Chasing The FanDuel Sunday Million
One of my goals for this NFL season was to find more time to do data analysis and turn that into creative content. I feel like I’ve been able to do that in a few ways this season:
Still, I wanted to do a regular piece that was a bit more fun and hopefully is able to portray how I go about using a lot of the information on the site.
I figured, hey why don’t I try to win a $1million in the process? After all, I’m surrounded by million dollar winners and starting to feel a bit left out.
SportsGrid CEO Jeremy Stein, who is making lineups for the DR Sweat promotion, has done it TWICE.
I was live sweating with Drew when he pulled it off:
FanDuel user and DailyRoto subscriber ‘skipbidder’ just joined the club this past Sunday:
HUGE congrats to DR sub 'Skipbidder' on his $1,000,000 score. Love the humility too.
"I did okay too. Thanks guys." pic.twitter.com/MPKbVJimte
— DailyRoto (@DailyRoto) September 10, 2018
Each week I’ll go through the process of how I made my lineups, and the macro (roster construction) and micro (specific player pool decisions) strategies employed. I’ll then evaluate the results and how I likely got screwed by variance with no other explanation for my failings.
I’ll document the results in here, although the goal isn’t to grind out weekly profit in this format. The goal is to give myself the best chance to win life changing money, something I feel got lost in my macro strategy this week (more on that later).
In general, I’ll keep my entry fees in the $1,000 – $2,000 range. So, Week 1 that allowed me to max enter 150 times. In Week 2, the entry fee is $15. I’ll probably enter around 100 lineups.
Without further ado, here was my first attempt of 2018 to win the FanDuel Sunday Million:
Week 1 Results
Entry Fees: $9 x 150 entries = $1,350
Entries Cashed: 35 (23.3%)
Gross Winnings: $560
Best Entry: 5,004 / 457,385 ($25)
Net Profit: -$790 (-58.5%)
What Went Right
Projections, NO-TB Stacks with Fitzpatrick
Overall our projections did well. The winning lineup had our fifth best QB value (Ryan Fitzpatrick), our Top 3 RB values (David Johnson, Alvin Kamara, James Conner), our first and 10th best WR values (Kenny Stills), and our top cheap defense (Miami Dolphins).
That’s part of what made this week so frustrating for me. Success was there for the taking. And I did have exposure to the TB-NO game. Ryan Fitzpatrick was my third highest exposure QB (13%). Alvin Kamara was my second highest exposure RB (32%). Michael Thomas was my highest exposure WR (26%). I played both Ted Ginn Jr. and Chris Godwin at a 13% clip, well above the field.
It didn’t work, however, for a few reasons. From a macro standpoint, I spread myself too thin (more on that later). From a micro standpoint, I was unwilling to play Kamara and Thomas in the same lineup, outside of Brees stacks (minimal as I felt cheap QB was the best play on FD). While I had a lot of Kamara, the field had more (35%). I also didn’t force diversity in my Tampa Bay stacks. While Godwin had a good game for his price and ownership, it meant nothing with Mike Evans and DeSean Jackson meaningfully outscoring him.
A small projection gap in favor of Godwin along with a cheaper price tag, left me with no Jackson, who ended up the key to the slate. It’s a spot where I should have focused a bit more on the upside of the overall situation and not taken such a big stance on two high upside options partaking in it, especially in light of my offseason look at WR expectations (linked in the intro) that indicated DeSean Jackson was probably very unlucky in YPT (yards per target) last season.
High Ceiling WR Plays
Using our floor/ceiling projections, I had double digit exposure to Emmanuel Sanders, Kenny Stills, Jamison Crowder, Tyrell Williams, and Ted Ginn Jr. These 5 WRs ranked in the Top 6 in 90th percentile projection under $6,500.
In particular, the floor/ceiling projections (methodology linked in the intro) have shown higher YPT guys with high team totals to have ceilings beyond what you’d expect for their base projection. Ted Ginn Jr. ranked 31st in our median projections but 23rd in our 90th percentile projections:
Crowder was a bust, but Sanders, Stills, Williams, and Ginn combined for 5 TDs and two 100-yard performances. This saved me from an even worse ROI week.
Heavy Gronkowski Exposure
Using our floor/ceiling projections, I simulated the tight end position 4,000 times to help me get a sense of how to attack the position. Here are the probabilities of finishing as the Top X tight end on the slate:
Gronkowski dominated the simulations. His probability of being the highest scoring player in the tier was 31.8%, or roughly the equivalent of the chances of the next seven tight ends combined. There was nearly a 50/50 chance he’d be Top 3 at the position. With pricing a touch loose on FanDuel, it was clear this was a smash play. Gronkowski ended up as my second highest exposure player at 41%, nearly double the ownership of the field.
