“I do all of my lineup building without any of the DFS optimizers.”—David Bergman, $2.5-million winner in the 2020 DraftKings’ Fantasy Football World Championship

New daily fantasy players often ask, “What is the best DFS lineup optimizer?”

That’s often followed by the question, “How do you use a DFS optimizer?

But those aren’t the questions new DFS players should be asking. What they should be asking is, “Is an Optimizer something I should even be usingl?”

Relying on “off-the-shelf” optimizers is virtually guaranteed to result in underperformance versus astute hand-crafted lineups, for the reasons that follow. We’d urge all readers to carefully consider the facts below before risking any money with an NFL DFS optimizer.

Reasons You Should Not Use a DFS Optimizer

1. You think they simply work…like magic.
   • Optimizers don’t think. They use common algorithms to generate random lineups. And those lineups aren’t even perfectly “optimized” (more on that below). Optimizers don’t know which players are not playing at a peak level. They don’t know player trends. They don’t know historical relationships. Those things are gathered from experience, research and modeling, which off-the-shelf optimizers won’t give you. Optimizers are anything but set-and-forget tools. Apart from faulty inputs (e.g., bad projections), operator error is their biggest risk.
2. You like to build just a few lineups
   • If your game of choice is single-entry contests, optimizers simply aren’t as useful — especially for cash gamesA cash game refers to a DFS contest in which roughly half the field wins a prize. The last place winner wins the same prize as the first place winner. While the prize payouts aren't as sexy as tournament contests (GPPs), the probability of placing in the money is usually higher. ... More. For double-upsA DFS contest where entrants must beat around 55-56% of competitors to win. Winners double their money. More and 50/50s50/50s are DFS contests where entrants win if they beat at least 50% of competitors. Winners receive $1.80 back for every $1.00 bet. With a 10% rake, 50/50 players must win 55.6+% of the time to grow their bankroll. More, you rarely need more than 1-3 lineups. After all, you’re searching for players with the highest floors. A proficient single lineup DFS player will easily beat your typical optimizer-generated single lineup the majority of the time. But wait, the pros swear by the optimizers they sell. How can they not work? Well, it’s not that the automated optimizers don’t work. It’s that they probably won’t work for you. Pros with fat bankrolls use optimizers to create dozens, even hundreds, of lineups. They rely on correlations and variance and build complex algorithms into their optimizers. In any event, you can be darned sure they’re not selling the secret sauce that makes them pros for $49 a month.
3. Optimizers rely too heavily on projections
   • All fantasy point projections are inaccurate—be they total points, ceilings, floors, or what have you, they’re inaccurate. Yet, those same projections are the main variable in optimizer calculations. Worse yet, most optimizers use projected fantasy point averages by default, with little regard for: (A) upside, and (B) how often a player hits his upside. Obviously, what everyone wants/needs are the least wrong projections — those with the least bias (since they’re all biased). A successful optimization system must have more consistent and accurate projections than the crowd. Otherwise, an optimizer will simply magnify projection errors — due to the large number of lineups being created. Relying too heavily on mass-market projections that hundreds of thousands of your competitors rely on dilutes your edge. That’s especially true in GPPs.
4. Managing variance isn’t easy
   • It’s not just projections you need to worry about. If you’re building multiple lineups in an optimizer, managing variance is essential. Among other things, it means adjusting player projections up and down by the right amount to get the optimal diversity of lineups. But here’s the problem. What if you adjust a projection higher because you think the opponent’s poor defense is not fully factored into the projection? But then again, what if it is actually factored in?  What you’ve just done is boost the projection error.
5. Optimizers hinge on exposures
   • Determining the correct percentage of rosters that a given player will appear in is fundamental to optimization. You don’t want to be overexposed or underexposed to anyone. Getting this part of the process wrong significantly reduces the chances of success. So does over-limiting the player pool your optimizer can pull from (i.e. excluding too many players from your “short list”). If you’re making these kinds of foundational decisions, you could just as easily spend that effort on perfecting a manual lineup construction process.
6. Optimizers rely on ownership optimization
   • You have to set limits on aggregate ownership (i.e., the total combined ownership of all players in your lineup). There’s more science than art in that. Moreover, game theory and projecting the crowd’s opinion of player value remain paramount. This is but one more complexity of optimization prone to error.
7. Optimizers are only quantitative
   • “Qualitative” factors often get dismissed by optimizer pushers. We touched on this above. Optimizers don’t adequately factor in variables like player and team motives, psychology, shifts in team strategy, environment (e.g., what sort of contests does a player thrive in), the impact of lost teammates, player matchup details, player utilization and so on. How can you find the best DFS QB, for example, if you’re not considering each player’s context that week? The reason optimizers are mainly number crunchers and not context analyzers is that qualitative factors are not like yards or TDs. They cannot be as easily quantified. As a result, qualitative factors can’t be readily plugged into an algorithm. That’s good for you, however, because it creates inefficiency in the market—and exploiting inefficiency is how you win. To put this all another way, optimizers rely too heavily on mass-market data. The more people who know a piece of DFS information, the less valuable that information becomes.
8. Most People Lose
   • Roughly 73% of DFS players lose, as of the time this is being written. The top 1% make 45% of the money on DraftKings, as of this writing. More interestingly, a 2015 report found that the top 11 DFS players spend $2 million+ in entry fees a year on average, and they’re all using optimizers of some sort. But if you’re trying to exploit Draftkings inefficiencies, you don’t want to toil away competing against stat geniuses who run multi-million-dollar quant models on a custom NFL DraftKings optimizer. You want to do something different, like find a consistent niche methodology (possibly a more qualitative approach) and refine it. And remember, increasing your entries with an optimizer may boost your win probability, but does it boost your probability more than it boosts your lineup costs? For most people, the answer is no.
9. You have limited resources
   • To make optimization worth it on a large scale, you need enough bankroll to deploy enough lineups. And enough time to manage it all. You can’t just create 150 lineups and not look at them before submission. For those with smaller bankrolls, searching for the best overall values can have a higher return on time.
10. You’re not getting the most optimized lineup
    • With practically infinite player combinations, processing power limitations prevent publicly available optimizers from generating true theoretical top lineups. What you get is an approximation of the most optimized lineup(s) using mathematical and programming shortcuts.

