DraftKings Week 12 Recap

This week was a bit special. The feature contest was the NFL 75th Millionaire Special on DraftKings. $75 entry and $1MM to first place.

CONDIA hasn’t played since November 5th and the pro’s of old are being washed away. I still believe DFS is a game of skill but it’s something to investigate during the off season.

This week’s top pro that we track was scout326 with 210.48. The overall winner was “ending” with 233.68 points and it was 150 to cash but the best possible lineup would have been a whopping 282.18 points thanks to a dominant performance by Julio Jones.

Let’s take a look at ending’s lineups and see what we can find out about his strategy

Ending wins with 233.68

Ending 233.68 Winning Lineup

Ending entered 150 lineups even with the $75 entry fee this week. Ending had some insane variance, a great thing for GPP’s. His lowest scoring lineup was 75.96 points and his best was 233.68! He won $998,075.

Points
count 150.000000
mean 133.588800
std 29.029659
min 75.960000
25% 111.830000
50% 131.480000
75% 151.735000
max 233.680000

Look at this beautiful one-sided fat tail.

Ending's Lineup Point Totals

Ain’t that some magic. If I we’re to write a book on GPP lineups, and maybe I will, this would be the perfect picture for illustrating how your lineups should look when engineered to exploit variance. Like seriously, it’s perfect.

We're not Worthy

Now let’s get to the nitty gritty, how he actually created his lineups.

Ending Created His Lineups Like CONDIA

We looked at his lineups and found that Ending plays most like CONDIA. For those of you new to our recaps, that means optimizer driven lineup generation with exposure limits. Generally these players use a player pool over 100, max single player exposure 35%, and stacks <20%. Single player utilization percentages generally favor RB-WR-TE with lots of touches. No big bets, just lots of little ones.

Percentage Player
39 Joe Mixon
35 Carlos Hyde
34 Travis Kelce
31 Rob Gronkowski
29 Todd Gurley II
27 Julio Jones
27 Michael Thomas
25 DeMarco Murray
25 Jaguars
22 Cooper Kupp
22 Mike Evans
19 Michael Crabtree
19 Emmanuel Sanders
15 Seahawks
15 Jets
15 Frank Gore
14 Carson Wentz

In my opinion, this strategy is best if you can max in your contests, otherwise you have to take a bit more risk like TommyG1979’s strategy (although he maxes most of the time :P).

Here’s ending’s 2 pairs.

Percentage Player
791 20 Carlos Hyde, Travis Kelce
761 18 Joe Mixon, Michael Crabtree
705 15 Carlos Hyde, Joe Mixon
683 14 Joe Mixon, Julio Jones
695 14 Joe Mixon, Travis Kelce
601 11 Joe Mixon, Mike Evans
609 11 Julio Jones, Michael Thomas
610 11 Michael Thomas, Rob Gronkowski
568 10 Jaguars , Joe Mixon
569 10 DeMarco Murray, Jaguars
574 10 Joe Mixon, Michael Thomas
505 9 Joe Mixon, Rob Gronkowski
509 9 Carlos Hyde, Cooper Kupp
514 9 Mike Evans, Travis Kelce
517 9 DeMarco Murray, Rob Gronkowski
542 9 Rob Gronkowski, Todd Gurley II
545 9 Eric Decker, Todd Gurley II
550 9 Emmanuel Sanders, Todd Gurley II
434 8 Carlos Hyde, Rob Gronkowski
437 8 Carlos Hyde, Julio Jones
439 8 Cooper Kupp, Joe Mixon
441 8 Cooper Kupp, Rob Gronkowski
442 8 Cooper Kupp, DeMarco Murray
443 8 Cooper Kupp, Travis Kelce
448 8 Carlos Hyde, Mike Evans
451 8 Julio Jones, Matt Ryan
453 8 Carlos Hyde, DeMarco Murray
455 8 DeMarco Murray, Michael Thomas
467 8 Carlos Hyde, Michael Thomas
481 8 Jets , Rob Gronkowski

His first QB stack wasn’t until Matt Ryan-Julio Jones at 8%. To be honest, this ending guy looks like a CONDIA clone on first take. I tried looking him up on Rotogrinders and he’s not tracked. We’ll keep an eye on him from now on.

Congrats on the win.

For an excellent book on things that benefit from variance, like GPP lineups, I recommend you read Antifragile: Things That Gain From Disorder by Nicolas Nassim Taleb. The book is endorsed by DFS Pro Jonathan Bales and an eye-opening read.

CheapFlightNow's Exclusive Roundtrip Flights Sale. Book now and get up to $15 Off* with coupon code CFNAIR15.

