Poker To DFS: How I Learned To Win At Daily Fantasy

I didn't find daily fantasy sports because I loved sports. I found it because a guy on a poker forum who was making more money than me announced he was quitting to play fantasy. That was the whole scouting report. If he was beating me at poker and he was leaving, the poker to DFS math wasn't complicated.

What I didn't expect was how little of my poker skill lived in the cards, and how much of it transferred anyway. Less of my edge came from reading a hand than I would have guessed. Most of it came from choosing which table to sit at, being honest about what I was actually good at, and building a process that could survive months of being right and still losing. DFS is that same game. Most players lose because they think daily fantasy sports strategy means picking the right players, when it mostly means picking the right opponents and then getting out of your own way.

One warning before we start: the sharpest lineup decision in this article is one where I deliberately rostered a player who gave me no projection edge of my own. I'll get to it. It'll look like a mistake until it doesn't.

Watch The Full Roundtable Video

This piece is built from an early Stokastic roundtable (recorded when the site was still Awesemo.com, now Stokastic.com), where the team walked through my background and how I actually approach a slate. Watch on YouTube.

What Poker Actually Teaches You About DFS

People assume the crossover is math. Mostly, it isn't. The transfer happens in two places: the lifestyle and the strategy.

The lifestyle piece is that both games hand you loud, constant feedback that's mostly noise. You win a night, you lose a night, and in the short run that happens regardless of how good you are. Learning to stay even keeled, to not get too excited about the wins or too bummed about the losses, is a job requirement if you want to make a career out of this. I'd put it near the top of why I'm still here.

The strategy piece is that both are games where you're trying to outsmart a field, not solve a problem. Poker is hard statistics. DFS is picking teams of fantasy players. Underneath, you're doing the same thing in both: finding the spot where the other people at the table are wrong.

Poker Concept	The DFS translation
Table Selection	Contest selection: who else is entering this GPP or head-to-head
Rake	Site fees, which is the number you actually have to beat
Variance And Tilt	The nightly swing that says nothing about your edge
Hand-Reading	Ownership projections: what is the field going to do
Bankroll Management	DFS bankroll management: never risking more than you can lose
Hand History Review	Back-tested slates and the distribution of your finishes

The row that matters most is the first one, and almost nobody acts on it. In poker I learned the ugly, load-bearing fact of every skill game: for you to win money, other people have to lose money. Read that as an instruction about where to sit.

DFS Game Selection Is The Whole Game

When I first heard about Draft Day, a site I came to through the CardRunners crew, my reaction wasn't "this looks beatable." It was "how can anyone beat a 10% rake?" That was twice as high as the poker games I was playing. The fee was the opponent.

So I did what a poker player does. I looked for the softest table. I kept a database of my head-to-head results against professional players, and across the whole sample, neither of us was really winning. We were both just feeding the rake. Two skilled players paying a fee to flip a coin isn't a business, and that's the mechanical reason it's easier to play contests full of recreational players than to take on the pros.

Head-to-heads have stayed a small part of my volume ever since. Part of that is logistics, because the slates make it hard to get the volume I'd want. Part of it is that if you don't cancel an unpaired match before it starts, you get paired against a pro, and a pro-versus-pro match is a fee with extra steps.

The same honesty applies to which sports you play at all. When I started, I assumed the highest-volume player on the leaderboards was the best, mostly because he was on top every night. I never tracked his results closely enough to know. What I do know is that he moved serious money into high-stakes hockey and NASCAR, which weren't his forte, and I don't know how that ended up for him. NASCAR is the hardest sport I've tried, and I still barely have a comprehensive strategy for it: too few drivers, too few stats. So I don't fight there. I try to find the fights I know I can win, and if I don't have a proven track record in a game, I don't put a real wager on it.

If you're still assembling the foundation under all of this, our DFS beginner guide and our guide to building a DFS bankroll cover the fundamentals I'm assuming here.

Build the pool the way I do it. The Stokastic Sims simulate a contest tens of thousands of times and generate the lineups out of that pool, so your build is optimized for win probability instead of one projected score. Ownership projections and Top Stacks feed it, and Late Swap keeps it honest after news breaks. Start with MLB and take 10% off your first month with code POKERDFS10: see Stokastic MLB pricing.

Worked Example: The Head-To-Head Lineup That Made No Sense

Here is the promised one.

Years ago I was playing the same recreational NBA player in head-to-heads nearly every night. He kept rostering Enes Kanter, back in Kanter's Thunder days, when Kanter was one of the most volatile players in the league. And he kept beating me more than he should have.

