How To Make DFS Projections

Accurate player projections were once the Holy Grail of daily fantasy sports. They were the crux of success and the primary concern of every top-ranked player. Used alongside game theory and lineup diversification, good projections still improve your decisions and lineup quality over a large sample. But the DFS landscape has changed. Now that you can get elite projections for an affordable price, you can win without spending hours a day buried in a sport and grinding on Excel spreadsheets. Plenty of top players do not build their own projections for every sport they play, whether that is MLB, NBA, NFL, NHL, PGA, or UFC. I know that because I can usually predict their cash lineups night after night. So how do you make your own DFS projections? Let's tackle the topic for the current season and beyond, but I will tell you up front that the best answer might be that you shouldn't.

The Short Version (TL;DR)

• You can absolutely build your own DFS projections, but for most players it is no longer the highest-value use of your time. Proven projections are cheap and the edge has moved into lineup construction.
• The fastest way to start a model is to let Vegas do the heavy lifting. Convert the game total and spread into implied team totals, then distribute that team total across the roster.
• Layer in advanced stats carefully. They add data points (swinging strikes vs. strikeouts, for example), but only if you know exactly how each stat is defined.
• Cross-check your numbers against sportsbook player props before lock, but trust team totals and moneylines more than props.
• The real edge today is what you do with projections, not just owning them: ownership, correlation, leverage, and late swap. That is the work Stokastic Sims is built to do for you.

Watch The Video

We walked through this exact build-your-own process on our show How To Make Your Own NBA Projections — customizing projection inputs, balancing your own numbers against market-based ones, and turning the result into playable edges. If you would rather see the workflow than read it, start here.

Watch on YouTube

How To Make DFS Projections From Scratch

When I started playing DFS, accurate projections were not readily available, so I made a good return simply by building my own. I realized within days how limited the public projection sources were. I would see hitters in a Coors Field game projected as bad plays while the sportsbooks had that same game as the highest-scoring slate on the board. There are still limitations to projection sources, but the commercialized ones have gotten pretty good. Building projections was once an essential skill to crush DFS. It is no longer something you need to make your primary focus. That realization is a big part of why I started Awesemo.com (now Stokastic.com): to give players proven projections so they could spend their time on the parts of the game that actually move the needle.

The safest route for most people is to start from a proven projection set, then add your own analysis on top. When thousands of people have access to the same projections, it is wishful thinking to expect that picking the highest projected lineup will print money. You need to differentiate from the field. Now that everyone is sharp at finding the best plays, you have two real options: hunt for errors in the industry consensus, or lean on other edges in lineup construction.

The most obvious ways to build lineups beyond raw projections are to account for ownership and correlation, depending on the sport. Because most GPP lineups are built by optimizers, it is easy for people to stack their rosters with strong values, but it is hard for them to weigh multiple factors at once. As a result, the highest projected lineups are usually suboptimal in tournaments, barring unusual circumstances. It is better to find lineups that are strong across all factors. The ones optimized purely on projection tend to end up too heavily owned. That is exactly the trade-off the Stokastic Ownership projections and Top Stacks tools are built to surface, so you can see where the field is piling in before lock.

Aggregating DFS Projections

Averaging multiple projection sources, also called aggregating projections, is a popular DFS strategy. Aggregating can help because if you build and evaluate a large pile of lineups from one single projection set, you introduce hidden biases into your process. The most common one: if you cap how much exposure any player can get, lineup one grabs all the best plays, lineup two grabs the second tier, and so on. That cycle repeats indefinitely, so your preferred plays clump together and those lineups all look like the best you generated. Using different projections to optimize and select lineups mostly solves this.

There is also a theory that averaging many opinions produces a stronger consensus prediction than any single one. I am a bit more skeptical of that, because it is subject to its own biases. In NBA, where projections carry more weight than in other sports, I like to project a questionable player as the midpoint of both outcomes. If you only grade the players who end up active, that makes my projections look worse than others. If you grade every player in the pool, they look better. Same projection, opposite scorecard, depending on how you measure.

