Analyzing NBA DFS Perfect Lineup Trends

Our algorithms here at numberFire run a script every day that gives us the "perfect lineups" from the previous day of NBA DFS. Our own Michael Worthington built a script to compile the optimal lineups from previous days, and our DFS guru, Sam Hauss, puts it out in article form and includes all the major DFS sites. We've been doing it a little over a month now, so I wanted to take some time and analyze some trends from the data.

A couple notes before we start. There are a lot of great articles and books that analyze thousands of user lineups and give data to the readers based on those. For example, Jonathan Bales, in his great books, looks at a lot of user data from DraftKings and analyzes what it takes on average to cash in each type of game. This is not really the same type of article. We're not looking at the minimum score needed; we're looking at how to get the highest score possible.

Also, I'm not going to break down different types of DFS games -- H2H's, 50/50's, or GPP tournaments -- because a perfect lineup is a perfect lineup. We're looking at how to maximize points. This might be more useful in GPP's, admittedly, as some of the "perfect" plays weren't safe. However, the lineup data and analysis should provide some insights and clues into roster construction among all DFS games.

Ok, let's begin with FanDuel.

FanDuel Data

First, I looked at the average salary and average fantasy points for each position. I merely pulled the salary for each perfect lineup we've posted and averaged them. Each position other than center has two slots -- for our purposes, I counted the higher-salaried player as the "1" spot in the perfect lineups.

And here is the breakdown by percentage of the salary cap, which is $60,000 on FanDuel. Put simply, the PG1 spot -- average salary percentage of 14.1% -- means that on average, the PG1 spot takes up that amount of the cap in a lineup.

However, the average doesn't really tell us the whole story -- the center position average salary allocation is about $6,200. But what does that mean? Does that mean you should optimally look for centers around that price? Not necessarily, as our perfect lineup could have had a lot of $8,000 and $4,000 centers -- still an average of about $6,000.

To look deeper, I broke down the perfect lineups by position and looked at which type of players were found at each position. I broke it down into four categories, which I made up myself to make things easier -- Super Saiyan ($9,800 or more), Stud ($7,100-$9,700), Mid ($5,100-$7,000), and Value ($5,000 and lower). Here's what the percentages looked like.

The pie charts show some really interesting trends that could perhaps shape the way we build our teams. Let's look at what the data is telling us.

Pay Up for Point Guards

A whopping third of perfect lineups included super-high priced point guards ($9,800 or more). And when the lineups weren't taking the top guys (SS), they were taking Studs right below them often. None of the perfect lineups had values at both point guards positions -- there was always a top PG.

And the second point guard (for our purposes, I always counted the lower-priced PG as the second one) had the highest percentages of Studs at 33%. That means one-third of the perfect lineups had both point guards spots filled with guys $7,100 or more. Pay up for your point guards.

The why -- I'm musing here, but I'd venture to say that this trend is just due to being a ball-dominant player. It's the same reason the top small forwards, Durant and LeBron, are so valuable -- they handle the ball. A lot. A fair parallel would be the quarterback position in NFL DFS. They have the ball in their hands almost every play, so they have many more opportunities to score fantasy points.

Take a Value Small Forward

Look at that pie chart for the second small forward spot. Such a higher number of the perfect lineups -- 86% of them -- included a low-priced value at the small forward spot. While it seems like it would be a good idea to fit both KD and LeBron in your lineup, the perfect lineup data says no. In fact, only the point guard spot had any perfect lineup where there were two SS ($9,800-plus) guys at the same position.

The data also had an interesting trend that I'd like to bring up briefly...

Does Stacking Work in NBA DFS?

This has been a much-debated topic in daily fantasy world. It is clearly a viable strategy in the rest of the major sports -- NFL, MLB, and NHL -- but the consensus has always been, in NBA DFS, don't do it. It's fairly intuitive -- in MLB, for example, it raises your potential lineup upside. If you've stacked consecutive hitters, if one of them is on base while the other hits a home run, you get points from the home run plus the extra run from the player on base. When you stack a team, if that team goes off for a lot of runs, you do well with a stacked lineup.

I found that 86% of FanDuel perfect lineups and 90% of DraftKings perfect lineups included at least one pair of teammates. There is probably a lot of noise to those numbers, but I'm going to continue to study this trend as the season progresses and try to research this topic more in-depth. I don't have a direct answer on whether you should stack in NBA DFS, but the data is telling enough to look deeper.

DraftKings Data

Here is the average salary and average fantasy points for the perfect DK lineups. I slotted players into positions by salary -- if there were three point guards in the perfect lineup, the highest-priced one I slotted in the "PG" spot, the second highest-priced went into the "G" spot, and the lowest-priced point guard went into the "Utility" spot.

Again, the breakdown by percentages.

And again, the groupings by positions.

Pay Up for Point Guard and Power Forwards

The point guard data for DraftKings was fairly similar to FanDuel -- a lot of lineups included a high-priced point guard. However, there was an even more important position on DK: power forwards.

Only 9% of the perfect lineups had a value player at the power forward spot. It had the highest percentage of Super Saiyans at 38% and Studs at 29%. Lots of value could be found at the small forward, center, or utilities positions (G/F/Util), but often the perfect lineup had a high-scoring power forward.

Another thing to remember about DraftKings as opposed to FanDuel -- 3-pointers are worth an extra half-point and double-doubles are worth an extra 1.5 points. That gives shooting point guards (like a Damian Lillard, for example) and double-double big men (Blake Griffin) extra points for just putting up their regular statistics. Additionally, rebounds are worth 1.25 fantasy points on DraftKings and just 1.2 points on FanDuel, giving the bigs a slight bump on DK.

Don't Pay Up for Small Forwards

The FanDuel lineups had a good number of small forward Studs. However, rarely did the DraftKings lineups have high-priced small forwards. In fact, over 3/4 (76% to be exact) of the lineups had a player $7,000 or less in the small forward spot.

Again, if I may muse for a moment, I think this might have to do with positional scarcity. On FanDuel, you can afford to take a luxury (when there are only two or three players that are elite, you can consider it a luxury -- consider the tight end position in fantasy football, for example) because the salary cap at $60,000 is higher. LeBron at $10,000 on FD is only 16.7% of your cap; on DK, $10,000 would be 20% of the cap (DK's salary cap is $50,000).

It's definitely OK to take a center in your utility spot

The Utility spot almost always included the lowest-priced player of the lineup, and it was almost always a minimum-salaried player, or close to it ($3,000 on DK). Here's how often each position was in the Utility spot.

The reason that the center spot is a little more important on DK than FD is probably due to the extra points for double-doubles and small bump in rebounding, as mentioned above. Even a mid-range or value center that is getting a spot-start for an injured player can get close to 10 and 10. If he hits it, he easily exceeds value and gets the bonus.

Again, we've only been doing our perfect lineups for about a month, so it's definitely possible that some of this data will change as we get larger and larger samples. I'll revisit this a bit later, but don't hesitate to contact me and ask any questions you may have. Good luck!