How to Build Winning MLB DFS Lineups with SaberSim
Hey. I’m Andy Baldacci. I’m the CEO of SaberSim, and in this video, I’m going to show you everything you need to know to build winning lineups with our optimizer. We have built a one of a kind algorithm that simulates every single game, play by play, thousands of times. And this gives you the best projections and optimizer on the market. The simulations power absolutely everything that we do, but it also makes our three-step lineup building process a little bit different than the way the traditional, outdated optimizers work. But don’t worry. It is easy to understand, and once you give it a try, you’re not going to want to go back to the old way of doing things.
The first step is reviewing and adjusting projections. Second is choosing your build settings, and third is dialing in your exposures. We have a free course over on our website, under the resources section, that walks through each part of the process in a lot more detail than I’m going to cover right now, and I definitely recommend checking that out. But over the next few minutes, I’m going to share everything you need to know to at least get started and comfortable with the process.
The biggest thing I want to stress is that if you want to win money at DFS and you can’t spend all day building lineups, then you have to focus your limited time where you can add the most value. And with SaberSim and our three-step process, that’s going to be the first and last step of it. Our team of data scientists is constantly working to improve our models, but no model is ever going to be perfect, and that’s why it’s always important to review and adjust the projections.
And with this first step, there’s actually a couple of approaches you can take. If you are comfortable just working directly with projections, you know what to look for, then you can just review them right here on this screen, and make your adjustments without doing anything else. But if you’re like me and maybe you don’t really know where to even begin as far as narrowing things down, or you just don’t have the time to dig through an entire slate of players, you should start with something we call a test build, and see what impact these projections actually have on the lineups.
And so I’ll talk a little bit more about these settings in just a second, but for the test build, all we have to do is just pick something middle of the road, as far as the contest goes. And I like to build 150 lineups just to get a bigger sample, but this isn’t some exact science. What I’m trying to do right now is just quickly get a feel for where SaberSim is at on the slate, and narrow down the players and teams that I should focus on.
And this is really where the power of SaberSim shines, is because with a traditional optimizer, if you just went in and hit build, you wouldn’t actually learn anything from the result, because the lineups that they build wouldn’t be representative of anything that you or your opponents would enter into a contest. To get good lineups from a traditional optimizer, you have to set dozens of rules and create all kinds of groups. Otherwise, your lineups would just be dead in the water. But by running thousands of simulations, we get data that allows our optimizer to understand correlations and variants, and incorporate that in your lineups, along with ownership, so you can get a foundation of high upside lineups without spending hours programming exactly what you want.
And let’s jump over here into this stack tab, and I’ll show a little bit of what I’m talking about. I didn’t put in any rules or change any of the default settings, other than to tell the optimizer that I’m playing a GPP. And look at the stacks that I’m automatically getting. No other optimizer does this, because no other optimizer actually understands what it takes to win a contest, and helps you build lineups that do that. And that’s why you have to fight them to get the kind of lineups that you need if you want to actually win.
And so you still have to put in some work of your own if you want to win consistently, but because of our powerful data, SaberSim acts as your advisor. We’re not just a dumb spreadsheet. So instead of just forcing in whatever players you’re looking for and doing all of the work by yourself, let us help you by building you a strong foundation that’s in line with what you’re looking for.
That being said, let’s just jump in and talk about what we’re actually looking for in the test build. And the short of it is, you’re trying to find places where you aren’t as confident in SaberSim’s suggestions, and you want to do a little bit more digging. So these are typically the places where we’re going to be much higher or lower on a team or player than you’d expect. And if you don’t have strong expectations, you don’t have a strong opinion on a slate, that’s fine. What you should do is focus on the players and teams that simply we have a lot of, as well as players where we have significantly more exposure to them than the field does. So this is what you can check by looking at the projected ownership. That’s where we’re saying, “We think the field in an average contest, 18.7% of the lineups in the field will have Cory Seager in them.” You can look at that and see if you are way over or under the field.
And also, I like to look at the higher value players, especially in a sport like baseball, which is not like basketball, in that in basketball, you’re going to have some backup players that are actually going to start. They’re going to get a lot of minutes, and they are going to be a clear, obvious value play. That’s not even close to anything else out there. But for hitters in baseball, there’s just a ton of variants involved, and so you just always want to at least give those high value plays a second look, to make sure that you actually agree with why we’re assigning them that much value.
On DraftKings, to actually see your lineups, you do have to make two or more changes, but right now we’re not going to dig through all the lineups. That takes a ton of time and doesn’t give you the picture that we’re really looking for here. Right now we’re just getting a feel for the exposures.
