DFS Lineup Optimizers Are Obsolete. You Need a Simulator.

The traditional DFS lineup optimizer is broken and simulators are the only solution. Discover how the right daily fantasy sports simulator can be your DFS secret weapon. Use it to build the BEST DFS lineup and give yourself a better chance at winning.

Transcript

Read Transcript:

I'm going to cut right to the chase. The traditional DFS tools from average projections to optimizers are outdated and they're costing you time and money by forcing you to do unnecessary busywork and make expensive mistakes. We built SaberSim to change all of that by giving you the only tools built from the ground up to beat today's tough games.

And all of it is made possible by a one of a kind complete game simulator. And I'm going to talk a little bit more about this in a minute, but I want to stress that not all sims are created equal. And while more and more tests are using them practically as marketing gimmicks, we have the only complete game simulator.

And that means that we're simming out a full game, one play at a time, keeping track of the score, the clock, and dozens of other factors that influence play, calling and performance, and no one else does this. My name is Andy Baldacci. I'm the CEO of Saversim. And over the next 30 minutes or so, you're going to learn what makes our complete game simulator and optimizer your DFS secret weapon.

Whether you're playing DFS to make some extra cash on the side and balancing with your job, your family, and everything else you have going on, or you're already crushing the game and have all the time in the world, you need to hear this because I guarantee what I'm covering today is going to change how you think about DFS and help you win more money at the game.

So let's just jump right into it. If you want to beat DFS today, you need to build high upside lineups. And that means lineups that have more potential than their average projection suggests. And you do that by leveraging correlations, ownership, and a player's true variance or their range of outcomes.

Traditional tools cannot do that when I'm talking about traditional tools. What I mean is, I mean, average projections in old school lineup optimizers. And before you take out the pitchforks, it's not that the tools don't work. It's that the job they're designed to do isn't the job you actually need them to do, which is why you have to spend hours programming dozens of settings and rules.

If you want lineups with any chance of winning. Let me start with the average projections. Projections, as you know, are the foundation of any DFS process. The trouble is that we take those projections for granted, and we don't actually think about what they're telling us. The projections that you see everywhere are simply the scores someone says each player is going to get on average.

And this seems reasonable at first, but to maximize upside, you don't care about a player's average performance, you care about their top performance and how often they get there. And averages do a horrible job of showing you that because they don't represent the reality of sports. Averages are useful when a range of outcomes is what's called normally distributed.

And that just means the range of outcomes follows a bell curve. This is what a bell curve looks like. Basically what it's saying is that the average or the mean is also the most frequent outcome, which is also known as the mode, and also the middle outcome, which is also known as the median. It also says that the outcomes the same amount above and below the average occur with the same frequency, but that's just not how sports work in reality.

And let me jump into Sabrestone quickly to show you how this looks using our simulation data. So I'm going to pick George Kittle as the example here. Um, what you see on this chart that we're bringing up is, um, this is all based on the thousands of data points from our simulations, and it shows how often Kittle is going to score a range of points.

So on the bottom here, we have fantasy points on the left. We have the frequency. So what this is saying is. And I highlight any of these right around 10, so there's probably like 7 to 10, 8 to 10. It's not an exact science the way we have this shown. What it's just trying to do is give you a good look at all of those thousands of data points to see what a player's range of outcomes looks like, what their variance looks like.

And so according to average projections, a tight end with a projection of 15 points, which is about what Kittle has, is as likely to score 20 points as they are to score 10 points. Both are 5 apart from the average. But that's not true. So he's going to get 20 points around 8 percent of the time. He's gonna get 10 points.

Closer to 14 percent of time, 13, 14 percent of the time. So that's honestly about a 50 percent difference. I'll actually a little more. Um, and what you'll see with tight ends across the board is that they're going to score below their average more than half the time. And this is a big oversight on its own, but the most dangerous part of the equation is that averages dramatically underestimate a player's upside while overestimating their downside.

So let's go a little further here. Kittle is going to get 30 points. About 3 percent of the time, he's going to get 0 to 2, about 2 percent of the time. So he's getting 30 points, at least 50 percent more than he's going to get a true zero. Um, but to go even further, he's basically never going to get negative points.

Um, but there are a lot of scenarios where he's going to get more than 30 points. And so if the average is 15. Say, um, what is going to be 20 points higher than that? 35 20 percent 20 points lower, negative five, that negative five outcome almost never happens, but he's going to get 35 points or more, a real part of the outcomes.

