How Sharks Approach Daily Fantasy Sports: Defenses

In this installment I’ll talk about how professional daily fantasy players — think of players like CONDIA, Saahil Sud, and Jonathan Bales — approach defenses on Draft Kings.

I’m only focusing on the GPP’s (large tourneys), specifically the Millionaire Maker’s and these claims are backed up by 3 weeks of evidence. If they are invalidated I will go back and keep this up to date.

Classifying Draft Kings Sharks

I’ve done a few recaps on DFS to date (2017 Week 1, Week 2, Week 4). So far I’ve found that professional DFS players or sharks come in 2 distinct camps: the Fat Tony’s and the Dr John’s of the world — characters from Nassim Taleb.

First, an introduction on these characters using a coin flip.

Coin Flip

From Wikipedia:

A third party asks [Fat Tony and Dr. John] to “assume that a coin is fair, i.e., has an equal probability of coming up heads or tails when flipped. I flip it ninety-nine times and get heads each time. What are the odds of my getting tails on my next throw?”

  • Dr. John says that the odds are not affected by the previous outcomes so the odds must still be 50:50.
  • Fat Tony says that the odds of the coin coming up heads 99 times in a row are so low that the initial assumption that the coin had a 50:50 chance of coming up heads is most likely incorrect. “The coin gotta be loaded. It can’t be a fair game.”

The core of this debate is whether probability should model ever-changing beliefs, Fat Tony, or frequency of regularly occurring events, Dr. John.

In Draft Kings, CSURAM88 and CONDIA make up the Dr. John’s. Every week they follow certain rules. Limit player exposure to ~25%, take a lot of chances on different QB-WR stacks that make up 8% of their lineups. So far CONDIA has missed 2 out of 3 times but when this strategy hit he made millions. The Dr. John’s don’t get overly impressed by “must play” players and play the same style every week.

Saahil Sud and Jonathan Bales take big risks. They are the Fat Tony’s. They don’t mind putting a risky player like Jared Cook in 35+% of their lineups. When/if they hit they can receive 10x+ multiples higher return than Dr. John’s but Fat Tony may go a long time without hitting.

If you’ve read Nassim Taleb, Antifragile and The Black Swan are good places to start, you’ll recognize these characters. Although it’s not a traditional match because these pros all compete in highly nonlinear contests, their approach to the game is reflected by the characters from Nassim Taleb.

Sidenote: here’s Bales mentioning he’s influenced by Taleb

CONDIA’s Approach

In Week 1 CONDIA used 13 out of a possible 30 defenses! Max exposure 21% Bills

lineup_hash dk_salary dk_points points/salary Percentage week_num pair_size dst
6929 Bills 3900.0 9.0 2.307692 21 1 1 Bills
23072 Steelers 3600.0 18.0 5.000000 20 1 1 Steelers
13383 Texans 3800.0 0.0 0.000000 18 1 1 Texans
34262 Colts 2700.0 2.0 0.740741 9 1 1 Colts
16121 Rams 3200.0 28.0 8.750000 7 1 1 Rams
28314 Panthers 3500.0 15.0 4.285714 7 1 1 Panthers
30804 Packers 2500.0 9.0 3.600000 7 1 1 Packers
24342 Eagles 2700.0 22.0 8.148148 3 1 1 Eagles
25598 Bengals 3100.0 4.0 1.290323 3 1 1 Bengals
35536 Cardinals 2800.0 8.0 2.857143 2 1 1 Cardinals
34518 Ravens 2800.0 25.0 8.928571 1 1 1 Ravens
34774 Redskins 3000.0 12.0 4.000000 1 1 1 Redskins
35792 Cowboys 2900.0 12.0 4.137931 1 1 1 Cowboys