I’d Do It Again
One spot that hurt me a lot was going contrarian at the RB position. For the most part, I think the process was right, although I’ll point out a couple of spots where I should have treated the information I was using with a bit less confidence than I did.
There were three main contrarian choices for me:
Underweight on Alvin Kamara
This choice did the most damage to my ROI this week given that Kamara put up 38.6 FD points at 35% ownership. Given Kamara’s high projection, I still had a lot of him. However, we projected him at 42% ownership. We were pretty close. That’s a really high number for an expensive skill player to justify given NFL variance – TD variance, injury potential, etc.
It almost worked fine, as Kamara had a low 20s game for a while, but a frantic fourth quarter led to a gimme TD from the 1-yard line (Michael Thomas tackled at the 1) and another receiving TD later in the quarter. This was certainly within Kamara’s range of outcomes, but it was a ceiling game with an unpredictable game script, buoyed by Ryan Fitzpatrick playing the game of his life and a Tampa Bay D/ST TD leading to an extra possession and negative game script for New Orleans.
We had Kamara at a 50/50 chance to exceed 19.5 FD points, and at those odds, I was willing to go underweight at 42% projected ownership. However, I should have taken into account the margin of error on the projected ownership, which was still very high but lower at 35%, something Drew and I suspected might happen in the Saturday Update for premium subscribers.
Heavy David Johnson
We saw some of the risk with David Johnson realized under a new coach, new QB, and a year removed from injury. Still, he was our highest projected player on the slate, with Kamara projected to out-own him 2.85 to 1. Based purely on the projection and ownership information, I was happy to take a stance on Johnson.
However, it’s a situation where some of my inclination that the two would be owned a little bit closer to one another and understanding that there was a bit more guesswork with Johnson’s projection could have led to a better decision (even exposure between the two). Regarding the former point, Kamara ended up out-owning Johnson at a 1.43 to 1 clip, about half the projected ratio.
Saquon and Shady Contrarian Plays
One of the strategies I wanted to employ this week was taking advantage of the field overreacting to bad matchups given the unpredictability of the NFL, especially in Week 1. Volume is king in the NFL, and with Saquon Barkley and LeSean McCoy I felt I was getting a lot of volume from two really talented backs at extremely low ownership.
We had McCoy and Barkley projected for the fifth and sixth most carries respectively and the 10th and seventh most catches at the position as well. Their MS (market share) of Rush TDs was also projected to be in line with other top backs, albeit for teams with much lower totals:
The McCoy play was an utter disaster. As bad as the perception of the Bills was heading into the game, it was tough to predict the complete lack of competence. It’s the type of play you make realizing the downside, and the downside hit. We’ll adjust our baselines for the Bills offense as a whole, which will alter the decision process moving forward, but the projected volume relative to the ownership spot was in a position I felt confident attacking.
The Barkley play almost went equally as poorly, until his game changed on a dime with an electric run:
— Onward State (@OnwardState) September 9, 2018
While the run by itself was a lucky result for me, you’re not expecting consistent success in a difficult matchup. You are expecting lots of opportunities for a generational athlete to make something happen.
Like with the Kamara-Johnson decisions, the overall process I was happy with on the Barkley-McCoy plays. However, at that level of contrarian I could have easily been over the field at 10-15% exposure. Instead, I went up to around 20% on each player. This would have allowed me to mitigate the damage of one of these players in a difficult matchup hitting a floor game, like McCoy did. Simultaneously, that would have cleared more exposure for me to devote to simply our favorite values, like John Connor (more on him below).
What Went Wrong
One of the struggles I’ve had as I’ve dug into more data and played more MME recently is balancing the variance of individual sports and DFS contests with the type of events you need to happen to actually win.
The high degree of variance is something that makes you want to spread out your ownerships and pick up small bits of leverage against the field in a lot of different ways. It’s important to acknowledge and truly comes to grip with just what variance means.
Rather than look at it from a data perspective or without bias, we often view results with hindsight bias. If a chalk player goes off that you faded, it’s tempting to think you got too cute and still needed to use the “best plays”. If a chalk player stumbles, it’s tempting to think you should have accepted the variance of the situation and picked up leverage on the field. Unfortunately, it’s not so black and white.
For example, I’m kicking myself for not playing even more of Michael Thomas, who was still my highest exposure WR, given the high-value projection we had for him relative to the rest of his peers:
Yet, prior to making my lineups, I simulated his chances of scoring at least X amount of points for every point value between 10 and 29 points:
Even with our best projection of the week, we gave Thomas only an 8.5% chance of scoring 29-plus points. Heck, we had his odds of failing to score in double digits at about triple that.
So, while it’s true that variance can mess with our heads, both before and after the fact, and is a very real thing, I ultimately believe this has led to a recent mistake in my MME strategy of spreading myself too thin.