So I went and did the work, and the answer inverted everything I assumed about head-to-head lineup construction. In a head-to-head, my edge is my projection, and projection edges only cash if the night is boring. What Kanter gave my opponent wasn't points. He gave him variance, a lever that let a worse lineup leap a better one often enough to erase my advantage.

The fix wasn't to out-project him harder. The fix was to roster Kanter myself.

Once Kanter is in both lineups, he can't differentiate us. My opponent's variance lever disappears, the night gets boring, and a boring night is one where the better projection simply wins. When I looked into it, rostering him came out better than fading him. I never wrote down the sample, so take that as the logic rather than the proof. The logic is the part that generalizes anyway.

That's the whole poker lesson in one roster spot. I wasn't picking a player. I was picking what kind of game we were going to play.

Cash Vs GPP: Two Lineups From The Same Projections

That example only works because head-to-heads and tournaments want opposite things, and running one build in both is an expensive habit.

In a head-to-head or a cash game, I put out the lineup with the highest projection, and that's nearly the entire thought. Beat one person, or beat half the field, and floor is the currency.

In a GPP, projection is table stakes and the real inputs are ownership and how much diversity I want between my own lineups. I'm no longer asking "who scores the most points." I'm asking "who scores the most points that the field doesn't have," which is a different question with a different answer. Our DFS tournament guide and our NFL DFS leverage breakdown go deeper on that split.

I almost never hand-exclude players. The projections plus the ownership read do the filtering: if a guy is too highly owned relative to what he projects for, he falls out of the lineups I actually enter. There's one exception I'll admit to. The player I always pulled out by hand was Tiger Woods, because my opinion of him always ran so much higher than his actual skill level. Knowing the exact shape of your own bias is worth more than pretending you don't have one.

Ownership And Leverage, Or Why I Had 20% Of A 1% Player

The clearest version of that idea I have ever gotten was an NFL slate where the entire industry had decided Alex Smith could not have good games. He came in around 1% owned. I ran 20% ownership of him across my lineups, and he went off.

I want to be careful here, because this is where DFS writing usually goes wrong. That night probably had more to do with luck than analysis. One slate proves nothing. The repeatable part isn't "fade the consensus," it's the structural thing underneath: I have an affinity for players the field has decided have no upside, because "it hasn't happened yet" and "it can't happen" are wildly different claims, and the field prices them the same.

The field isn't a person. Treat it as a pricing mechanism with a blind spot, and treat ownership projections as how you find out where that blind spot sits tonight.

DFS Projections: Build Your Own, Or Borrow And Bend

Everyone wants to know whose projections to trust. Here is the honest answer nobody in this industry likes to give.

Building your own projection system from scratch is a full-time job. It takes expertise most people don't have and access to data sets that are genuinely hard to get. When I started, I made my own projections my obsession, precisely because I believed doing that better than everyone else would be a big source of edge. It also consumed my days.

A lot of the guys I've talked to at live events seem to do something smarter: they take a projection source and add their own twists. I want people to do exactly that with ours. The numbers on Stokastic come directly from my own projections, though I'll be straight with you, I'm not handing out my exact numbers, because those are important to my edge and I'd have to charge far too much to justify it. What you get is a real baseline, and the trade is that you spend your scarce hours on the intuition layer instead of rebuilding the engine. If you want to see how that engine works, our DFS projections walkthrough opens it up, and the DataHub is where the numbers live.

How much intuition you add is sport-dependent, and this is where most cross-sport players get lazy:

• MLB: the data is rich and the roles are known. You know roughly how much everyone plays and where they hit in the order, so I put very little intuitive touch on it. The strategy in baseball lives in how you build the lineup around the projection, which is why MLB stacking is the sport's real skill.
• NBA And NFL: you have to spend real time researching context, because how a player fits on his team drives the projection before any stat does. For a football model, the core is yards per play by play type and the number of plays a team runs, and you work up to yardage and touchdowns from there. Red zone usage is a genuine factor, not noise: bigger players get more red zone targets, and a model that ignores that will misprice touchdowns.
• NHL: the information arrives late. Most of what matters isn't available until about three hours before lock, so I do everything then and accept I may be leaving a little edge on the table.
• NASCAR: the information barely exists. Too few drivers and too few meaningful stats to build a comprehensive strategy around, which is a different problem entirely from hockey's timing one.