Week 2: 16/32 defenses, max exposure 17% Ravens

lineup_hash dk_salary dk_points points/salary Percentage week_num pair_size dst
10570 Ravens 3700.0 17.0 4.594595 17 2 1 Ravens
4628 Raiders 3500.0 9.0 2.571429 13 2 1 Raiders
21256 Seahawks 4000.0 8.0 2.000000 11 2 1 Seahawks
32371 Chiefs 2900.0 11.0 3.793103 11 2 1 Chiefs
28464 Panthers 3600.0 10.0 2.777778 10 2 1 Panthers
13802 Chargers 2800.0 3.0 1.071429 9 2 1 Chargers
17030 Buccaneers 3200.0 19.0 5.937500 9 2 1 Buccaneers
24750 Steelers 3000.0 8.0 2.666667 9 2 1 Steelers
33385 Rams 3000.0 2.0 0.666667 3 2 1 Rams
34652 Redskins 2900.0 7.0 2.413793 3 2 1 Redskins
35163 Titans 3300.0 9.0 2.727273 1 2 1 Titans
35671 Jaguars 3100.0 -1.0 -0.322581 1 2 1 Jaguars
35927 Dolphins 2600.0 2.0 0.769231 1 2 1 Dolphins
36438 Cowboys 2600.0 2.0 0.769231 1 2 1 Cowboys
36694 Broncos 3400.0 13.0 3.823529 1 2 1 Broncos
36950 Cardinals 4100.0 10.0 2.439024 1 2 1 Cardinals

Week 4: 17/32 defense, max exposure 22% Jaguars

lineup_hash dk_salary dk_points points/salary Percentage week_num pair_size dst
9279 Jaguars 3600.0 15.0 4.166667 22 4 1 Jaguars
19399 Cardinals 3200.0 6.0 1.875000 11 4 1 Cardinals
23543 Bengals 3500.0 8.0 2.285714 11 4 1 Bengals
15232 Chargers 2500.0 1.0 0.400000 9 4 1 Chargers
26747 Titans 3300.0 0.0 0.000000 9 4 1 Titans
12029 Cowboys 2500.0 -3.0 -1.200000 8 4 1 Cowboys
1516 Vikings 3100.0 7.0 2.258065 4 4 1 Vikings
34362 Buccaneers 3400.0 0.0 0.000000 4 4 1 Buccaneers
28019 Raiders 2300.0 5.0 2.173913 3 4 1 Raiders
30308 Broncos 3300.0 9.0 2.727273 3 4 1 Broncos
31574 Patriots 3400.0 5.0 1.470588 3 4 1 Patriots
32841 Browns 2400.0 4.0 1.666667 3 4 1 Browns
35382 Eagles 3300.0 4.0 1.212121 3 4 1 Eagles
29038 Lions 2700.0 12.0 4.444444 2 4 1 Lions
36648 Giants 2900.0 1.0 0.344828 2 4 1 Giants
28275 Ravens 3000.0 3.0 1.000000 1 4 1 Ravens
35890 Steelers 3500.0 14.0 4.000000 1 4 1 Steelers

This leads me to believe CONDIA’s max exposure “rule” of no more than ~25% exposure applies to defenses as well as skill players.

And he typically picks a good defense. Every week one of his top 3 defenses scored at least 4 points/1k salary.

It’s still too early to tell but CONDIA seems to have some sort of “exclude” list on his defenses based on his knowledge of football matchups. Still, it’s far more general of a rule than I’d expect. To me picking 3-5 defenses should be enough but not for this guy. He’d rather take his chances on more players without negative covariance from touchdowns scored against his own starting D.

Saahil Sud’s Defense Selection

As a Fat Tony, Saahil takes more thought out shots at defense selection. He killed it in Week 1 picking the two best defenses

lineup_hash dk_salary dk_points points/salary Percentage week_num pair_size dst
3947 Jaguars 3700.0 28.0 7.567568 36 1 1 Jaguars
10358 Bengals 3100.0 4.0 1.290323 30 1 1 Bengals
6919 49ers 2100.0 4.0 1.904762 26 1 1 49ers
11122 Bears 2200.0 2.0 0.909091 4 1 1 Bears
10614 Rams 3200.0 28.0 8.750000 2 1 1 Rams
11378 Seahawks 2900.0 7.0 2.413793 2 1 1 Seahawks

He only picked 6 D’s compared to CONDIA’s 13 but Saahil only had 50 lineups entered in the contest.