Conclusion: DFS optimizers

Too many optimizer users put too much faith in the simple logic that spits out optimizer recommendations. They see the shiny marketing and the big names associated with these tools, and then they click and pray….and lose.

An optimizer doesn’t just magically give you skill. An optimizer reflects skill. You have to make multiple decisions correctly to leverage an optimizer. Even top pros manually select 2-4 players as locks in all of their lineups. So if you’re going to do all that analysis work for 2-4 players, you might as well perfect your process and do it for the rest of the positions, with a hand-built lineup. At least that way you don’t have to get both your player picks right and get the optimizer settings right.

Don’t get me wrong, optimizers are an incredibly powerful tool in the right hands, with the right logic built in. In fact, once you reach pro level, one might argue that optimizers are the best way to scale your winnings. They’re practically indispensable for the algorithmic selection of 10+ lineups. For the best of the best who may submit 150 lineups in a given contest, imagine the gargantuan effort of creating all those rosters by hand!

But until you graduate to that point, to the point where you’ve refined your winning formula to beat the house edge, where you’re routinely cashing and handily beating the rake, optimizers are better left for another day.

People make a healthy living by hand-curating lineups using reliable disciplined methods that marry qualitative and quantitative analysis. Focus hard on learning what stats matter and improving your process. Whether your system is manual or automated, the DFS game boils down to successfully finding inefficiencies — inefficiencies that most people are too casual or inexperienced to exploit. An optimizer simply won’t be an optimal way for you to return on your investment.