So what I’ll usually do is jump into the team stacks first. And this shows me just the team stacks. It says what percent of our lineups has at least two players from each of these teams. And so right away, we are incredibly heavy on the Dodgers and the Rockies game. 90% of our lineups have a Dodger stack in it. 36% have a Colorado stack in it. I’m definitely going to look into that game.
And something else that jumped out at me, this isn’t a huge deal, and just the more kind of you get familiar with the sport, the more you’re following things, some teams might surprise you that you don’t see more of them or any of them at all. And so I just had expected to see a little bit more of San Diego. So I’m going to just look into that as well. And what I’m doing is just off screen, I’ve got a little Notepad file where I just keep track of a handful of players or teams that I want to look into. And the more time I would have in this build, the deeper I would go, and the more players and teams I would actually look into.
Now, let’s talk a little bit about the individual players. And so when I’m looking through this list, I’ll actually start with the pictures. So I’m looking to see where we’re out of line with the ownerships, and that’s not always a bad thing. Oftentimes you have to take some stance on players that are a bit more under the radar, or just more than the field, if you really want to stand out in these big contests. So it’s not inherently a bad thing, but that’s just where I’ll usually start looking. And in general, yeah, we’re getting a ton of Pablo Lopez, and he’s also really high value. So I’m just going to make a note that that’s a player I’d want to just be a little bit more cautious of when we have two times the field on him.
Going through, yeah, we have a bit more Heaney than the field, but it’s only 40% overall, nothing too crazy similar with Kelly. Yeah. The more time I have, the more I’m going to look into these players, but to keep things simple for this video, I’m just going to focus on Pablo Lopez. And let’s jump over at the batters.
So for this, like we’ve already talked about, we have a ton of Dodgers and Rockies stacks, so I’m not going to necessarily look into the players individually. I want to look at that in game on the whole, and make an opinion on that game. But what I’m going to do is sort by value and then just see, “Okay, who is jumping out at me?” And there’s a couple specifically, primarily on Seattle. So I see Jake Fraley, highest value. Not an insane amount, but at least twice the field. We’ve got Taylor Trammell. I’m just making a note of that one as well. And then let’s also check out … Yeah. Sam Hilliard, we have a lot more than the field, and Jon Burr too. Okay.
And so again, there are more players that I would look into if I had more time, but what I’m looking for is, okay, who are the players that would surprise me to be getting this much of? And I just want to do a little bit of work to make sure that I should be getting it. Because this is, again, where having SaberSim as an advisor to you is helpful, because we’re not just randomly trying to fit as many people in your alliance as we can. We are looking at the players’ upside. We’re looking at their range of outcomes. We’re looking at how they correlate to everyone else. So there are going to be a lot of under the radar plays that we put in there that you wouldn’t have naturally thought of.
You can kind of think of us as an architect. You’re building your house, you tell the architect what you want, and they come back and they’ve got a blueprint for you. And yeah, sure. You’re going to have to make a good amount of changes. Especially in that first round, you’re going to have to give feedback, have them throw out a few things, add a few other things. But along the way, just drawing on their expertise, that architect is going to think of some things that you wouldn’t have thought of on your own, and you want to make sure you don’t just immediately discount them. So that’s why I’m not just going in here and being like, “Oh, I don’t like this guy. I don’t like that guy.” And just mess with all the exposures. I’m really focused on giving the instructions to the architect to build the right blueprint for my lineups.
If really all you have is 15, 20 minutes, then your process is probably going to look pretty similar to what I’m doing right here. It’s finding those biggest outliers and focusing on them, because it all goes back to the 80-20 principle, where 80% of the outcome comes from 20% of the effort. Finding those biggest outliers is where you’re going to get the biggest reward for your research. If you have more time, you can do more research into some of the marginal plays, and this and that, but that just won’t move the needle nearly as much as those bigger outliers.
And I’m not going to walk through my exact research process for how I verify all this. We have a lot of videos on that subject. And one I would recommend for anyone watching this is a new one we put out, The Six Secrets to Beating Daily Fantasy Baseball. And that’s where DFS Pro and SaberSim partner Max Steinberg walks through his research process and talks about the types of things he’s looking for. We also had Danny Steinberg, Max’s brother, and Matt Hunter, a SaberSim founder on that video, and we went really deep into the research process. So if you have questions about that, watch that video.