In Week 2, Saahil picked 5 defenses with only one good pick. This time he had 150 lineups.

lineup_hash dk_salary dk_points points/salary Percentage week_num pair_size dst
16872 Chargers 2800.0 3.0 1.071429 35 2 1 Chargers
8152 Buccaneers 3200.0 19.0 5.937500 31 2 1 Buccaneers
25892 Redskins 2900.0 7.0 2.413793 23 2 1 Redskins
19915 Bears 2200.0 4.0 1.818182 10 2 1 Bears
26402 Patriots 2500.0 2.0 0.800000 1 2 1 Patriots

Like with the rest of his play, Saahil takes chances. I don’t have Week 4 results because Saahil only played in the Primetime contest. Looks like Saahil may have a ~35% rule on his max exposure.

Analyzing Saahil, I could not find remarkable DST-RB stacks or things of that nature.

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

Bales

Bales defensive selection strategy looks similar to CONDIA. 150 lineups entered, 10 defenses.

lineup_hash dk_salary dk_points points/salary Percentage week_num pair_size dst
5602 Bills 3900.0 9.0 2.307692 16 1 1 Bills
10837 Colts 2700.0 2.0 0.740741 16 1 1 Colts
16469 Steelers 3600.0 18.0 5.000000 16 1 1 Steelers
21932 Texans 3800.0 0.0 0.000000 16 1 1 Texans
27562 Falcons 3300.0 5.0 1.515152 16 1 1 Falcons
30708 Jaguars 3700.0 28.0 7.567568 9 1 1 Jaguars
33449 Bengals 3100.0 4.0 1.290323 5 1 1 Bengals
34918 Cardinals 2800.0 8.0 2.857143 4 1 1 Cardinals
31218 Panthers 3500.0 15.0 4.285714 1 1 1 Panthers
31722 Rams 3200.0 28.0 8.750000 1 1 1 Rams

Remarkably, he couldn’t really pick between his first 5 defenses and played them all equally.

In Week 2, 13 defenses, 150 lineups.

lineup_hash dk_salary dk_points points/salary Percentage week_num pair_size dst
7778 Buccaneers 3200.0 19.0 5.937500 13 2 1 Buccaneers
11646 Chargers 2800.0 3.0 1.071429 11 2 1 Chargers
18517 Panthers 3600.0 10.0 2.777778 11 2 1 Panthers
3185 Ravens 3700.0 17.0 4.594595 9 2 1 Ravens
14807 Steelers 3000.0 8.0 2.666667 9 2 1 Steelers
27195 Raiders 3500.0 9.0 2.571429 9 2 1 Raiders
21232 Cardinals 4100.0 10.0 2.439024 7 2 1 Cardinals
23850 Patriots 2500.0 2.0 0.800000 7 2 1 Patriots
29811 Rams 3000.0 2.0 0.666667 7 2 1 Rams
31754 Redskins 2900.0 7.0 2.413793 5 2 1 Redskins
33487 Seahawks 4000.0 8.0 2.000000 5 2 1 Seahawks
35219 Bears 2200.0 4.0 1.818182 5 2 1 Bears
36235 Jaguars 3100.0 -1.0 -0.322581 3 2 1 Jaguars

In Week 4 Bales only entered 19 lineups but 7 different defenses!

lineup_hash dk_salary dk_points points/salary Percentage week_num pair_size dst
2149 Patriots 3400.0 5.0 1.470588 21 4 1 Patriots
750 Bengals 3500.0 8.0 2.285714 16 4 1 Bengals
3156 Falcons 3200.0 4.0 1.250000 16 4 1 Falcons
3921 Cardinals 3200.0 6.0 1.875000 16 4 1 Cardinals
4671 Chargers 2500.0 1.0 0.400000 16 4 1 Chargers
1258 Jaguars 3600.0 15.0 4.166667 10 4 1 Jaguars
2405 Titans 3300.0 0.0 0.000000 5 4 1 Titans

Bales takes a pretty conservative approach to defense like CONDIA. Saahil is the crazy guy so far.

