In case you missed it, I had a nice weekend on DraftKings. Well, really, I had a very lucky weekend. Here’s what I posted on Facebook prior to the Monday Night game:
This past weekend Sonya and I were supposed to go to Seattle with Brandonand Caitlin for 4 nights, mainly due to a charity auction package I won throughThe V Foundation for Cancer Research‘s ESPY day eBay auction. The package stated that we could go to a Monday Night Football game of our choice and get VIP treatment – field passes pregame, tour production area, meet the cast and grew (Jon Gruden!!). Unfortunately, communication broke down a bit and we had to reschedule the experience til December 14 in Miami (poor us lol) and cancel our Seattle plans.
As a result, we ended up staying in Las Vegas until Sunday (yesterday) night rather than leaving Friday and heading to Seattle. We made the most of it and had an awesome trip nonetheless. Also, I woke up at 5:30am PST yesterday morning Vegas time and made my weekly DFS lineups. At around 9:30am PST I learned that Jay Cutler was starting and put in a few last minute Cutler stacks, thinking it’d be a good underowned spot. And here we go:https://www.draftkings.com/contest/gamecenter/10470315…
I need 7.65 points from the Seattle D to win the above tourney for $150k, and am plenty live to win a FFWC seat ($80k value) and a bit more in some other contests. Let’s go!!!
Aside from the winning lineup, I did quite poorly this weekend at NFL DFS – I played close to $20k of volume in HU matches with a negative ROI, entered 14 lineups in FantasyDraft GPPs and lost small, and cashed only 1 of my 14 DK GPP (Guaranteed Player Pool – the DFS equivalent of a poker MTT) lineups in the Blitz, Millionaire Maker, and some satellites. Luckily, the one winning lineup cashed for about $180k in equity and turned my weekend from a loser (would have been close to minus-50% ROI on DK and probably around minus-30% to minus-40% across all sites) into a big winner, with about an 850% ROI on DK and close to that across sites, since most of my volume was on DK. The difference between me losing ~$13k and winning ~$165k for the weekend was that one lineup, submitted about 20 minutes before kickoff. Glad I didn’t pack up shop early to head out to watch the games!
It’s intuitively obvious to me that DFS GPPs will have massive variance, but maybe less clear to some of you. This article on poker MTT variance should help put it in perspective some. FWIW, I think top NFL DFS players are making maybe 30-40% ROI longterm in fairly soft large-field GPPs like the Millionaire Maker (can be higher given overlay but otherwise I think it’s very difficult to be >40%), and ROIs are smaller in the higher buyin GPPs. For further reading on variance, Fooled by Randomness and The Black Swan, both by Nassim Taleb, are fascinating reads examinating variance across a wide spectrum of examples.
What are the important takeaways about DFS variance? For one thing, realize that GPPs, like poker tournaments, are not ideal candidates to be one’s primary source of income. There is plenty of profit potential for sure, but if one plays GPPs only – even if his methodology for making lineups is very strong – it’s very easy to have large downswings. The good news is that GPPs, unlike poker tournaments, often allow multiple entries per player (Millionaire Maker allows 500 per player), so players can reduce variance by diversification within a lineup portfolio. For example, when I play DFS MLB GPPs I tend to target several stacks (groups of hitters on one team – stacking is nearly mandatory for MLB GPP success) I like and build lineups around each of them. On a typical day I might have 8 lineups using 6 different stacks. If by contrast I decided to just make 8 lineups using variations of the same stack, my variance would be significantly higher. By diversifying lineups, one can greatly reduce downswings. Still, just understand that GPPs have an enormous amount of variance even if lineups are well diversified.
So should I play more cash games (in DFS heads up and 50/50 games are considered cash games) to reduce variance? Not necessarily – EV matters even more than variance. Many of the 50/50 games offered are populated widely by top pros who probably make better lineups than you. In the large-field multi-entry 50/50s, a top pro often has his best lineup entered the maximum amount of times (or a collection of his best few lineups). I’m not the foremost expert on DFS cash games (this MLB season I just stayed away since I wasn’t putting in tons of time and don’t have a model – some friends have told me there is actually soft action though), but to me it’s much more clear that there are solid edges to be had in GPPs than in 50/50 games. Heads up is another story, but the HU DFS economy is similar to a poker HU SNG economy – if you’re not good enough so that top pros lack incentive to take your games (10% rake at most stakes; must win 5/9 of games to breakeven after rake) then these pros will just gobble up most games you post. NFL has smaller edges than other sports, so luckily for me I’m mostly able to repel pros even if I’m slightly less good than them, but new players will quickly have any HU games they post in the lobby taken by pros until they prove that they can beat the pros 4/9 of the time basically. DraftKings does allow you to block 3 players from taking any games and also to check a box when creating contests to allow others to take only one game per buyin level, but this isn’t enough to safeguard new players from sharp action (for the true beginners, the beginner games are great, but you only get a small fixed amount of them).
My approach to NFL DFS this year has been to try to get as much non-t0p-pro HU action as possible and to supplement that with GPP entries, focusing on contests that look to be overlaying (this is why I targeted FantasyDraft this weekend and also why I entered all my lineups in the $300 Blitz). I dabbled with 50/50s some but found them to be quite tough and passed on them this past week, although I’m always re-evaluating everything. For MLB DFS this year, I’ve ignored cash games entirely and just focused on GPP play at my convenience (skipping 2 months during WSOP and plenty of other days due to travel/life/etc). It’s gone very well but I’ve run very well also. More importantly, there was a significant learning curve involved. I’ve been playing DFS since 2011, when I was an initial investor in DraftDay. I’ve spent countless hours reading, talking strategy, browsing forums, and thinking about strategy. DFS, like poker MTTs, offers the chance to luck out and get rich quick – that’s entirely true. If you want to make it a long-term profitable endeavor, though, it’s important to work hard and be willing to devote time, effort, and mental energy to it.
How do I get better at DFS? Here are a few quick links to get started – happy to answer any questions about them:
- NFL correlation matrix – data analysis that shows which positions correlate with each other and in which directions. You’re looking for positive correlations in GPPs to maximize ceiling/upside (the obvious are QB/WR and QB/TE stacks), but looking for minimal correlation in cash games to maximize floor.
- Jonathan Bales’ books – link is to the latest one, which I haven’t read, but the entire series is data driven and contains valuable insight that will help you to get from beginner to competent quickly. Just be mindful that the top pros are all well aware of all the information in these books, and that game theory matters a lot in GPPs – I built my last minute Cutler stacks because I knew given the late-breaking news that he was starting he would be criminally underowned in a plus matchup against Oakland. I pivoted heavily to Martellus Bennett in many lineups also since he became an elite value with Cutler healthy and figured to be somewhat underowned.
I would add that there is certainly no shortage of NFL info out there, and it’s important to find the right info. Trust data-driven analysis over qualitative opinions generally, although some NFL (and other sport) analysts do have some valuable insight to offer.