More Aggressive Approaches
Some more aggressive approaches could have yielded some huge results for me using our projections. As I noted above, DailyRoto user ‘skipbidder’ took down this very contest. How did he do it?
He locked the Top 3 value RBs (David Johnson, Alvin Kamara, James Connor), whittled his QB pool down to four players, and used stack settings to make sure his QB was paired with a WR and that a skill player on the opposing team was used as well.
As Colin Drew noted on Twitter, you also could have arrived at this lineup simply running straight optimals with the stack settings set by pairing your QB with a WR on his team and one from the opposing team:
How to win $1,000,000 with the DailyRoto optimizer.
— DailyRoto (@DailyRoto) September 10, 2018
Now, I don’t think I’m ever going to run that tight of a core. The goal is to get one lineup into the Top 1% and running a super tight core really ratchets up the variance of your week to week play. With that said, I think there’s a happy medium of taking the more contrarian, spread out strategy I employed that embraces volatility while still taking heavy stances, even sometimes on chalk players, when I feel strongly or it is in line with the projections.
Marrying the two concepts allows you to plant your flag so that if one condition is met (for example, stacking NO-TB goes off) you immediately have high-end upside in a number of lineups. Spreading the risk and acknowledging volatility with your secondary plays allows you to cast a wide enough net that when you hit on whatever stance you take, it’s not completely ruined because you tried to take heavy stances everywhere and one of those stances doesn’t hit.
This is similar to how I finished runner-up in the 2017 DraftKings Masters Millionaire Maker. I locked an underpriced golfer, who I felt was a phenomenal value and could help me win with a Top 5 finish (Sergio Garcia). Then I cast a pretty wide net outside of Garcia, that allowed me to be overweight on some low-owned golfers like Thomas Pieters.
Of course, no single strategy is going to guarantee you success. If that was the case, everyone would do it, and we’d all chop a million dollars each week. Some weeks you are going to whiff on your core. Other weeks your player pool might be lacking the low owned player that breaks the slate (DeSean Jackson).
Still, I hope by being more aggressive in a few different spots I give myself a better chance at the rare and life changing outcome of winning $1million.
Where should I have been more aggressive?
There are two very distinct spots I wish I went more aggressive this week.
The first is with Rob Gronkowski. Given my analysis of his odds to win the tight end position outright or even finish Top 3 at the position weighed against his projected ownership (19.4%, actual was 21.6%), this is a spot I may try to triple up the field on or even lock the player in the future.
The other spot I was disappointed in myself was with James Connor. While I’m happy we had Connor as a Top 3 value, this is a projection I actually feel like was a small mistake on our part. Early in the week when the Le’Veon Bell news started to hit, my inclination was that Connor would be a near lock play regardless of ownership if Bell was out, based on his successful preseason along with the Steelers’ history at the position when Bell has been out (see DeAngelo Williams’ success).
However, some jitters over a gaudy projection and a few Jaylen Samuels takes later, we ended up going a bit conservative on Connor: Here was our projection on Connor that had him as the third best value at the position and easily the best source of cap relief:
As well as that had Connor rate, the final projection still left room for him to fail or for his cap relief to not be necessary to be in the nut lineup if the raw production wasn’t Top 5. While Connor had a ceiling game, my initial instinct in regards to his volume would have had him projected even better, which is something you can fool around with using our customizable projections:
At those volume levels, Connor would have been a value on par with Johnson and Kamara, forming a clear tier above the rest of the pool. From there, with the cap relief he offers, he would have gone from a player I was planning on being even with the field on to someone that I could have easily gone double the field on or locked in, scaling back ownership on some of the contrarian plays like Barkley and McCoy in the process.
Connor ended up with an 89% MS of the team’s carries and a 21.7% MS of the team’s targets.
Of course these stances won’t always work out. As noted above, I did take stances on Barkley and McCoy, and there was no getting around the McCoy bomb for me based on my intentions for the week. Still, I could have reduced some of the stances I took on the super low owned players while making more stances on players that the projections simply liked. Also, I could have gone through a similar volume exercise with Royce Freeman or Dalvin Cook, two other backs where you could have made an argument that, similar to Connor, we were too conservative on from a volume perspective (and that turned out to be prudent).
I wasn’t the only one who maxed out the FanDuel Sunday Million contest. As part of our DailyRoto Sweat promotion, our two-time $1million winner CEO Jeremy Stein maxed out the tournaments, giving subscribers a chance to share in his winnings. Jeremy ran a 9 QB build set that paired a QB with two of his WRs and focused on the optimizer’s top values. The SWEAT lineups turned a very small profit and our best lineups finished 152nd in the Sunday Million and 12th in the Sunday Blitz respectively, with decently heavy Fitzpatrick exposure but most of that paired with Mike Evans and Chris Godwin. To be SWEAT eligible this week you must be subscribed as of Wednesday at 12pm EST.