NBA
How to Dominate Single-Game Daily Fantasy Basketball With Perfect Lineup Analysis
FanDuel's NBA single-game offering gives us a unique way to watch a game. How can past perfect lineups help us build better rosters?

Single-game fantasy sports on FanDuel are all the rage. Single-game slates allow us extra chances for fantasy glory and extra fun when watching the NBA.

But how should you really approach single-game NBA DFS? That's a good question. I didn't quite have the right answer, so I figured I'd do something about it.

That's why I went back and ran optimal lineups for 225 slates from the start of the 2019-20 season until the mid-March hiatus. That's quite a large sample of games, so surely we can learn a thing or two about approaching building optimal lineups for ourselves.

Here are some takeaways I uncovered when looking for high-level trends, stacking, and more.

FanDuel NBA Single-Game Basics

It's important to know the ins-and-outs of FanDuel's single-game format. So, here are the ins. Or the outs. I guess it depends on your perspective.

To make a single-game lineup on FanDuel, you roster five players with at least one from each team while staying under the $60,000 salary cap. It's really that simple. There are, though, three multiplier slots.

Your MVP gets his FanDuel points multiplied by 2.0, your STAR gets his FanDuel points multiplied by 1.5, and your PRO gets his FanDuel points multiplied by 1.2. Your two UTIL players get the usual FanDuel scoring treatment.

This means that the same five players in a lineup won't score the same number of FanDuel points if the MVP and STAR are swapped. This is the real MVP about the format. For being such a straightforward game, there are a ton of layers to it.

Foregrounding Some Terms and Things (and Such)

There aren't too many ways to complicate this study, but unlike with my study on single-game NFL optimal lineups, I can't really refer to positions. In NBA DFS, there's no equivalent to a quarterback, and for some teams, it's the center -- not the shooting guard -- who is the go-to option. For that reason, I treat players just as players, not as positions. I'll rank them by salary, projection, or popularity, though.

I will refer to the MVP, STAR, PRO as such and may distinguish between the two utility players as UTIL1 and UTIL2 when it's relevant.

In only 6 of the 225 lineups, there was a tie for some position inside the top three spots (e.g. two players had exactly the same score and were, thus, tied for MVP rights). In these cases, I considered the higher-salaried player as the tie-breaker. This will have minimal impact throughout and will really mostly affect position averages (only marginally). Four more lineups had ties between the two UTIL players; again, I used salary to consider who was the UTIL1 and UTIL2 for the optimal purposes.

I will use "optimal" and "perfect" lineup interchangeably throughout.

Let's continue.

Should You Use All the Salary Cap?

This is always a fun place to start. How many of the perfect lineups used all $60,000 of the salary cap, and how many were way below that mark?

Salary Used in
Optimal Lineup
Frequency
$60,000 21.8%
$59,500 18.2%
$59,000 11.1%
$58,500 6.7%
$58,000 8.4%
$57,500 5.3%
$57,000 7.1%
$56,500 3.1%
$56,000 4.0%
$55,500 2.7%
$55,000 or Less 11.6%


In total, 51.1% of the optimal lineups used at least $59,000 of the $60,000 cap, but that also means nearly half of the perfect lineups left at least $1,500 unspent, and 11.6% of the lineups left at least $5,000 on the table.

This doesn't mean that salary is irrelevant, but definitely don't prioritize spending as much of your cap as you can. There's no one right answer, but if you want to build unique lineups. leaving salary unspent is a way to help.

Are 3-2 Lineups or 4-1 Lineups More Common?

Stacking is not necessarily a super common strategy in NBA daily fantasy because points can be scored by only one player at a time (sure, there are assists, but an assisted two-point field goal will net you just 3.5 total FanDuel points). But how does that affect lineup builds overall? Is a 3-2 balanced lineup more common than a 4-1 onslaught stack?

You betcha.

Optimal Lineup Split Frequency
3-2 Lineup 80.0%
4-1 Lineup 20.0%


These numbers were closer to 60/40 in NFL single-game optimals. Basketball is just a different ball game. In fact, in just two of these slates did an optimal lineup initially return a 5-0 split (I obviously then added the restrictions to get at least one player from each team), but even if we had the choice, a full 5-0 stack isn't ideal.

Because the answer to "when should I go with a 3-2 lineup?" is "almost always," here are some facts about the 4-1 lineups.

Of the 45 overall 4-1 lineups, 22 came from the road team, but only 10 of the 22 road teams to get four players in the optimal were even favored. Of these 45 lineups, then, 23 were from the home team, and 20 of them were favorites. We probably can't glean a ton from that, but home underdogs weren't a big source of the rare 4-1 lineups.

Here's the data split into winner and loser buckets (e.g. 17.8% of the optimals had four players from the winning team and one from the losing team).

Optimal Lineup
Win/Loss Splits
Count Frequency
4 Winners, 1 Loser 40 17.8%
3 Winners, 2 Losers 105 46.7%
2 Winners, 3 Losers 75 33.3%
1 Winner, 4 Losers 5 2.2%


The 3-2 split is, again, most common, and often, it's the winning team that features the three players in the optimal.