And. The averages completely ignore this, and this might seem kind of nuanced. It might seem like a small mistake, but these mistakes happen with every single player, and they add up to become expensive errors when they're all combined in your lineups with how tough it's getting to beat DFS. You simply can't afford to make mistakes like that.

If you want to win, you need data that gives you a true look at what can happen in a given game. An average projection simply cannot give you that. Now let's talk about the other flawed DFS tool. lineup optimizers, and they have three massive blind spots that you need to be aware of. The first is that traditional optimizers by definition were literally built to optimize exclusively on one thing, and that is average projected points, completely ignoring a player's variance.

Even if you did have projections that reflected a true range of outcomes, Optimizers can't handle that because they only use a single data point for each player. And I've already shown how big of a mistake that is. Even if traditional optimizers so could handle this variance, these range of outcome projections, that's only one part of the problem because their second blind spot is that they ignore correlations.

And correlations, it's just a fancy word for saying the measurement of how each player's Performance impacts the other players in that game. And this is what drives the value of stacking. So when you build lineups with stacks, which really just means rostering positively correlated players, like a quarterback and his receiver or batters on the same team, you're capitalizing on the fact that when one of those players does well, it's very likely that the others in that stack are going to do well as well, giving your lineup additional upside that average projections alone would ignore.

The last one is again, if this already wasn't enough, this last point is that traditional optimizers ignore ownership and we're talking about ownership. We're referring to the percentage of lineups in a contest that roster a specific player. If the goal of DFS was just to be a specific score, then you wouldn't care who else had your players in their lineups.

But in DFS, you're not playing against the house. You're playing against everyone else in your contests. And to win, you need to beat all of their scores as well. Now, this doesn't just mean you should blindly ignore the most popular players, but on the extreme end of things, if you think about this way, if a hundred percent of lineups in a contest have the same plays as you, then it doesn't do you any good when they go off because all of your opponents benefit as well.

But say that you faded some 50 percent owned player and they had a bad night. That means half the field is now at a huge disadvantage to you. With traditional optimizers though, they don't look at this. So it's up to you to determine who's good shock and who is bad shock. To maximize upside, which really is what we were saying before, Is what you need to do.

If you want to beat DFS today, you need to separate yourself from the field and you do that by leveraging correlations ownership in a player's true variance. Traditional optimizers cannot do any of this on their own, and that's why you have to program dozens of rules and settings and groups to get lineups with decent stacks and ownership plays and can that approach work?

Of course it can. There are tons of players winning with those traditional tools, and I'm not trying to pretend that it is impossible to win with them because again, it happens every day, but to win with those traditional tools, it takes a ton of time if you want to do it well. And even then you simply cannot create enough rules to account for all of the edge cases in a game.

So you're left with rules of thumb that give up a ton of value because they don't apply in every situation, but you are applying them, your optimizer is applying them across the board. So these are rules like all lineups must have total ownership below X and no more than one player with ownership below Y and maybe no more than two players with ownership above Z.

The problem is this lacks all nuance because there are plenty of high owned players that are good plays and plenty of low owned ones that are bad and these blanket rules ignore all of that on the correlation side of it. In football, I know one of the common rules is all lineups must have a quarterback, two of their receivers, and one of the opposing receivers.

As a rule of thumb, this one's not bad, but it again, treats all stacks the same. All sex is having the same value when that's not true. And you lose so much nuance when you have these blanket rules, these rules of thumb, and you could create groups to try to work around this by saying, if I'm playing X, then play at least one of these players.

Or if I'm playing this player, don't play that one. But when you create all of those rules to cover every Practical combination of players. You're really not doing much better than just building a lattice by hand. Ultimately, all of these groups and rules and settings are trying to account for the fact that games are played in the real world, where players don't achieve their average results every single time, where their performance impacts other players, where ownership matters.

Software is supposed to get you better results in less time, and traditional optimizers completely fail at that because they don't have enough good data, so they force you to program exactly what you want, which leaves you wasting a ton of time obsessing over settings instead of dialing in your actual lineups when there weren't any other tools available.

You didn't have another option, but with SabreSim, that is no longer the case. SaberSim solves all of this by combining the only complete game simulator with the only optimizer that automatically handles correlation, ownership, and variance. By simulating every single game from start to finish, play by play, we get an accurate look at a player's true range of outcomes, their true variance, which you saw in the app with George Kittle.