The biggest limitation of any projection is that a single number cannot capture multiple scenarios. If a player is questionable, that one number is really a blend of very different outcomes: he could sit and post a zero, play limited minutes, or fill his normal role. How I handle that depends on whether the site allows late swap. On a late-swap slate, I roster a questionable player at his active projection, because if he ends up ruled out I can swap to the replacement-value play before his game locks, as long as I have left myself viable swap paths. The edge is that swap option itself, not optimism, and I never put two conflicting scenarios into the same lineup. Ideally you update each lineup by hand, but when time is tight I lean on the Late Swap feature inside Stokastic Sims to reoptimize the whole slate in seconds.

On sites without late swap, I like playing the average of both scenarios, because then I get some exposure to the outcomes where I guess the injury right. I prefer the player who benefits from a teammate's absence, because even in the worst case you still have a shot. Locking in a questionable player and hoping they go off is riskier. It badly cuts your odds of cashing the half of the time they sit. The trap is that most players are overconfident in predicting these spots, and the whole industry leans toward assuming questionable guys will play.

Using Sports Betting Markets To Build Your Model

There is a real barrier to entry to building your own model, because your model has to clear a certain accuracy threshold before it adds anything to your process. The ceiling is high, though, because the most accurate projections carry real value. The easiest way to start a model is to let Vegas work for you. Sportsbooks are a vital information source, especially for sports you do not know cold. Because the books are willing to take real money on set odds, you can infer a lot from the markets they offer. Pulling figures like team totals into your fantasy model is a great way to simplify the problem you are trying to solve. These odds carry some biases, but in high-liquidity markets like game totals and moneylines, they are very useful.

The process of setting a game total or a moneyline is called oddsmaking, or line-setting. Sportsbooks and their market-making traders open numbers using models, trader input, and comparable markets, but the true accuracy of Vegas does not come from a single model or even the collective effort of the books. Vegas owes its accuracy to sophisticated bettors who hunt inefficiencies in the markets and attack them. They put serious money on the lines, which balances out a public that leans heavily to one side.

Books set and move lines to balance pricing accuracy, hold, and liability against market demand, so a flood of money on one side moves the number for the next bettor. The more money they allow on each market, the more efficient the line becomes as professionals make up a bigger share of the action. The accuracy holds up as information emerges, in the form of injury news or weather. That is the single greatest benefit of using betting lines: the wealth of information already baked into one number, which means fewer factors you have to adjust for yourself. To get the full benefit, update these numbers close to game time, when they are sharpest.

Turning Totals And Spreads Into Implied Team Totals

Because the margin of error at any given moment is small, these markets are reliable inputs. Take a real NBA example: if a game has an over-under of 230 and a spread of 4 favoring the home team, you can deduce that the away team is expected to score 113 and the home team 117. Those are the implied team totals. In spread sports like the NBA and NFL, the math is clean: split the total in half, then add half the spread to the favorite and subtract half from the underdog.

Sport	Game total	Spread (favorite)	Favorite implied total	Underdog implied total
NBA	230	-4	117	113
NFL	44.5	-7	25.75	18.75

The NFL row is the one I keep coming back to when I explain this, because that 7-point gap turns a 44.5 total into a 25.75-vs-18.75 split, so the favorite is projected for roughly 37 percent more scoring than the underdog. That is exactly the kind of edge you want flowing into your quarterback and skill-position projections before you ever open a spreadsheet. From there you make educated guesses about pace and other game-level stats, then distribute that team total across the roster using each player's role and usage.

One important caveat: this half-spread shortcut only works where the spread tracks the expected margin. In low-scoring sports like baseball and hockey, a standardized -1.5 run line or puck line is not a 1.5-run expected margin, so do not split it the same way. There you derive the implied team total from the moneyline win probability and the game total instead, which is the more honest read of the market anyway.

