NBA DFS Projections Are BROKEN (We Fixed It)

If you're building your NBA DFS lineups this year with average projections, ownership projections, and hand builds or an optimizer, you are leaving money on the table. My name is Jordan Chan. I'm the head coach at SaberSim. And in this video, I'll dissect exactly why NBA DFS projections are misleading.

And how using them to build your lineups will cost you money this season. As a DFS coach, I've helped thousands of players improve their NBA DFS results over the years. And at SaberSim, our tools have produced eight different million dollar wins in the past year alone. So if you're playing NBA DFS this season, you need to hear this.

And make sure to stick around until the end of the video, where I'll unveil your NBA DFS secret weapon for this season. That can turn the limitations of NBA DFS projections into your biggest edge. Now to unlock the high scores you need in your NBA lineups to win, you need to understand what upside looks like for the different games on the slate.

And to understand this upside, we need to understand the NBA games. Rotations, matchups, player skill, coach tendencies, recent player form, foul trouble, and dozens of the other factors that end up going into an NBA game. Each time you decide to put a player in your NBA lineup, you're asking the question, what's the ceiling for this player and how often do they get there?

But here's the problem. Average projections are a single number. representing the mean of all possible outcomes, and they assume that players score that average 100 percent of the time. And they don't. NBA players have a range of possible outcomes. The most common outcomes are around the average, but that is certainly not all that can happen.

NBA players have downside, and more importantly upside, that isn't captured in the average projection alone. Yannis might be the top projected player on the slate with a 55 point projection, but how often does he score the 70 or 75 you need from him to take down a tournament? How often does Tatum, Durant, or Luka outscore him?

And to make matters worse, projections assume players are completely independent from each other, which again, they are not. Players playing in the same game affect each other's projections. performance and their results are correlated to each other. You're probably already used to the idea of correlation from sports like NFL or MLB.

And while LeBron James isn't going to be throwing a touchdown to Anthony Davis on the basketball court, these correlations do still matter in NBA, especially in the game outcomes where one player is having that big upside game. If LeBron is going off, it's going to be harder for Anthony Davis to achieve a huge game right alongside him.

After all, there's so many points, assists and rebounds to be had in an NBA game. I'm But the average projections alone are going to assume that these players upsides are completely unrelated to each other. Yet, there is a balance here. Just making a rule that says you'll never play LeBron and AD together in the same lineup is oversimplifying that problem.

What if both players are severely underpriced? What if they're playing in by far the highest total game of the night, and what if all the value on the slate is at the guard position and LeBron and AD make up the best two pay up options at forward and center? What I'm getting at here is the importance of understanding and building NBA DFS lineups based around NBA game scripts.

What do the individual games look like when the players in your lineup hit their ceilings? How many points do players score in those upside games, and how often do they get them? Many DFS players assume since because NBA players have a bell curve distribution, and that because correlations are more subtle than in other sports, that these things do not matter at all, and that is patently false, and building lineups based on these averages will cost you money.

NBA DFS player projections are broken because they assume that the average projection is the only outcome, and that these player correlations do not exist. But building lineups based on game scripts isn't all it takes to win in NBA DFS is a peer to peer game, and we need to beat the other lineups in our contests, not just play 8 players that all outscore their projection.

And you probably already know that high ownership is a bad thing for tournaments. The more lineups that roster a player, the less that player's points matter. If 100 percent of the field plays a player, his points literally do not matter. But NBA DFS is unique. It's the one sport where you'll get agreement from virtually everyone, that a player can sometimes be a smash play regardless of ownership.

Players that are highly owned are often going to be owned for good reason. Take for example the min priced backup point guard that's going to start and play 30 minutes. Do you really want to fade that guy? Yet we still need to find leverage somewhere. To finish in the top 1 percent of large field tournament contests, our lineups need to stand out from the others in those contests.

But the NBA DFS ownership projections most players use to navigate this problem are broken because they operate at the player level, but DFS is played at the lineup level. Saying Jason Tatum is a good play at a 50 point projection and a 20 percent ownership projection makes no sense in a vacuum. There are good Jason Tatum lineups that have the right amount of leverage in the rest of the lineup to stand out from the field.

And there are bad Jason Tatum lineups that don't. What we need to do to be successful is project the kinds of lineups our opponents will play, and then play the lineups best designed with the right amount of upside and ownership to beat them. But even if you solve these problems, NBA DFS is primarily a speed based game.

That does not stop at Locke. In fact, after Locke is often when the real game actually starts. High impact, slate changing news is often going to break within the last 15 minutes before each game starts. And if you're not reacting to this news with Lateswap, You are not going to win. The most important factors for a player having high upside in NBA are minutes and usage.

And a starter getting ruled out right before the game might make the backup that is now starting the best play on the whole slate because of the minutes and usage that player now soaks up. But these lineups will come out late, sometimes 15 minutes before the next game starts, which means the slate context can literally completely change you.

every time a new round of games is starting. And we're not only getting new information about the games yet to start, but we're also getting new information about the games in progress. Remember our Jace and Tatum lineups from before? Well, imagine he was expected to be 20 percent owned, and then he actually comes in at 8 percent owned.

And if he was projected for 50 points, well, what if he has 25 in the first quarter? That changes the ranges of outcomes of the lineups that he's in significantly, and also our late swaps should adjust for that information too. The inefficiencies of both average projections and ownership projections that I talked about are magnified by how often the slate changes, and how little time you have to react to these changes before the next game's lock.

