DraftKings Week 5 Recap

One thing we keep seeing is that the $1MM winners typically play more than 1 lineup at a time, with Week 2 being the only exception. Lineup optimizer’s that help you generate the most efficient lineups are the way to go. If you’re code savy here’s our free python lineup optimizer. It’s made to work with Sports Data Direct data but you can easily plug in your own salary and projections info.

Anyways, let’s look at what happened last Sunday night.

GrindersCal Wins Big

GrindersCal won $1MM with a little used Brian Hoyer-George Kittle stack.

That stack was only used in 2 of his 30 lineups! Is this guy a 49er’s fan because that was a great pick.

In hindsight, 49er’s-Colts was a great game for the Niner’s offense to break out under new coach and offensive guru Kyle Shanahan. The 49er’s previous 4 games were against good defenses that can stop TE’s (CAR, @SEA, LAR, @ARZ).

Below is a summary of all of his lineups.

count 30.000000
mean 133.934000
std 31.121901
min 74.140000
25% 115.265000
50% 131.150000
75% 154.365000
max 217.420000

Wisely, he took bigger risks than pro’s like CONDIA who can afford to set their max player exposure to ~25% because they play 150 lineups each week. GrindersCal. Here’s his player utilization that drew heavily on the Packer Offense.

Percentage Player
53 Dez Bryant
53 Aaron Rodgers
47 Aaron Jones
43 Austin Seferian-Jenkins
40 DeVante Parker
40 Randall Cobb
33 Melvin Gordon
33 Martellus Bennett
30 Antonio Brown
30 Torrey Smith
23 Bills
23 Le’Veon Bell
23 Carlos Hyde
20 Dolphins
20 Jordy Nelson
20 Jaron Brown
17 Brian Hoyer
17 Jay Ajayi
17 Davante Adams
17 Cardinals

On quick glance, 53% and 47% add to 100% so you might think he played either Aaron Rodgers or Aaron Jones but not both. It turns out he played them together in 10% of his lineups which is surprising because Jones is not a pass catching RB (in two games he has 1 catch for 9 yards) and QB-RB’s typically exhibit negative correlation.

Normally I don’t show triple stacks but in GrindersCal’s it is relevant.

Percentage Player
20 Austin Seferian-Jenkins, Dez Bryant, Randall Cobb
20 Aaron Rodgers, Dez Bryant, Randall Cobb

That Rodgers-Cobb-Bryant stack is a big bet on the Packers Cowboys game being high scoring and if you realize it ahead of time you can actually hedge this (scroll down to learn more about hedging DFS plays).

Double Stacks

Here’s GrindersCal’s biggest double stacks.

Percentage Player
37 Aaron Rodgers, Dez Bryant
30 Aaron Jones, Austin Seferian-Jenkins
30 Austin Seferian-Jenkins, Dez Bryant
30 Aaron Rodgers, Martellus Bennett
30 Dez Bryant, Randall Cobb
27 Aaron Jones, Antonio Brown
27 Aaron Rodgers, Randall Cobb
23 Aaron Jones, Dez Bryant
23 DeVante Parker, Dez Bryant
23 Aaron Jones, DeVante Parker
23 Austin Seferian-Jenkins, Randall Cobb
20 Austin Seferian-Jenkins, Melvin Gordon
20 Dez Bryant, Melvin Gordon
20 Austin Seferian-Jenkins, Torrey Smith
20 Austin Seferian-Jenkins, DeVante Parker

How He Probably Did It

As far as how this guy made his lineups, I’m guessing he used a lineup optimizer because he had 7 different defenses and 66 players in his player pool but didn’t use any of the constraints like force this QB-WR pair. Or if he did, he generated half of his lineups with a Packers heavy offense and the other half differently.

His best play was Brian Hoyer-George Kittle. He must have removed lineups generated based on some metric like predicted points because, at the low salaries Hoyer and Kittle had, you’d expect them to be far more prevalent across his lineups. Perhaps he just got a lucky draw from the generator and doesn’t even know who George Kittle is.

Best Hotel Deals - Save up to 75% Off* on Hotels

At Least 70,812 Lineups Better Than GrindersCal

Using our lineup optimizer, we found room for improvement over the actual played lineups in Week 5’s Millionaire Maker. The best possible lineup was 271.34 points.

player_name position score salary team_name lineup_id
6 Deshaun Watson QB 35.54 6200.0 HOU 0
0 Aaron Jones RB 23.40 5100.0 GB 0
4 Leonard Fournette RB 34.40 7000.0 JAC 0
1 T.Y. Hilton WR 27.70 6000.0 IND 0
5 A.J. Green WR 33.90 8200.0 CIN 0
7 Davante Adams WR 25.60 5800.0 GB 0
3 Ed Dickson TE 25.50 2900.0 CAR 0
8 Melvin Gordon FLEX 37.30 6000.0 LAC 0
2 Jacksonville Jaguars DST 28.00 2800.0 JAC 0

Honestly, I’m surprised no one played it. None of these players are completely random flukes, Aaron Jones and Ed Dickson stand out, but a reasonable player could have had all these players in their high upside player pool. On the prediction websites I use, both players were projected around 8 points which is pretty good considering their salaries.