Pro Meta

This week the pro’s accounted for 6% of the player pool. As a group, they paid $276,600 and lost $101,900.

payout lineups_entered payout/num_entries
screen_name
reztes757 12625.0 13 971.153846
scout326 7325.0 150 48.833333
teejayortj 3100.0 75 41.333333
shocae 2050.0 6 341.666667
bric75 1200.0 150 8.000000
moklovin -7700.0 150 -51.333333
chipotleaddict -7875.0 150 -52.500000
kingfi -8100.0 150 -54.000000
petrgibbons -8250.0 150 -55.000000
bdholla89 -8725.0 150 -58.166667

scout326 did the best out of the pro’s I’ve heard of, and ChipotleAddict did the worst of pro players I’ve heard of.

TGII received the most usage in spite of his high $8,800 salary.

Percentage Player
30 Todd Gurley II
29 Carlos Hyde
25 Julio Jones
25 Cooper Kupp
25 Michael Thomas
25 Kareem Hunt
24 Rob Gronkowski
23 Joe Mixon
22 Tevin Coleman
19 Jack Doyle
17 Travis Kelce
16 Sammy Watkins
16 Brandin Cooks
15 Delanie Walker
15 Corey Davis
14 Jaguars
14 J.D. McKissic
14 Doug Baldwin
13 Tom Brady
13 Russell Wilson
12 Eagles
12 Demaryius Thomas
12 Devontae Booker
11 Larry Fitzgerald
11 Bengals
10 Mike Evans
10 Matt Ryan
10 Dion Lewis
10 Panthers
10 Falcons

I’m surprised how little they used Leonard Fournette, smart play by them. Their top defenses we’re Jaguars and Eagles. Top QB Tom Brady. Picking Joe Mixon was big and pro’s did it.

As for stacks, the pro’s looked for a bounce back week by Rob Gronkowski, stacking him with Tom Brady in 9% of lineups, and Cooks in 8%.

Percentage Player
9 Cooper Kupp, Michael Thomas
9 Rob Gronkowski, Tom Brady
8 Julio Jones, Matt Ryan
8 Brandin Cooks, Tom Brady
8 Doug Baldwin, Russell Wilson
7 Cooper Kupp, Todd Gurley II
7 Carlos Hyde, Cooper Kupp
7 Michael Thomas, Sammy Watkins
7 Michael Thomas, Todd Gurley II
7 Carlos Hyde, Michael Thomas

But pro love for Matt Ryan – Julio Jones was right in-line with how the winner “ending” played it.

Scout326 Won $7,325

This is the first time we’ve looked at Scout326. On first take, he looks similar to SaahilSud.

Percentage Player
62 J.D. McKissic
59 Julio Jones
51 Michael Thomas
44 Cooper Kupp
39 Joe Mixon
34 Bears
30 Todd Gurley II
30 T.Y. Hilton
28 Brandin Cooks
27 Rob Gronkowski
27 Panthers
27 Jack Doyle
20 Sammy Watkins
19 Travis Kelce
19 Kareem Hunt
18 Corey Davis
18 DeMarco Murray
15 Carlos Hyde
15 Russell Wilson
15 Doug Baldwin
15 Jimmy Graham
15 Marcus Mariota

He managed to win despite playing the Bears D in 34% of his lineups. Julio Jones did some great things last Sunday..

Scout326’s stacks look optimizer driven rather than deliberate like TommyG. But he places a bit more weight on players than SaahilSud or CONDIA. Going as far as having some 10% 4-player pairs (most pro’s don’t have anything over 5% on their 3+ player groups).

Percentage Player
36 J.D. McKissic, Julio Jones
34 J.D. McKissic, Michael Thomas
32 Cooper Kupp, Michael Thomas
26 Cooper Kupp, J.D. McKissic
25 Joe Mixon, Julio Jones
25 Julio Jones, Michael Thomas
22 Cooper Kupp, Julio Jones
22 Bears , Julio Jones
21 Joe Mixon, Michael Thomas
20 Brandin Cooks, J.D. McKissic
20 Michael Thomas, Todd Gurley II
20 Bears , J.D. McKissic
19 J.D. McKissic, Todd Gurley II
18 Julio Jones, Panthers
17 Cooper Kupp, Joe Mixon
17 J.D. McKissic, Joe Mixon
17 J.D. McKissic, Rob Gronkowski
17 Bears , Michael Thomas
16 Julio Jones, Rob Gronkowski
16 J.D. McKissic, Panthers
16 J.D. McKissic, Jack Doyle
15 J.D. McKissic, Travis Kelce

His player pool was 81 players which suggests he manually excluded some players or set exposure targets rather than ceilings.