Let's imagine the typical optimizer bro navigating an NBA DFS slate. The final starting lineups for the first round of games come out 15 to 20 minutes before lock. In that final 15 minutes, he has to put together all the nuanced rules to account for the correlations between players in games. He needs to set randomness settings, or something else entirely, to account for the fact that players are not going to score their average projection 100 percent of the time.

And he needs to spend a bunch of time crafting each lineup to make sure it correctly balances upside and ownership. 15 minutes before lock, the Sixers scratch Joel Embiid, and Andre Drummond is going to start at 3. 5k salary, is expected to play 30 minutes, with huge fantasy point per minute numbers. Now what?

Either this player doesn't react at all, or runs out of time trying to, and misses out on the best play of the slate. Or he has to redo a ton of work, accounting for the new correlations and game scripts in the Sixers game, but also making sure that none of the previous work he did on other games interferes with his ability to get the right drum and lineups in.

And all this has to happen 15 minutes or less before the game actually starts. And then 20 minutes after lock, 10 minutes before their game, the Charlotte Hornets decide to let us know that LaMelo Ball will not be playing basketball tonight. Now he's got to do it all over again. Plus. He might already now have a hundred plus different lineups in play in a variety of different contests where he now knows how some of his players are performing in the first quarter, the actual ownerships for a bunch of players in those games, and ideally would want to swap those lineups accordingly based on that new information as well.

This pattern would continue throughout the entire slate, and this might sound extreme, but I promise you there will be NBA slates this year that play out exactly like this. It is incredibly hard to make the optimal decision accounting for all these important variables in the 10 15 minutes that NBA DFS gives you to do it.

So, enter Saberson. Your NBA DFS is secret weapon, the tool you need to appropriately balance upside ownership and time in your lineup building process. This year, average projections can't account for upside and correlations on their own. And this is why Sims are valuable on Saber Sim. Your NBA lineups are built with simulations.

We simulate each game on the slate play by play a hundred thousand times accounting for all the important factors that matter in the NBA. And each lineup you build on Saber Sim is the highest scoring lineup. for a set of simulations of the way that the games on the slate might play out. Because of that, these lineups answer all the complicated questions of NBA DFS naturally.

How much does correlation matter on that particular slate? How often should you roster the backup point guard that's thrown into the starting lineup? Should you pay up for studs in games with high blowout risk? The sims will help identify the answer to these questions by simply seeing how often these players or lineup constructions are the highest scoring lineup.

And our contest sims will solve the problem of ownership projections accurately and precisely. On Sabersim Ultimate, we generate an expected field for 13 different contest types in the lobby to project exactly what kinds of lineups that you'll be facing off against in different entry limits and stakes.

And after you build your lineups on Sabersim based on our NBA game sims, We'll pit those lineups against these fields and simulate out your DFS contests a hundred thousand times, identifying the lineups that have the right amount of leverage and upside to be profitable against the opponents that you're actually playing.

So while your opponents battle their optimizers using ownership, some rules, or just gut feel about how much ownership is good or bad. You're bringing mathematically. Optimal lineups to the contest. And most importantly, we make it easier than any other tool to adjust a new information in the limited time you have before the next game.

Start as starting lineups come out for each game. We'll rerun our simulations of that game based on that new information. generating new play by play game simulations and new expected field lineups in just a couple minutes after news breaks. And as the games in progress are playing out in the court, we run a new simulation for those games from the current point that that game is in every five minutes to adjust player projections to new live projections.

So if it looks like Victor Wembenyama is a must have player after a half of basketball, our live sims will pick up on that. Showing you how to maximize your profit for your lineups that have him and also how to save your lineups that don't and as these basketball games lock in your DFS contests, we're also updating our field lineups that you run your contest Sims against.

So if when Bonyama comes in at 20 percent owned instead of 40 percent owned. We'll adjust for that, so you're contest simming your lineups against field lineups that increasingly resemble the actual lineups your opponents are playing in your contests. It can sound like a lot, so in practice, here's what it actually looks like.

Every time an NBA game starts, we intake the real, actual ownerships in the contest you're playing, and then readjust the field lineups we use in the contest sim. Every five minutes that passes in the NBA game, we run new live simulations from that point in the game on and use those new live simulations in your contest sims to account for how players are playing on the court.

Every time we get a new starting lineup for a team whose game has not started, we adjust our game simulations for those games plus the field lineups to react to the new information we have given those starting lineups. And when you go to late swap on SaberSim, We swap each lineup that you're already entered a few dozen times.

So you'll have say 20 or so different swap options for your lineups based on these updated simulations. We pass each of these swap options through the contest sim. that is accounting for that live data to identify the most profitable swap for each original lineup you're playing. With this strategy, you're beating average projections by accounting for the real upside in correlation by using NBA play by play game simulations.

You're beating ownership projections by seeing what lineups you're actually competing against and finding the swaps and lineups that best exploit them. and you're doing it all in the short amount of time that the MBA schedule tends to lend you. In other words, you are maximizing your MBA DFS edge at every single opportunity you have.

Now, in another video released before MBA actually starts, I'll walk through the steps on SaberSim, step by step, so you can see exactly what it looks like and learn how to do it yourself. But I wanted to first make sure you don't waste your MBA season trying to build lineups with average projections, ownership, and an optimizer.

These tools don't account for the full range of outcomes of players. They don't account for correlation. They don't exploit the field correctly. And they don't make it easy to account for all the new information that you have coming in during a live NBA slate. We've reset everyone's trial for the start of an NBA DFS season here at SaberSim.

So whether you've tried us before or not, take advantage of this opportunity for the start of the NBA season with a free five day trial on our site, SaberSim. com. In the meantime, good luck, and I'll see you in the next video.

‍