The player pool for these winning lineups is below

Name Utilization
8 Melvin Gordon 0.905
4 Leonard Fournette 0.585
3 Ed Dickson 0.579
5 A.J. Green 0.537
2 Jacksonville Jaguars 0.534
12 George Kittle 0.421
6 Deshaun Watson 0.387
1 T.Y. Hilton 0.353
7 Davante Adams 0.316
21 Antonio Brown 0.309
16 DeAndre Hopkins 0.297
0 Aaron Jones 0.284
19 Miami Dolphins 0.274
14 Alex Smith 0.256
10 Marquise Goodwin 0.250
9 Will Fuller 0.247
11 Nelson Agholor 0.243
13 Javorius Allen 0.207
24 Brian Hoyer 0.192
18 Torrey Smith 0.187
17 Marlon Mack 0.185
15 Duke Johnson 0.181
20 Charcandrick West 0.166
22 Orleans Darkwa 0.159
23 Tavon Austin 0.141
25 Robert Woods 0.136
29 Green Bay Packers 0.130
26 Brandon Tate 0.120
27 Kyle Juszczyk 0.104
30 Kevin Hogan 0.091
28 Shane Vereen 0.087
31 Josh McCown 0.064
32 Dallas Cowboys 0.062
33 Joe Webb 0.010

Keep an eye on these lists because you should always chase the best possible score rather than over-analyze what the winning lineup happened to be. Especially this week when the winner is 55 points below the top score!

CONDIA: Another Week as Usual

True to form, CONDIA’s QB-WR stacks topped out around 7% and a player pool over 100.

  • Dak Prescott-Dez Bryant 7%
  • Aaron Rodgers-Randall Cobb 7%

He didn’t win big but won a handy $2,925 on his 150 lineups ($3,000 entry, unlike how DraftKings presents it, we believe winning is how much money you received after your bet… On DK this would show as winning “$5,925”.)

CONDIA’s point totals show a healthy “positive skew” versus a pure normal position. Basically when some of his player scored well, other players also scored well due to positive correlation.

CONDIA's Point Totals

Sidenote on Stacking

Just because the best possible lineup doesn’t use stacks, stacks are a great way to increase your probability of a winning lineup even though stacks don’t guarantee you the best score. In today’s contests, winning lineups are significantly below the best possible lineups, so stacking and positive correlation should continue to work. Otherwise you’re essentially gambling.

CONDIA’s Player Pool

CONDIA used more exposure on Le’Veon Bell than we’ve seen in past weeks. Typically he caps out at ~25% but this time he felt comfortable using 32% on Bell. Still, this is in-line with him using exposure on consistent plays.

Percentage Player
32 Le’Veon Bell
27 Eagles
26 Melvin Gordon
22 Jordy Nelson
22 Aaron Rodgers
21 Dak Prescott
21 Dez Bryant
20 Todd Gurley II
20 Jarvis Landry
19 T.Y. Hilton

Interestingly, he has two stacks that have matched exposure within the stack (Rodgers-Nelson and Prescott-Bryant). I did not see this weeks 1-4 but it’s worth keeping an eye on.

Still, he didn’t enforce any hard rules like Saahil Sud would do, keeping his stacks to 7% as we showed earlier.

RKennedy23 Does Not Repeat

count 27.000000
mean 118.958518
std 15.922730
min 83.340000
25% 108.130000
50% 119.040000
75% 131.050000
max 144.840000

Hot off his $1MM win last week, RKennedy23 gave Week 5 another go. Thankfully he didn’t double down and played a similar number of lineups. He lost $500 on 27 lineups while using a combo of Ezekiel Elliott and the Ravens defense at 85% utilization each.

We’ll stop tracking this player unless he has another great finish.

Saahil Sud Remains the Ballsy Pro

Saahil Sud is pissed of losing. This week he went all out on Aaron Rodgers and Jordy Nelson, placing them in 100% of his 150 lineups!

Programming Category (English)468x60

While this didn’t work in Week 5, you can imagine how big a payout it will be if the blip around 150 points in the chart below moves out 60 points.

Saahil Sud's Point Totals

It’s about time we saw a lock by a pro but too bad it didn’t pan out. Still, Saahil only needs about 1 win and he will be swimming in money. His player pool had only 44 players compared to CONDIA’s 119. Saahil is truly a fat tony.

He lost on Jordy, Engram, and Elliott all having bad days.