Al_Smizzle

Entered 50 lineups, losing $2,000. He missed the mark with Corey Davis and the Bengals D.

Percentage Player
47 Bengals
47 Corey Davis
43 Jack Doyle
41 Carlos Hyde
37 Julio Jones
35 Brandin Cooks
33 Eagles
33 Sammy Watkins
33 Devontae Booker
31 Jimmy Graham
31 Kareem Hunt
25 Tevin Coleman
25 Cooper Kupp
23 Todd Gurley II
21 Russell Wilson

But he did well with Jack Doyle. Again Al_Smizzle focused on his player pool. He got down to 43 players throughout his 50 lineups. Interestingly, his player pool size was about the same last week when he played 150 lineups. So, this independence of player pool size and lineups suggests Al_Smizzle is definitely a player pool driven DFS Pro.

Here’s his biggest pairs

Percentage Player
35 Bengals , Corey Davis
27 Corey Davis, Jack Doyle
21 Bengals , Jack Doyle
21 Carlos Hyde, Corey Davis
21 Jack Doyle, Julio Jones
21 Brandin Cooks, Carlos Hyde
21 Bengals , Jimmy Graham
21 Bengals , Tevin Coleman
21 Corey Davis, Marcus Mariota
19 Carlos Hyde, Eagles
19 Brandin Cooks, Eagles
19 Jack Doyle, Sammy Watkins
19 Corey Davis, Tevin Coleman
19 Bengals , Devontae Booker
19 Eagles , Kareem Hunt

All and all, Al_Smizzle missed the mark but we learned something concrete about his GPP play style.

TommyG1979  Entered 50 and Lost Big

TommyG lost $3,200. Will be interesting to see if his bankroll is trending downwards because he didn’t max last week either. His point totals had the right shape but needed to be shifted a bit to the right.

TommyG Point Totals

His single player utilization tells most of the story

Percentage Player
69 O.J. Howard
53 Jaguars
51 Todd Gurley II
51 J.D. McKissic
45 Kareem Hunt
39 Corey Coleman
35 Marquise Goodwin
25 T.Y. Hilton
23 Alshon Jeffery
23 Brandin Cooks
21 Russell Wilson
19 Jared Cook
17 Kenny Stills
17 Tom Brady
15 Christian McCaffrey

O.J. Howard and Todd Gurley were big misses. And due to his heavy stacking approach this affected a large percentage of his lineups.

For reference, Tommy G played 79 players across his 50 lineups. Nearly double Al_Smizzle. And Tommy G used big stacks

Rest of Pro’s

Bales had his best DraftKings week all season. Just barely losing money at -$50. He still doesn’t come close to maxing, only entering 13 lineups this week.

Not to mention CSURAM88 played only 16 lineups losing a whopping $1,075. Bales and CSURAM88 mirror each other pretty closely as far as entries go — their lives are probably correlated due to their friendship but at least their lineups are different. I wonder if he uses Fantasy Labs… or is focused on other things probably the latter.

ChipotleAddict did terrible in Week 12, losing $7,875 on 150 lineups. His player pool wasn’t all that bad. He had Carlos Hyde, Gronk, and Joe Mixon is his top 3. But Tom Brady – Rob Gronkowski didn’t do well enough to beat the rake and he had them in 25% of his lineups.

Programming Category (English)468x60

Pro Absences

CONDIA looks to be taking a break while SaahilSud has moved his action to FanDuel’s Sunday Million contest.

We can only speculate why Saahil left DraftKings but I’d like to think it has something to do with how FanDuel is harder to get information about FanDuel’s contests for an amateur — not that Saahil does this but be advised that DraftKings and FanDuel may engage in supplying “generally available” information to sponsored players and sites in an easier to program feed — here’s an old link on Rotogrinders saying they had a custom feed 4 years ago.

