players of the same position except at Center.

My goal generally is to find a median projection of 270 total points. There are a few ways you can do this, but generally I will always look to go stars and scrubs method. This means I will be using 40-60% of my salary on a given night on 2 or 3 stars, and allocating the remainder on scrubs. My expectation is that if I'm going with 2 stars, they will combine for 100 points, my 2 higher end mid-level players will combine for 80 points, and my 4 scrubs will combine for 80 points.

This is what I refer to as a 3-tier stars, high-end mids, scrubs lineup. Generally when I say "stars" I mean anyone priced in the $9000 and up range, "high-level mids" are those priced in the $7000-$8400 range, and scrubs are generally going to be $3000-$4500. You will notice I am generally ignoring most of the players in the $4500-$7000 range. There are definitely nights when there will be some good plays from this range, but for this particular slate of games, I choose to go with this method.

Lineup Example

The lineup I constructed last night was the following:

PG: Raymond Felton $3400 Scrub
SG: Eric Bledsoe $8300 High-End Mid
SF: Jared Dudley $3000 Scrub
PF: Anthony Davis $10,600 Star
C: Brook Lopez $7400 High-End Mid
G: Wes Matthews $4400 Scrub
F: Lebron James $9700 Star
UTIL: Alex Len $3200 scrub

Salary Cap: $50,000
Salary Used: $50,000

Percentage Breakdown

Cap Percentage on Stars: 40.6%
Cap Percentage on Mids: 31.4%
Cap Percentage on Scrubs: 28%

Notice that I used the full $50,000 of the cap. This isn't always necessary, but generally I do come very close to full cap when optimizing.

What you'll notice is that I'm deviating from the mean salary. I'm doing this on purpose to create a high floor. If I constructed a team of 8 players near the mean salary of $6250, I would have a very low ceiling and floor. The players that are priced at that level have very little potential to exceed their projected totals outside of the standard deviation, and also have higher potential to fail to exceed the projected total below the standard deviation.

Note: The standard deviation is a mathematical term derived from statistics used to calculate variance. It is usually calculated by taking the square root of the variance.

Let's say Anthony Davis is projected to get 51 points. This isn't a prediction of his actual score, it is simply the median (average) of his range. What is his range? Usually you have to pay for sites that will tell you projections for floor and ceiling. Let's say his ceiling is 76 and his floor is 32. The lowest point from the mean is about 32 which represents his floor. The highest point from the mean is 76 which represents ceiling.

Of course, there can be outliers. If Anthony Davis gets hurt in the first few seconds of the game and doesn't return, we have a 0 output. We have to remember that statistical models don't have an accurate way of predicting outliers. If they did, whoever figured it out would have billions of dollars within a couple months, even if they only started with $1.

Now suppose Anthony Davis also has a career high night and scores 85 DraftKings points. That would be 9 full points above the standard deviation. That is the kind of outlier we want, not the alternative. Anthony Davis has that kind of potential. Nobody else on his team is going to have anywhere near that kind of ceiling on any given night.

Projections

So, I calculated my projected points last night and arrived at a number of 272. This is ideal as 272 is usually a good number that will win cash games in NBA most nights, especially on a night like last night, which only featured 12 teams playing.

My total score on the night? 292.5. A very nice score. Most of the players I selected performed slightly less or slightly more than his expected production. However, Alex Len achieved 43 points at a salary of $3200. He achieved 13X value which is basically an outlier. 10X value is usually considered A+ return, so 13X is A+++. Len's ceiling night helped bump me from the standard projection of 272 up to 292.5.

Why was Alex Len priced so low? Because he generally doesn't play enough minutes to produce. Minutes = opportunity. Opportunity is EVERYTHING in basketball. If your player isn't on the court very much, there is nothing he can do to accrue a large number of points in limited minutes. There are no 90 yard touchdowns off 1 target or grand slams from 1 AB in the NBA. Production is steady, slow, and more consistent.

We had information that Alex Len was going to start last night because Tyson Chandler was injured. Len suddenly goes from a reserve player, to a starter, which means a huge jump in opportunity. The DraftKings algorithm doesn't take this into account, so we had a great opportunity value play last night in Len. We would have been happy to see him hit 7X value and get us 21 points or so, but he basically doubled his expected output and that was the difference.

So how did my lineup perform based on goals?

Stars Expected Points: 107 Actual Points: 108.25 Result: +1.25 (Average)
High End Mids Expected Points: 85 Actual Points: 71 Result: -14 (Below Average)
Scrubs Expected Points: 80 Actual Points: 113.25 Result: +33.25 (Well Above Average)

I achieved almost exactly the median from Lebron's 47.25 and AD's 61.
The High-End Mids underachieved and resulted in -14 points below average.
The scrubs overachieved and resulted in +33.25 above average.

Projected Average: 272
Projected Average 272 + 1.25 (Stars) - 14 (Mids) + 33.25 (scrubs) = 292.5

So you can see in this case, I relied on the consistency of my stars, got less than expected from the mids, and way more than expected from the scrubs.

Information is Key

Notice how I said we had information that Alex Len was going to start? That is one of the biggest keys to NBA lineup construction. DraftKings doesn't adjust pricing on the fly when injuries and lineup adjustments happen, so we have to be able to in order to find our "scrubs" who are suddenly not so scrubby.

If you aren't prepared to monitor news right up until the games start, you're taking an enormous risk. To succeed in NBA DFS, you have to be able to adjust your lineups on the fly based on up to the minute information. Don't use twitter? You need to. You need to follow the sources who provide this information because it will always be on twitter first.

I hope you enjoyed this post and found it informative. I will continue to explore my thoughts and strategy on NBA DFS, and I'll still talk about NFL as well.

-- DF

Derek Finley

Existing in a world of psychotic monkeys. Writing about sports for passion. Thirsting for knowledge and beer.