MLB is the sport I find most engaging for exactly the reason it needs the least intuition. Nobody wins that slate by projecting best. They win it by building best.

Judge The Process, Not The Night

Every part of this collapses if you can't tell whether it's working, and DFS is unusually cruel about that.

Ask the question honestly: how many months of results do you need before you can call yourself a profitable player? The answer is longer than anyone wants. There's enormous variance even across a full season. Whether you win a big tournament five times or ten times over a career is both entirely realistic and a completely different bank balance, and the dollars will tell you a story about your skill that the dollars don't actually know.

So don't read the dollars. Read the distribution. The measurement I've wanted to build is an adjusted ROI based on your percentile finishes rather than your realized winnings, because it strips out some of the luck and shows you the shape of your play instead of the outcome of it. Your distribution of finishes teaches you something. Your winnings mostly flatter or scare you.

Two habits get you closer:

1. Go back and re-run old slates. You can take a completed contest and test what a different strategy would have done: a 5-3 stack instead of a 4-4, a different exposure setting, a different pool. One night tells you nothing, but a few dozen back-tested nights start to draw a picture you can actually improve against. The Contest Sims attack the forward-looking version of the same question, simulating the slate before it happens rather than reconstructing it after.
2. Don't sweat the scoreboard. I'm not a robot, and I'll check my winnings during a night if I let myself. The experience is more bad than good. I'd rather have fun and look at the end of the night, because nothing I learn mid-slate changes a lineup that's already locked.

That instinct connects back to the fee I couldn't stop thinking about at Draft Day. Everything I've described, the table selection, the honest self-scouting, the boring head-to-head lineup, is one long argument with the rake. Our piece on DFS variance versus sports betting variance is a useful companion here, because the swings look similar and behave nothing alike.

In Summary: The Edge Moved, SO Move With It

Zoom out and DFS is running the same life cycle poker ran. Information became more available, and the edge keeps getting smaller and smaller. I expect the sites to eventually cut fees to keep people playing volume, possibly through rewards programs rather than a straight rate cut, because it matters to them that the games stay winnable.

None of that is a reason to quit, and it's certainly not a reason to quit your job. I found the business side of this considerably harder than I expected, and I'd tell anyone that playing DFS full time is a worse plan than it looks from the outside. What it does mean is that you should stop looking for the edge where it used to be. The edge no longer lives in knowing a stat nobody knows. It lives in choosing games you can beat, being ruthlessly honest about which sports those are, and measuring yourself on the distribution instead of the dollars.

My process never got simpler as I got better. It went from simple to elaborate, year after year, and that's the tell. In poker, the last frontier was the tracking software that told you exactly how a specific opponent played, and DFS still doesn't really have that. Whoever builds it will land where I landed the first time I heard about a 10% rake and thought it was unbeatable: the players were never the puzzle. The table was.

Skip the part where you rebuild the engine. The Stokastic Sims run the contest tens of thousands of times, build your pool, and set your exposures, with ownership projections, Top Stacks, Boom/Bust, and Player Compare underneath them. Use code POKERDFS10 for 10% off your first month: start with Stokastic.

FAQ

Do poker skills actually transfer to DFS? The valuable ones do. Table selection becomes contest selection, bankroll discipline is identical, and tolerating variance without changing your process is the same skill in both games. Hand-reading doesn't transfer directly, but the instinct behind it, modeling what everyone else is about to do, becomes ownership projection.

What is one of the most expensive DFS game selection mistakes? Playing head-to-heads against professionals. Two skilled players in a head-to-head are usually just splitting losses with the site's fee. Recreational-heavy contests are where a real edge gets paid.

Should I build my own DFS projections? Only if you can treat it as a full-time job. Building a projection system requires expertise and hard-to-source data. Many of the profitable players I've talked to start from a trusted projection source and add their own adjustments on top.

How long before I know if I am a winning DFS player? Longer than a season, and the dollar figure is the wrong measurement. Look at your distribution of finishes rather than your realized winnings, because tournament outcomes are lumpy enough to make a break-even player look great or terrible for months.

Which sport is the hardest for DFS? NASCAR, in my experience. There are too few drivers and too few meaningful stats to build a comprehensive strategy around, which is exactly why I don't put serious volume there.

Do you ever exclude players from your pool by hand? Rarely. Projections plus the ownership read filter almost everything on their own. The one exception I made was Tiger Woods, because I knew my own opinion of him always ran well ahead of his actual skill level.