It also matters which way the odds lean on the total. Adjust your number toward whichever side carries the more negative price. In NHL, if the total is 5.5 with the over priced at -125 and the under at +105, that pricing implies the market expectation is shaded above 5.5. You can estimate how far by comparing different books and alternate totals over time, since they do not always post the line at the same number. Our DataHub for each sport pulls these market figures and our projections into one place, so you are not stitching together odds screens by hand.

Applying Advanced Statistics

Once you estimate the game-level stats, you compare them to a team's average performance and adjust each player's expected output accordingly. The baseline for any player projection should be their historical stats. Traditional per-game or per-minute numbers work fine, but to find an edge you want to sprinkle in advanced or non-traditional stats. Some have stronger predictive value than the raw box-score events. Others are worthless. Advanced stats are a great way to add data points to your sample. Swinging strikes versus strikeouts is a good example: there are far more pitches than at-bats in a game, so you get more information to work with.

There is a sea of advanced stats out there, but the most important thing is knowing exactly how each one is defined, because many do not mean what you would assume. Take usage rate in basketball. You would think it simply counts how often a player has the ball, but it is really an estimate of the share of his team's possessions he ends with a shot, a free-throw trip, or a turnover. Interesting to talk about, easy to misread if you assume it means raw touches. Worse, some sites calculate the same stat differently. Pace of play in NBA has several definitions depending on whom you ask. Whenever you use a stat, make sure the definition is consistent across every source you pull it from.

Finally, once you have your projections, check them against other sources. Even the best-prepared DFS players miss information that swings a projection. Your best reference point is sportsbook player props. Now that sports betting is legal in regulated U.S. markets where available, books like DraftKings and FanDuel offer plenty of these, partly because they appeal to DFS players. They cover things like passing yards for a quarterback or points for an NBA player. The numbers are informative, because the book would not offer them if they were easy to beat. Still, do not give props the same respect as team totals or moneylines. The prop market is less efficient: lines are not as widely posted, books cap the dollar amount per bet, and they limit sharp users. So a prop does not carry the accuracy a game total does.

Want elite DFS projections without building a model? Stokastic Sims runs on our own projections and ownership, then simulates each contest across many thousands of outcomes to find the lineups with the best win probability, correlation and leverage built in. New users get to try the DFS sims for free, and code STOK10 takes 10% off your first Stokastic Sims payment if you subscribe. Start with the free DFS Sims.

Why Owning Projections Is No Longer The Whole Edge

The dream in DFS is having the most accurate projections all to yourself, because then you know roughly who the field is on while everyone else only knows the consensus. If you are confident about why your projections differ from the crowd, you can capitalize on the so-called wisdom of the crowd. That is theoretically possible, but in practice it is a pipe dream, and it probably is not the most lucrative way to spend your time.

DFS players want to outsmart the field by picking better players than everyone else. But as more people gain access to accurate information, the game changes. It turns into a poker game, where knowing what your opponents will do matters as much as the strength of your own predictions. That is why DFS is in such a fun place right now. Every contest you enter has different dynamics based on who is in it, the size of the field, and the format. Put real thought into the best strategy for each tournament and you gain a big edge on the players who are not paying attention.

This is exactly where I point people toward our DFS Sims. Rather than optimizing for one projected score, the Sims build your whole player pool, simulate each contest across many thousands of outcomes, and surface the lineups with the best win probability. They factor in ownership, correlation, and leverage at the same time, which is the part humans struggle to juggle by hand. For GPPs that is the work that wins. For cash games, where you are just trying to beat roughly half the field, I keep it simpler and build the highest-floor lineup straight off projections rather than the simulated tournament pool. The Sims pool is a tournament tool, not a cash tool. You can dig deeper into that split in our guides on DFS ownership and leverage, cash vs. GPP lineups, and how to make NFL DFS projections. Running the Sims is a few minutes of work per day rather than the hours of spreadsheet building I used to put in.

All of these inputs live on the same set of tool surfaces, so you are never stitching a model together by hand: projections, ownership, and Top Stacks sit in the DataHub for each sport, the contest simulations run in the DFS Sims, and if you would rather see the finished output than build it, our today's MLB DFS picks hub shows the plays those same numbers point to. When you are ready to put your own money behind a slate, code STOK10 takes 10% off your first Stokastic Sims payment.