And not only that, but we can precisely measure how players are correlated with one another in a specific game, Rather than try to make guesses based on historical averages alone, and you're probably asking, like, why haven't others done this? It's because to get Sims right, it is insanely hard when you're just using spreadsheets or whatever it may be to make your projections.

You can run some equations on historical data, make a few tweaks here or there. And honestly, you'll get some decent average projections, but by now, you know, those aren't worth much on their own to make our SIM so powerful. We're not just thinking about the players and their stats. We have to actually model the game itself.

And that means play calling. That means clock management. That means bullpen management rotations in so much more. Other sites are talking about Sims, but again, these are little more than marketing gimmicks. We've got a video up on YouTube that actually breaks down our MLB Sim and the work you put into it in recent months.

So go check that out if you're curious what goes into this. We're actually going to be putting another one out in the next week or two on our NFL Sim. But to give you an idea of some of the work that goes into it. Will is our full time data scientist. And yes, I, we have a full time data scientist. I don't know if any of those other companies do.

Um, but will is working on a new timeout model for football because called timeouts actually have a really huge ripple effect on a game. They affect the pace of play. They affect what plays are likely to be called after the timeout and the list goes on. So once he's done with this, what we're going to do, run some back tests on it.

And if it's an improvement. We're going to add it to our sim and then will is going to move on to the next item on our list. If it's not an improvement, he's going to keep working on it until it is, because we know that this seemingly. Uh, kind of inconsequential factor of a game, which I don't actually know if anyone else publicly is modeling this, we know it's important.

And so we're investing the research, the effort, the time to getting it to where we want it to be. And really no one else is doing this sort of work because instead of investing in more programmers and data scientists, they're hiring more touts and taking Hawaiian vacations if a site is talking about Sims, but they're not modeling the clock or timeouts or how play calling changes based on the score.

Those aren't real Sims. While they can be an improvement over average projections, they are not even in the same ballpark as our simulator. They're not playing the same game. And not only that, but if their optimizer can't horn handle correlations, ownership or range of outcome projections. They are missing the point entirely.

You need the data, but then you need a tool that can actually use that data. If you want to get an edge on the field with Sims, you, again, you need the Sims that can model the entire game from start to finish, and you need an optimizer that can actually use that data to build lineups. And no one else has that.

I'm obviously biased though. Um, but to, to end my little rant there, what I'm going to do is jump into the app and just show you how all of this comes together. So let's open this up here. And we've got a ton of videos on our YouTube channel that does a deep dive into all the different features and functionality we have here.

So like, I'm not going to try to walk through every nook and cranny of the app. What I want to do is just show you at a high level, how this works. The home screen doesn't really look much different from the traditional tools, but there's a lot of power here. That's very important to understand. So what you see.

is our average projections for teams and for players up top, you can actually edit the projected total for a team. And what we'll then do is filter our simulations to get a set that match the neutral that you put in and automatically update the projections for all of the players in that game. And that's just one of the cool things we're able to do with all of the simulation data is make it really easy for you to say, You guys are different from Vegas.

I'm trusting Vegas. I'm going to put in a new total here and then we'll adjust all the players in that game accordingly. It gives you a fast way to make. Really important adjustments and I did mention that yet. These are the average projections that you see because frankly, that is a good way to quickly compare players.

But as I showed before, when you click on a player, that's where you see their full range of outcomes. And you also see Their correlation to the other players in that game, and it's this data is this complete data that our optimizer is actually using to build lineups. So we show you the averages so you can get a quick look, but the optimizer pulls in from the range of outcomes.

And over here, you'll also see our percentile outcomes, which can be a quick way to just get a feeling for players ceiling and floor. So 25th percentile, that just means it's a score that the player is going to exceed 75 percent of the time, and that's generally thought of as their floor, while the 95th is the score they're going to exceed 5 percent of the time, and that's generally thought of as their ceiling.

So other sites I know may talk about percentiles, um, having looked at them, I think there's some spots that are good at some that they miss, but without that complete game simulator. Oftentimes you're just not going to have nuanced enough data to make these be more than a guess when you're ready to build your lineups.