The Herd Meta

The herd played markedly similar to the Rotogrinder’s Top 50.

index Points_herd Points_pro
count 56833.000000 3688.000000
mean 128.801143 129.019767
std 27.252495 27.612265
min 41.740000 55.340000
25% 108.980000 109.065000
50% 127.680000 127.240000
75% 147.560000 147.590000
max 233.680000 223.460000

Like seriously, as a group they look to be formed from the same generating distribution, aka herd/pro performance identical in Week 12. The players they picked we’re different but had similar results.

lineup_hash Percentage Pro Percentage Herd pro_favoritism
T.Y. Hilton 8 20 -12
Kenny Stills 7 19 -12
Eagles 12 22 -10
Tyler Kroft 5 14 -9
Russell Wilson 13 20 -7
Alshon Jeffery 6 13 -7
Tevin Coleman 22 29 -7
Mohamed Sanu 7 12 -5
Dion Lewis 10 15 -5
Devin Funchess 5 9 -4
Falcons 10 14 -4
Marcus Mariota 6 9 -3
DeMarco Murray 8 10 -2
Brandin Cooks 16 18 -2
Delanie Walker 15 17 -2
Corey Davis 15 17 -2
Jaguars 14 16 -2
Doug Baldwin 14 16 -2
Cam Newton 5 7 -2
Christian McCaffrey 6 8 -2
Jacoby Brissett 5 7 -2
Duke Johnson Jr. 9 10 -1
Devontae Booker 12 13 -1
J.D. McKissic 14 15 -1
Jimmy Graham 6 6 0
Sammy Watkins 16 16 0
A.J. Green 6 6 0
Marqise Lee 5 5 0
Kareem Hunt 25 25 0
Julio Jones 25 24 1
Matt Ryan 10 9 1
Jack Doyle 19 17 2
Cooper Kupp 25 23 2
Tom Brady 13 11 2
Patriots 9 7 2
Jared Goff 7 5 2
Larry Fitzgerald 11 8 3
Panthers 10 5 5
Demaryius Thomas 12 7 5
Bengals 11 6 5
Rob Gronkowski 24 16 8
Todd Gurley II 30 21 9
Travis Kelce 17 7 10
Joe Mixon 23 11 12
Carlos Hyde 29 16 13
Michael Thomas 25 9 16
Mike Evans 10 <5% <5%
Rex Burkhead 8 <5% <5%
Emmanuel Sanders 8 <5% <5%
Frank Gore 8 <5% <5%
Tyreek Hill 7 <5% <5%
Austin Hooper 6 <5% <5%
Drew Brees 6 <5% <5%
Alex Smith 6 <5% <5%
Michael Crabtree 6 <5% <5%
DeSean Jackson 5 <5% <5%
Marquise Goodwin 5 <5% <5%
Jets 5 <5% <5%
Alvin Kamara <5% 17 <5%
Corey Coleman <5% 11 <5%
Mark Ingram <5% 11 <5%
Dede Westbrook <5% 8 <5%
Zay Jones <5% 7 <5%
Leonard Fournette <5% 7 <5%
Jared Cook <5% 6 <5%
Jarvis Landry <5% 5 <5%

Tevin Coleman was the top herd pick, beating the pro’s 30% utilization to Todd Gurley. The herd also did well playing Alvin Kamara way more than pro’s.

The herd faded the Patriots a little harder than the Pro’s. Russell Wilson-Doug Baldwin was their top stack at 9%

Percentage Player
9 Doug Baldwin, Russell Wilson
8 Cooper Kupp, Tevin Coleman
6 Julio Jones, Matt Ryan
6 Eagles , Tevin Coleman
6 Russell Wilson, Tevin Coleman
6 Corey Davis, Tevin Coleman
6 T.Y. Hilton, Tevin Coleman
5 Julio Jones, Tevin Coleman
5 Julio Jones, Kareem Hunt
5 Cooper Kupp, Kareem Hunt
5 Delanie Walker, Tevin Coleman
5 Jack Doyle, Tevin Coleman
5 Eagles , Julio Jones
5 Cooper Kupp, Eagles
5 Eagles , Kareem Hunt
5 Sammy Watkins, Tevin Coleman
5 Kenny Stills, Tevin Coleman
5 Eagles , Kenny Stills
5 Brandin Cooks, Tevin Coleman
5 J.D. McKissic, Julio Jones
5 Rob Gronkowski, Tom Brady
5 Brandin Cooks, Tom Brady

Not much Josh McCown love by either parties. But as you’ll see in the Best Lineup Possible section, Josh McCown was a beast.