How To Make DFS Projections: A Simple Step-by-Step

1. Pull the market. Grab the game total and spread for each game on the slate. In a 230-total NBA game with the home team favored by 4, the implied team totals are 117 home and 113 away.
2. Distribute the team total. Spread that team total across the roster using each player's usage and role, then convert expected production into fantasy points.
3. Set the baseline from history. Anchor every player to their historical per-minute or per-game stats, then adjust toward the game-level estimate.
4. Sprinkle in advanced stats. Add a few non-traditional metrics you actually understand, like swinging strikes for pitchers, and make sure each definition is consistent across sources.
5. Cross-check against props. Compare your numbers to sportsbook player props, weighting team totals and moneylines more heavily than the thinner prop market.
6. Decide build vs. buy. If your model does not clearly beat an established projection set, lean on that set and put your time into ownership, correlation, and leverage instead. The mistake I see most often is a player pouring hours into a home-grown number that trails the market, then still getting beaten on ownership by someone who bought the same projection and spent that time on leverage.
7. Late swap. Update for news before lock. A ruled-out player or a batting-order change can change everything, so use Late Swap inside Stokastic Sims to reoptimize once the lineups post.

Prefer to tail a proven card instead of building your own? The value-first habit above still applies; you just outsource the legwork. DFSnDonuts Picks on Tails is a DFS-native expert (#8 on the MLB Hot Streak leaderboard as of July 16) whose plays come with the receipt every Tails expert carries: a full graded track record you can check before you tail. Start with the free expert picks feed, where free picks post every day.

Frequently Asked Questions

How Do You Make DFS Projections?

Start with the betting market. Convert each game's total and spread into implied team totals, distribute that team total across the roster by usage, anchor every player to their historical stats, then adjust with a few advanced stats you understand. Finally, cross-check your numbers against sportsbook player props before lock.

Should I Build My Own DFS Projections Or Buy Them?

For most players, buying proven projections is the better use of time. Building a model only adds value if it clearly beats a commercialized projection set, and that is a high bar. The edge has largely moved from owning projections into how you use them: ownership, correlation, leverage, and late swap. That is the work Stokastic Sims handle for you.

What Are Implied Team Totals And Why Do They Matter?

An implied team total is how many points the market expects a team to score, derived from the game total and the spread. In a 230-total NBA game with the home team favored by 4, the home team's implied total is 117 and the away team's is 113. They matter because the market has already digested injuries, weather, and pace into one efficient number, which simplifies your projection.

Are Sportsbook Player Props Good For DFS Projections?

They are a useful cross-check, but not as reliable as team totals or moneylines. The prop market is less efficient, since lines are not as widely posted, books cap bet sizes, and they limit sharp bettors. Use props to catch information you missed, then trust the more liquid markets more.

Can I Make DFS Projections Without A Spreadsheet?

Yes. Tools like the Stokastic Projections, Ownership projections, and Sims give you proven inputs and simulate contests for you, so you can skip the spreadsheet grind and focus on lineup construction, leverage, and late swap.

Start Building Better Lineups With Stokastic Sims

You can spend years getting good at building DFS projections from scratch. I did, and it is a lot of fun. But for almost everyone, the smarter move is to start from established projections and spend your time on the parts of the game that separate winners from the field. New to Stokastic? Stokastic Sims gives you our DFS Projections, Ownership projections, Top Stacks, and the Sims that simulate each contest across many thousands of outcomes to find your highest win-probability lineups, with correlation and leverage built in. You can try the DFS Sims for free first, and code STOK10 takes 10% off your first payment if you subscribe. See Stokastic Sims pricing and start free.

DFS is high-variance. No process, projection set, or tool can promise a profit, and even the best pre-lock lineup can finish near the bottom on a given night. The goal is to give yourself the best long-run edge and let the process work over a large sample. Play within your bankroll, and only with money you can afford to risk.