For now, I’ll just kind of talk about what I would do with this. And in short, it’s not that complicated. If you feel like we are too high on a team or player, lower their projection. If you think we’re too low, increase their projection. Let me start with making adjustments to teams, because this is actually one of our coolest features, where over on the left, you have our game projections, which shows the number of runs we expect each team to score on average from our simulation data. And if you think we’re too low on the Padres, which right now I do, rather than trying to figure out how much to boost each player on that team and then lower the opposing pitcher and all that, I can just increase the average projected run total for them right here, and we’re going to use our simulation data to accurately adjust all the players in that game accordingly. So I will add 0.7 runs to them, and that’s actually a pretty big adjustment.
And this isn’t an exact science. You’re not trying to get these adjustments exactly right. Your goal is to move things in the right direction. For game adjustments, I don’t recommend changing a team’s total by more than a run, while for players, which we’ll talk about in just a second, you can use their value as a guideline.
A good place to look for this is just compare our projected totals to Vegas. Vegas is obviously very good at setting odds. We would put our numbers against theirs any day on average, but there are going to be spots that each of us excel at, and so it’s always helpful to leverage different points of data and use that as ways to figure out, “Okay, do I want to get closer to Vegas, or do I think SaberSim is right on with this one, where they are a little bit different, but maybe there’s a reason to it?”
And one other important thing to note is, we saw that we were getting a ton of the Dodgers. I don’t actually think that I should lower their projected run total here, because when I looked into it, we were already lower than Vegas on them. So if I was using Vegas as kind of the gold standard, which I don’t, but I do use it as a good point of information, I would increase their run total rather than decrease it. And so if I don’t disagree with the projection after doing it, I’m not going to change it here. That doesn’t mean that in my final lineups, I’m going to just have 90% of the Dodgers as a stack. That’s something that we can account for in the last step of the process.
What we really want to do here is adjust the projections in the right direction, if we think that is actually what the expected outcome is going to be. Otherwise, if it’s more just managing exposures, we’ll save that for the last step of the process. And so for pitchers, if I think Pablo Lopez is too high, which I do, what I will do is look at all the pitchers, and to see the range of values. So the top plays seem to be right around two. Pablo Lopez is significantly above that. And then there’s a handful of guys that are much closer to one, and then you get into like the relief pitchers below one.
What I’m going to do is lower Pablo’s projection based on my thoughts here. So if I thought he was just truly a horrendous choice here, I may lower him to a 1.5 value, but to go even to that or lower than that, I would need pretty strong evidence before making that big of a change. Because SaberSim is saying this is clearly the best value play for the slate. And so to completely disregard that, it’s not to say we’re always right, because we aren’t, but to completely disregard that, I would need a good amount of evidence.
On the other side of the coin, if I thought that they were an even better value, I would just want to be cautious, because they’re already the highest one by a good margin. So theoretically you’d go to three, but that’s 50% higher value than anyone else on the slate. So this to me seems like the top end that I would ever leave them that. But if it was someone you were trying to raise, you thought we were actually way too low on them, keep that top end in mind. I do think going to a little bit higher than the other top values on the slate is okay, but you just don’t want to get out of whack. Because again, these aren’t random numbers. We put a lot of effort into our models. So you want to tweak them, move them in the right direction, but don’t completely throw out what we’re giving you.
So right here, what I’ll do is put him at 14. I don’t think he is a horrible play. I do think he should be a contender for one of those top picks, but I don’t think he’s the standout play. So I’m going to lower him to 14, which is two times value. And I should have mentioned this earlier, but the way you calculate value is you figure out how many thousands of dollars does a player cost, so in this case, it’s seven. To get two times value, you multiply seven by two. That gets us to projection, which is 14. And again, though, what I’m really trying to emphasize is that we’re not trying to get this perfect. Projections are obviously a huge part of DFS, but they aren’t the only part, and we do a lot of those other parts, like looking at correlations, looking at ownership, and looking at upside in general, we do those very well. So don’t obsess here, because this is not the only work you’re doing. You’re touching all parts of the process to consistently add value. And really here, you’re trying to just move things in the right direction.
So a few of those other hitters that we talked about, typed out Fraley, same thing here. Three is the top end, but a lot of players are a little bit below three. Fraley, near the bottom of the order. I don’t want to, again, completely discount it, but I don’t think he necessarily should be as high up there. Trammell a little bit higher in the order. A little bit cheaper, though, but maybe we could put him at … Just call it. We want to get them to three. So that’ll be 6.6. He’s actually leading off, or projected to lead off, so I’m not going to touch that one. And Hilliard, similar to Trammell. Let’s go 6.8, so a little below three. And if I had more time, I would probably just look at all of these high value players, but I just want to show you the basics and how to get started. And I’m really just focused on the ones actually came up in the build.