Percentage Player
100 Aaron Rodgers
100 Jordy Nelson
92 Dez Bryant
48 Ezekiel Elliott
45 Marvin Jones Jr.
40 Evan Engram
37 Torrey Smith
33 Jarvis Landry
31 Austin Seferian-Jenkins
28 Ravens

Better luck next time Saahil. I’m rooting for you.

The Herd’s Picks

Draft Kings won $878,145 last week, down slightly from their usual $1MM.

Here’s how everyone did with their lineups.

count 241414
mean 118.901162
std 22.728308
min 0.000000
25% 103.140000
50% 118.120000
75% 133.940000
max 217.420000

135 points was just about the threshold to make money.

Herd Player Pool

Austin Sefarian-Jenkins received a surprising amount of utilization at 28%. And Jordy Nelson/Randall Cobb over Davante Adams (which wound up being the wrong play)

Percentage Player
33 Le’Veon Bell
32 T.Y. Hilton
30 Dez Bryant
28 Austin Seferian-Jenkins
27 Todd Gurley II
23 Jordy Nelson
18 Duke Johnson Jr.
18 Aaron Rodgers
16 Ezekiel Elliott
16 Jaron Brown
16 Bills
15 Melvin Gordon
15 Jarvis Landry
15 DeVante Parker
14 Carlos Hyde
14 Randall Cobb
13 Pierre Garcon
13 Larry Fitzgerald
13 Bilal Powell
12 Antonio Brown

People also thought Jaron Brown would repeat his great Week 4 performance but he didn’t live up to the hype.

QB Stacks

  • Aaron Rodgers-Jordy Nelson 8%
  • Jacoby Brissett-T.Y. Hilton 6%
  • Dak Prescott-Dez Bryant 6%

I’m seeing the Packers passing offense is much preferred to the Cowboys passing offense in spite of the strong performance a Brice Butler or Cole Beasley lineup would be.

Some of the most herd-like lineups involved Sefarian-Jenkins and a normal play like Le’Veon Bell

Percentage Players
12 Austin Seferian-Jenkins, Le’Veon Bell
11 Le’Veon Bell, T.Y. Hilton
11 Austin Seferian-Jenkins, Dez Bryant
10 Austin Seferian-Jenkins, T.Y. Hilton
10 Dez Bryant, T.Y. Hilton
9 Dez Bryant, Le’Veon Bell
8 Aaron Rodgers, Jordy Nelson
8 T.Y. Hilton, Todd Gurley II
8 Dez Bryant, Todd Gurley II
8 Austin Seferian-Jenkins, Todd Gurley II

So avoid these players given “equivalent” players that are more under the radar. Ed Dickson will probably be played heavily in Week 6, so it’s worth considering playing a Delanie Walker or Kyle Rudolph type TE. Still, if the price is right, sometimes you can’t avoid the “herd pick”.

Sports Betting Strategy: Hedging Your DFS Plays

Last weekend, we saw GrindersCal win big by playing a heavy dose of the Packers and the Dallas Cowboys. One way to reduce your downside risk and keep yourself in the running for the $1MM prize is to hedge your bets.

I won’t go into too much detail but the basic premise is this. Say you are betting on players from two teams facing each other — Dez Bryant, Aaron Rodgers. If you take the under line, betting that the total score of the Packers Cowboys game will be less than 52 on a sports book, you can still win if your DFS plays had a bad day (which would typically happen if the game goes under). If the Over/Under goes over, you’re likelihood of your players getting a lot of points increases significantly.

Another opportunity exists for player prop bets. For example, if you see an Over/Under line on Ezekiel Elliott getting 100 yards, and choose him in a lot of your lineups, you can take the under and benefit from the Draft Kings +3 points for a player getting over 100 yards.

Be Like Saahil If You Don’t Max Your Lineups

The cold hard truth is that it’s very hard to win the Millionaire Maker if you don’t take some chances. CONDIA is able to enter $3K a week and profits from a repetitive strategy but you can’t do that unless you play a ton of lineups. If you can only play 20 lineups, take some chances. Think of events that will change many players fantasy point totals. Will the Packers-Cowboys game be a shootout? Who will get the points and who won’t?

Programming Category (English)468x60

If you try to do a CONDIA without putting down lot’s of money, you could have a rough a time.

Update on Contest Recap Tool

Thanks everyone for voting last week. I will start looking into creating a contest recap tool but it probably won’t be ready until next season.

I’ll keep you updated with what I’m thinking of including as time goes on.

Also, subscribe to our mailing list so you are notified when we publish first. I time posts on reddit and sometimes publish these recaps a few days in advance before posting there.

Matteo Hoch

Matteo Hoch is the founder of Sports Data Direct and maintains a personal blog called Ergo Sum where he writes about data visualization and coding. You can generally find him in the DraftKings Sunday $3 early only contests under username mhoch2

Leave a Reply

Your email address will not be published. Required fields are marked *