Best Lineup Possible: 282.18

There were at least 120,856 lineups better than ending’s 233.68. The best lineup possible used Josh McCown and you guessed it Julio Jones

player_name position score salary team_name
Josh McCown QB 29.08 5100.0 NYJ
Alvin Kamara RB 39.80 7900.0 NO
Joe Mixon RB 28.50 4400.0 CIN
Jermaine Kearse WR 26.50 4000.0 NYJ
Cooper Kupp WR 22.60 5000.0 LAR
Robby Anderson WR 35.60 5700.0 NYJ
Zach Ertz TE 29.30 6600.0 PHI
Julio Jones FLEX 53.80 7700.0 ATL
Carolina Panthers DST 17.00 3300.0 CAR
Totals 282.18 49700.0

The player pool for these 120,865 better lineups with utilization is shown below

Name Percentage
Julio Jones 1.000
Alvin Kamara 0.870
Robby Anderson 0.712
Josh McCown 0.497
Jermaine Kearse 0.397
Joe Mixon 0.396
Zach Ertz 0.395
Blake Bortles 0.281
Carolina Panthers 0.264
Cooper Kupp 0.261
New England Patriots 0.261
DeShone Kizer 0.208
Mohamed Sanu 0.206
Rob Gronkowski 0.197
Marshawn Lynch 0.194
Devin Funchess 0.188
Rex Burkhead 0.174
Brandin Cooks 0.160
Tevin Coleman 0.160
Tennessee Titans 0.138
Ricky Seals-Jones 0.135
Frank Gore 0.132
Arizona Cardinals 0.130
Cody Latimer 0.124
Josh Reynolds 0.122
Buffalo Bills 0.108
Jack Doyle 0.106
Dion Lewis 0.105
Albert Wilson 0.102
Bennie Fowler 0.102
Jaron Brown 0.100
Miami Dolphins 0.099
Delanie Walker 0.098
Peyton Barber 0.091
Isaiah Crowell 0.074
Jalen Richard 0.072
Nick Vannett 0.063
Travaris Cadet 0.051
David Njoku 0.046
O.J. Howard 0.043
Charles Sims 0.035
D.J. Foster 0.033
Todd Gurley 0.029
Giovani Bernard 0.026
Joe Webb 0.007
Cody Kessler 0.007

The biggest surprise was that Josh McCown and Jets offense were the best stacks despite the insane performance by Julio Jones. Matt Ryan didn’t even make our list because he only scored 19.78 points on DraftKings.

Josh McCown is a BAMF if you don’t know already. After being cut in 2011 by the 49ers, he was a high school gym teacher/football coach. His high school team lost in the playoffs and 2 weeks later he was on the Chicago Bears doing this to the Packers at Lambeau.

Josh McCown Dunking

Not to mention he had a sweet spin move into the end zone right before this.

Announcements

We won’t do much of week by week advice

We said to fade TGII in our article last week and we were right but missed on a few other predictions. When pressed, we said Leonard Fournette had higher upside though and it doesn’t look like that was case last week.

[mini-rant] I don’t want to start making advice columns because I feel that’s a shoddy business. Because the big question I have to answer is, if the advice is so good why don’t I use it to make money. Like why does Hammer DFS offer a “DFS consultant” for $39.95 a month? If they can guarantee profitability, why not use it to make over $39.95/month and not deal with clients?

Programming Category (English)468x60

I think we’re starting to really key in on genuine approaches to GPP’s through these articles that you can use to update your play style. But for me to sell these recaps as the golden ticket to beat the sharks would be naive. If I had the golden ticket I would be a pro. Down the line, we may offer a book, tutorial, or something else but under no circumstance will I ever bill something as the golden ticket.

Supreme Court to Decide on Legality of Sports Betting

Read this Chicago Tribune article for more

NBA Recap Post Coming Just in Time for Christmas

We started these posts with every week being an investigative journey. I think we’re finally at the point where we know what data we need to look at each week so we’ll be taking some time to improve our internal processes to churn out quality content at a higher rate than we do for the NFL.

What NBA Contest should we focus on? I was thinking we look at on the NBA $350K Excellent 8’s on a day when most NBA are playing. Are there any bigger contests that happen every week and are accessible to lots of players, like the $1MM Maker for NFL?

Are GPP’s Skill?

I’m still surprised at how poorly the Rotogrinders top 50 do. 3 made money last week and 5 this week — including the bias of some one-time big winners. GPP’s are difficult to analyze because most people are all about getting that one win that makes everything worth it.

Still, no one has ever looked at skill vs rake (how much of a cut DraftKings takes). There is definitely a rake percent where DFS is a lottery not skill.

Maybe the Rotogrinder’s Top 50 is a bad ranking system, maybe we need more samples, I don’t know. Something to think about before the 2018 NFL Season.

IT Certification Category (English)468x60

Less Focus on Tracking All Pro’s in Weekly Recaps

With the data suggesting pro’s can be just as hit or miss as the rest of us, and the abundance of pro’s, we’ve limited our individual player analysis this week. Instead we focused on general trends and reporting.

Going forward we’ll do player profiles as their own series unless a player played remarkably in a given week.

Hate this change? Let us know.

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 *