The Average Single-Game NBA DFS Optimal Lineup

Here's the highest-level look that I think I can provide. Here are some averages across the 225 optimals.

Optimal Lineup
Averages
Salary Overall
Roster
Rate
Rostered as
Exact Optimal
Position
Rostered
as MVP
Rostered
as MVP,
STAR, PRO
MVP $14,558 66.7% 34.1% 34.1% 62.0%
STAR $13,140 55.5% 18.2% 18.6% 48.9%
PRO $11,893 45.0% 11.4% 10.5% 34.0%
UTIL1 $10,020 31.4% 14.9% 4.1% 16.6%
UTIL2 $8,371 22.2% 14.5% 1.2% 7.8%


We see that the average salaries for each of the five positions do cascade in descending order, as we should expect. The MVP, on average, is the highest-salaried player in the lineup. However, the MVP's average salary rank is 1.79 within the lineup, not that close to 1.00. This is the first bit of information that should help us be okay avoiding the highest-salaried option as the MVP (more on that later).

That being said, the MVPs from these perfect lineups were rostered as the MVP by 34.1% of all lineups playing that slate. They were rostered, on average, on 66.7% of overall lineups -- at any position. That means that 34.1% of lineups correctly predicted who the MVP would be but that two-thirds of lineups rostered the optimal MVP at some level (and 32.6% of lineups rostered the optimal MVP at a different position).

This means a few things. Firstly, the MVP in the perfect lineups was rostered on 66.7% of lineups -- and at 62.0% in one of the three multiplier slots. That does mean a third of rosters didn't contain the true MVP of the slate. However, the MVP isn't -- on average over a large sample -- someone who comes out of nowhere.

The STARs across these optimal lineups were also rostered on at least half of all rosters, but they were rostered as the MVP only 18.6% of the time and were rostered as a multiplier on nearly half of the rosters.

So far, this kind of feels like it's saying to be okay to be chalky.

Here are averages with FanDuel points and projection ranks.

Optimal Lineup
Averages
Raw
FanDuel
Points
Multiplied
FanDuel
Points
Actual In-Game
FanDuel Point
Rank
Projected
FanDuel
Point Rank
Projected
Minutes
Rank
MVP 55.1 110.3 1.01 1.72 2.07
STAR 45.4 68.1 2.06 2.11 2.31
PRO 39.5 47.4 3.20 2.59 2.44
UTIL 1 33.4 33.4 4.69 2.68 2.71
UTIL 2 28.2 28.2 6.63 3.23 3.20


These look pretty much as expected.

What we do see here is that the UTIL2 isn't necessarily the fifth-best scorer in the game but can occasionally be lower than that in order for us to jam in the bigger performances. Let's expand on that quickly by looking at the frequency with which each scoring rank actually makes the optimal.

Actual In-Game
FanDuel Point Rank
Times in
Optimal Lineup
Frequency
1 223 99.1%
2 215 95.6%
3 202 89.8%
4 152 67.6%
5 140 62.2%
6 85 37.8%
7 56 24.9%
8 20 8.9%
9 12 5.3%
10 8 3.6%
11 6 2.7%
12 4 1.8%
13 2 0.9%


The top-two scorers find their way in the optimal nearly 96% of the time, and the third-highest scorer gets there around 9 of out 10 times. After that, it tapers off, and fairly often, the sixth- and seventh-highest scorer makes it. Why? If a stud puts up 70 points, it's likely more important to get him in your lineup than it is to get from 20 to 25 points for the bottom of the lineup.

This is an oversimplification, but: yes, while our goal should be to hit the top-five scorers, we shouldn't overlook ceiling from superstars just to get a better UTIL2.

Digging Into the MVPs

So, we know that you have to play the right five players to put up a huge lineup, but you also have to hit the right order of the multiplier slots. Let's look closer at the MVP picks and see what we can find.

Because of the way NBA DFS is constructed, it's a little tricky to bucket players, so I'm looking at things two ways: based on salary and then based on projections (from your dudes at numberFire, of course).

Here are the average MVP roster rates for players based on pre-game salary rank compared to the frequency with which they were in the optimal lineup's MVP spot. (For example, the highest-salaried player in a game has a salary rank of 1.)

In-Game
Salary Rank
Rostered
as MVP
Actual Optimal
MVP Rate
Leverage
1 53.0% 35.6% -17.5%
2 24.7% 24.9% 0.2%
3 10.6% 10.2% -0.4%
4 5.0% 9.3% 4.4%
5 2.1% 4.9% 2.8%
6 1.4% 4.9% 3.5%
7 0.6% 3.6% 3.0%
8 0.4% 1.3% 0.9%
9 0.2% 1.3% 1.1%
10 0.1% 0.4% 0.3%
11 0.1% 0.9% 0.8%
12 0.1% 0.9% 0.8%


All right, so the highest-salaried player is rostered, on average, 53.0% of the time at the MVP position. That makes sense. However, the highest-salaried player is the actual optimal MVP just 35.6% of the time. Is the most likely salary rank of an actual MVP 1? Yes. Is the highest-salaried player rostered as the MVP far more frequently than he is actually the optimal lineup's MVP? Also yes.