All you have to do is click this button right up here, and this is where you're going to see the standard settings. You're going to see minimax salary, the number of lineups you're looking for. But you also see our sliders, and this is where we're incorporating all of that simulation data. I'm going to enable manual mode to show you how this works, but I'll talk in a minute about how you don't have to worry about that.

So, correlation, this controls the impact of correlation on your lineups. A higher weight is going to favor players who are positively correlated, which again just means players who do well when the other does well. The higher your correlation slider, the better. The higher, the more enlarger your stacks are going to be because we are simulating every game play by play, we can precisely quantify just how correlated any two players are with one another rather than treating them all the same rather than putting in a blanket rule of saying, I want to have all quarterbacks with their wide receiver.

We can say, okay, like, yes, in most cases, that does make sense. But there are some quarterbacks and wide receivers that are much more highly correlated than others and others that are lower, and you don't have to build rules for each different team. You don't have to try to have different exposures to the ones that you like the most.

We will automatically do that for you because we have the data. So if you've ever been unsure of when to include pass, catching running backs in your stacks, what positions you should run back with in your NFL game stacks, or What the best batting or positions are for your MLB stacks. Honestly, the questions and the nuance, it really doesn't end.

But any of those things we just answer for you. We take out the guesswork because we precisely measure all of those relationships. Then we've got ownership fade, and this just controls how much of a factor you want ownership to be in your lineups. The higher you set it, the more your ownerships will fade, which is just another word for avoiding.

So you want The more you will fade the players with the highest ownership projection. And the opposite is true. The lower you go, uh, we don't actually target the higher on players. If you set it on the bottom end, but we just don't consider ownership as much of a weight or any weight, if you turn it all the way off, it's actually much more powerful than that because high variance, highly owned players are going to be faded more aggressively by the optimizer automatically than low variance.

Because if a highly owned player is low variance, which just means they consistently achieve their average projection or above, then you really don't want to fade them as much when compared to a player who has a similar projection, but doesn't reach that as often. They have much more variant outcomes, and we use our simulations to determine the variance of the individual players.

And the last one is sim variance, which is how we're able to build lineups using a range of possible outcomes using a player's variance. The way it works is that for every lineup we build you, we randomly choose a bucket of sims and optimize based on the average score for each player in that bucket.

Then we randomly choose another bucket for the next lineup and so on. What the slider does is it controls how large that bucket is. So if you were to turn this all the way off. What that would mean is that we're going to just have one bucket for all the sims, and we will use the average across that, which basically just saying we're going to use the average projection, and this works like a traditional optimizer, and it can be useful for cash games.

If you put some variants all the way up to 10, we'll go the other way. So what we'll do is each lineup. We'll have one simulation for it, and we will just use the score from that one simulation. Traditional optimizers usually have a randomness setting, and that gives you, uh, gives your lineups diversity, but it comes at a significant cost because you're literally just randomly adjusting projections.

A lot of work seemingly went into creating them, and so just adding noise to that. Doesn't seem like a great idea, but that's all they can do because they don't have a player's variance. They don't have that range of outcomes from a complete game simulator, which we do have. And like I said, there's a lot of power here, but again, our goal is to automate as much of this as possible.

So you can focus your time, adding value to the process rather than memorizing different settings and rules and all of that. So you don't have to mess around with these because what you'll do. is using these drop downs. You will just tell us what type of contest you're playing. Um, and we'll then automatically give you the best sliders for that contest.

And not just that contest. We'll also look at the size of the slate. So if you're on a 13 game slate versus a four game slate, even if you're playing the same contest, we will adjust the sliders because a small single entry contest is going to require different settings in the Millie maker. And all contests are going to require different settings on a two game slate.

Versus a 10 game slate and rather than force you to figure out for your own, all the best combinations for all the different scenarios, we've actually just gone in back, tested them all and set the defaults to the best ones. We do allow you to set stacking rules, which you can do just over here. Honestly, this is not necessary unless you know you want some extra control.

over the process. And the reason for that is because our sliders again are automatically taking into account correlation and variance in ownership when we're building a lineup. So you don't have to tell us what makes a good lineup. We already know. So we're automatically going to give you stacks because we have all of that data.

You can also do things like exclude the tight end and flex. But again, these are things that if you want the control, It's there, but you don't have to tell us every little thing to do because we're not going to put some random bum in the flex like a traditional optimizer may we are going to look at all the variables in build you lineups that follow the principles of building high upside lines, the principles of correlation.