CSURAM88

CSURAM88 takes a mildly aggressive approach. Max exposure around ~26% but fewer defenses used than Bales and CONDIA.

Mobile Apps Category (English)468x60

In Week 1, 9 defenses used, 150 lineups. He hit big on Rams. Missed big on Texans.

lineup_hash dk_salary dk_points points/salary Percentage week_num pair_size dst
16863 Texans 3800.0 0.0 0.000000 26 1 1 Texans
8230 Rams 3200.0 28.0 8.750000 25 1 1 Rams
21413 49ers 2100.0 4.0 1.904762 13 1 1 49ers
30033 Bengals 3100.0 4.0 1.290323 13 1 1 Bengals
25610 Bills 3900.0 9.0 2.307692 12 1 1 Bills
32876 Giants 2600.0 2.0 0.769231 8 1 1 Giants
33639 Lions 2600.0 16.0 6.153846 2 1 1 Lions
33895 Steelers 3600.0 18.0 5.000000 1 1 1 Steelers
34151 Packers 2500.0 9.0 3.600000 1 1 1 Packers

In Week 2, 8 Defenses, 150 lineups, did pretty poorly

lineup_hash dk_salary dk_points points/salary Percentage week_num pair_size dst
6839 Chargers 2800.0 3.0 1.071429 21 2 1 Chargers
13201 Jaguars 3100.0 -1.0 -0.322581 20 2 1 Jaguars
18586 Steelers 3000.0 8.0 2.666667 17 2 1 Steelers
28622 Raiders 3500.0 9.0 2.571429 15 2 1 Raiders
23361 Rams 3000.0 2.0 0.666667 14 2 1 Rams
32512 Chiefs 2900.0 11.0 3.793103 11 2 1 Chiefs
32768 Buccaneers 3200.0 19.0 5.937500 1 2 1 Buccaneers
33024 Broncos 3400.0 13.0 3.823529 1 2 1 Broncos

In Week 4, CSURAM88 used 7 defenses, 16 lineups, and didn’t do well.

lineup_hash dk_salary dk_points points/salary Percentage week_num pair_size dst
2508 Broncos 3300.0 9.0 2.727273 31 4 1 Broncos
3778 Chargers 2500.0 1.0 0.400000 19 4 1 Chargers
509 Jaguars 3600.0 15.0 4.166667 12 4 1 Jaguars
1276 Jets 2600.0 4.0 1.538462 12 4 1 Jets
3018 Ravens 3000.0 3.0 1.000000 12 4 1 Ravens
765 Vikings 3100.0 7.0 2.258065 6 4 1 Vikings
4034 Cowboys 2500.0 -3.0 -1.200000 6 4 1 Cowboys

Recapping Our Findings

Ranking Pro’s by Defensive Prowess

  1. CONDIA/SaahilSud
  2. Bales
  3. CSURAM88

By Dr. John-ness

  1. CONDIA
  2. Bales
  3. CSURAM88
  4. SaahilSud

As you can see, CONDIA looks to be solid at picking defenses so far this year but Saahil has done well with less defenses used. I’d suspect as we do more of these position analyses that CONDIA will be solid all around but not exceptional. It’s his mathematical approach that helps him win whereas Saahil is a ballsy dude.

In general, I haven’t seen any notable stacking besides the anti-stack of not facing your own defense by these pro’s.

This post took me a while to write but I hope to continue these position group analyses as the season progresses. Let me know what position you want to see next. Leaning towards QB.

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 *