This means that a very easy way to be unique is to lessen your exposure to the slate's highest-salaried stud in the MVP spot.

Here's the same chart but with our pre-game projections ranking the players instead of salary.

Pre-Game
Projection Rank
Rostered
as MVP
Actual Optimal
MVP Rate
Leverage
1 57.6% 40.4% -17.1%
2 22.9% 20.0% -2.9%
3 9.9% 14.2% 4.3%
4 4.4% 10.7% 6.3%
5 1.8% 3.6% 1.7%
6 1.3% 3.1% 1.8%
7 0.6% 1.8% 1.2%
8 0.3% 1.8% 1.5%
9 0.2% 0.4% 0.3%
10 0.1% 1.8% 1.7%
11 0.1% 1.3% 1.2%
12 0.1% 0.9% 0.8%


It's very similar, but we do see the top-projected player finish as the MVP 40.4% of the time. That's still about 17 percentage points lower than how the public rosters that player. So, again, it's most likely that the MVP is the top-projected player of any other rank, yet the field (i.e. any player not projected for the most FanDuel points) finishes as the MVP in 59.6% of the lineups.

And let's do one more: let's rank the players by in-game popularity (e.g. the player on the highest number of rosters is ranked first this way).

In-Game
Popularity Rank
Rostered
as MVP
Actual Optimal
MVP Rate
Leverage
1 60.1% 33.8% -26.3%
2 20.4% 26.2% 5.9%
3 8.9% 11.6% 2.7%
4 4.0% 6.7% 2.7%
5 2.5% 8.4% 5.9%
6 1.3% 3.6% 2.3%
7 0.9% 1.8% 0.9%
8 0.5% 1.8% 1.2%
9 0.4% 0.9% 0.5%
10 0.3% 2.2% 2.0%
11 0.2% 1.3% 1.1%
12 0.1% 0.4% 0.3%


Oh, yeah, that's the stuff. The chalkiest pick is on around 60% as the MVP pick but is actually the MVP in just over half of that number of the optimal lineups. I don't need to hammer this home more. Placing the stud in another position other than MVP is very justifiable.

What About the Overall Chalk?

I wanted to do one last look here at the most popular plays. There were 628 players who actually played, were inside the top-five in salary for the game, and were on at least half of the rosters on that slate. You know: the really chalky picks who weren't just value plays.

Here's how frequently they ranked first through fifth (and sixth or worse) in in-game FanDuel points.

Rostered on at Least
50% of Lineups
& Top-5 in Salary
Frequency
1st in FanDuel Points26.4%
2nd in FanDuel Points20.9%
3rd in FanDuel Points14.0%
4th in FanDuel Points11.5%
5th in FanDuel Points7.0%
6th in FanDuel Points or Worse20.2%


There's a 47.3% chance that these really chalky plays finish first or second on the slate in FanDuel points, but as we've seen throughout, there's a lot of leverage available in just not playing the most obvious lineup.

You don't need to be fading super stars, but be different elsewhere if you roster them at MVP, or just play them at STAR or PRO, and you can hit it big when the game breaks that way.

As always in fantasy sports, we don't know as much as we think we know, so we should act accordingly.

Summary and Important Notes

- Only 21.8% of the optimal lineups used all $60,000, but 51.1% of the lineups used at least $59,000.
- Just 11.6% of lineups spent $55,000 or less in total salary.
- A 3-2 stack existed in 80.0% of the optimal lineups, and 46.7% of the optimal lineups featured 3 players from the winning team and 2 from the losing team.
- On average, the optimal MVP is rostered on 66.7% of lineups at any position but as the actual MVP just 34.1% of the time.
- You still want to target players projected for minutes and FanDuel points; differentiate by shuffling players around in the multiplier slots rather than playing too many bench players.
- Whether we rank players by in-game salary rank, pre-game projections, or overall draft percentage, the top play of the slate is generally rostered as the MVP more frequently than he winds up in the optimal MVP slot.
- In only 2 of the 225 optimal lineups (0.9%) did the in-game FanDuel point leader not make the optimal.
- In only 1 of the 225 optimal lineups (0.4%) was the MVP rostered on under 10% of overall lineups at any position.

Related News

FanDuel PGA: Analyzing Trends in Optimal Lineups

Brandon Gdula  --  Dec 14th, 2020

FanDuel Soccer Strategy: Does Stacking Work in EPL DFS?

Austan Kas  --  Dec 14th, 2020

How to Dominate Single-Game Daily Fantasy Football With Perfect Lineup Analysis

Brandon Gdula  --  Dec 14th, 2020