Ownership and variance. This is what makes SaberSim so much better than the traditional tools. Our one of a kind complete game simulator and optimizer actually understand upside. And that means we're not just a dumb spreadsheet you have to fight. We're a trusted advisor that's going to point you in the right direction and save you a ton of time while getting you a better result.

I think of, honestly, the other tools kind of like Photoshop. Some might be a little bit more rough around the edges, might be more similar to an etch a sketch, but that's kind of beside the point. Photoshop can get the job done with a lot of work, but it's your job to tell it exactly what to do because it's not helping you at all.

It's not trying to give you advice because it doesn't understand what you're ultimately trying to create. With SaberSim, it's like working with an expert designer. You tell us what you want and we'll do our best to give it to you. This doesn't mean you're going to get a masterpiece from us that's going to make you rich with just a few clicks, because you still have to give us some feedback to dial it in.

You still have to do some work to make the lineups the exact way you want them, but you save a crazy amount of time and get a much better result in the end. Some people, they can make amazing art on an Etch A Sketch. But those people, frankly, are pretty weird and they have way too much time on their hands.

Why would you knowingly handicap yourself by wasting time with outdated tools? That aside, let's just jump in and look at these lines quickly. Um, again, you saw I didn't set any rules, um, anything at all like that. Let's click over here to go to the stack types, which will give us like the best look at how our lineups are shaping up.

So this is what SaberSim is giving us without putting any rules in at all. We're seeing, um, 10 percent of lineups have a QB plus 2 stack, 25 percent have QB plus 1, and then one has a bit of a bigger stack, the QB plus 3, with one on the opposing team. If you were to do a build like this without any rules, without changing any settings in a traditional optimizer, you wouldn't get anything close to this because all it's going to do is just jam in the players with the highest average projections while staying under the salary cap.

And I know I'm beating a dead horse on this point, but like, I just want you to see how much power is here. Did you get this? And do you get how much power there is in the fact that right out of the box, you're able to get the strong foundation. And it's not to say. You don't have to make any adjustments, but it's to say, all right, don't worry about messing with all the settings.

Don't worry about getting everything just right before you start. We'll give you the strong foundation and you can build from there. Even if you were to set a basic stack rule in a traditional optimizer, you would have to spend so much time dialing in the settings for all those specific teams and positions to get stacks to actually make sense.

We did all of this without you having to touch a thing other than telling us the contest type that you are playing. This completely changes the lineup building process. Instead of spending all of your time worrying about settings, you can quickly get a foundation of strong lineups and then invest your time.

Getting those lineups dialed in just the way you want instead of taking a dull ax to the problem. We're giving you a sharp scalpel so you can quickly make finely tuned adjustments. No one else gives you this power or control. And don't get me wrong. Again, I'm not trying to make this seem as though it's some money printing machine because it does still take work to win.

But instead of wasting all your time programming those rules that computer could do for you. And frankly, do better. You can invest your time building better lineups. If you're just starting out, our tools level the playing field and give you a shortcut to DFS success. Not only that, but to make sure that you get up to speed, we have daily office hours to answer your questions live.

And we have the only support team staffed exclusively by winning players because we are committed to making you a better player yourself. And if you're already a winning player, And you aren't taking advantage of the only complete game simulator and the only optimizer that can use that data, then you are leaving money on the table.

There's no way around it. And with the games getting harder every season, you simply can't afford to do that. If you were trying to compete on the pro tour for golf, you wouldn't be using wooden clubs. You wouldn't even use wooden clubs. If you're just trying to compete with your buddies on the weekend, if you actually want to make money at DFS, why would you use outdated tools?

Sims are the future of DFS and we are the only one doing them right by combining the complete game simulator. With an optimizer that leverages all of its variance, correlation and ownership data to show you how confident we are in our tools. We are also the only site out there with a free seven day trial, so you can make up your own mind to start that free trial, head over to Sabresim dot com.

You can click a button up in the right corner and you get started in under a minute. If you have any questions at all, you can always email me personally at Andy at savers. com or shoot our team a message at support at savers. com. And we are always happy to help with all that. Thank you. And good luck this season.

Unlock Expert Knowledge with SaberSchool

Master strategies, insights, and tools to outsmart the competition. Sign up for SaberSim Premium today!

Free for 5 days
Cancel any time
Everything included