So now let’s actually go to the next step, which is choosing your build settings. This step, to be completely honest, is almost a formality. Like I said before, because we’re simulating every game thousands of times, we’re able to understand correlations and variants, and incorporate that information, along with ownership, into the lineups that we build you, and we automatically adjust how we’re doing this based on the contests that you’re playing using these sliders right here. So you just simply tell us the contest that you’re building lineups for, and we will automatically adjust the sliders accordingly, based on what we have predetermined to be the best settings on average for that contest.
But it also goes beyond that. We have different defaults for every sport, and even within the sports, we have different defaults for all the contest sites, obviously, but they change based on the slate size. So a six-game slate will have different settings than a 12-game slate, than a two-game slate. We have spent a lot of resources into testing needs, and we’re always working to continue improving them. But this is just a spot where you can let us handle this part of the process, and then you can focus your time on those other steps, where you can actually add significant amounts of value.
And I’m not saying just completely ignore these. If you do want to know more about what’s going on behind the scenes, check out this video, and feel free to play around with it. But like those projections, you don’t want to make massive departures without really understanding what’s going on behind the scenes. And this video will do a great job of helping show what is going behind the scenes.
When it comes to actually building our lineups, there are generally two approaches that we recommend. One is just doing a single build for all of your contests. And in that case, I would just select, sort of like we did for actually the test build, I would just select something that’s based on all of my entries, about middle of the road. And the other alternative, though, is doing a separate build for each entry limit you’re in, meaning doing one build for your single entry contest, another for three max, and so on. I typically will do that second approach, and in that course I talked about, I walk through how I do that and why, but the reason for this is that the strategy for a single entry contest is going to be noticeably different than a 150 max.
And so I want to capitalize on those differences by not treating all of the lineups the same. But there is no right or wrong approach here. I probably wouldn’t recommend going all the way in that direction, of just doing a separate build for every single contest you’re playing, because the differences won’t be that big. You’re trying to, again, utilize 80-20. What is a way where I can quickly get a lot of impact for my efforts? And this is really where it’s going to be. For this video, though, what I’m going to do just for sake of time is a single build. But again, if you want to see me do a build for each entry limit, check out that course over on our website.
And another point that I want to touch on is, I’m going to build a unique lineup for every entry that I have, in order to spread out my risk, and also just give myself more chances at a first place prize. If you’re not as concerned about risks, though, and you just want to bet more on your top lineups, you can definitely take a more focused approach, and enter some of your lineups multiple times. There is no right or wrong approach here, again. So try not to overthink this, but if you are trying to grind out wins more consistently, and you’re not trying to go as boom or bust, I would recommend doing a unique lineup for every entry that you have. I’ve already reserved some entries, so I have 240 out there. This is pretty middle of the road for what all my lineups look like in terms of the contest, not messing with any of the settings, and hitting build.
As we get to this last step, dialing in your exposures, another thing I want to point out is that rather than just building you the exact number of lineups that was requested, we build you a pool of lineups much larger than that, so that you can make these exposure adjustments immediately rather than having to build an entirely new set of lineups like with traditional optimizers. If you say you want a different exposure of, say, the Dodgers, we will then sort through this lineup pool and find the best set of lineups that match the exposures that you’re looking for right away. You don’t have to go back and start from scratch over and over again.
We often refer to this step as quality control, because what you’re first doing is checking to make sure the adjustments that we just made worked as planned. And then we’re making some fine tuned adjustments from there to get everything dialed in the way that we want. If you find yourself making major changes here, then it’s really best to go back to the home screen and make your adjustments there rather than trying to fight what SaberSim has given you. Because again, if the foundation that we give you doesn’t match what you’re ultimately trying to build, start over rather than trying to build on top of it anyways.
So let’s just kind of quickly go through what we got. We’ll start with the teams. I’m still getting a lot of the Dodgers, and that makes sense. I didn’t adjust their projections at all. This is the type of larger adjustment that it makes sense to do here, because I’m not saying I disagree with the projections. It’s that I just don’t want this much exposure to a single stack. If this were an under the radar play that I really had a lot of confidence in, I could see keeping it to around maybe 50%, 60%, but it’s the Dodgers at Coors. This isn’t a shocker. This is going to be a really popular stack. And while we are accounting for ownership, I wouldn’t be surprised if especially this early in the season, people are just purely following rules of thumb and heuristics and just stacking cores really heavily. So I do still want exposure to them, because I think there’s a reason people like this situation, but I’m going to lower it to 40%.
We can also see that I am getting a lot of San Diego, and this, it’s a bit more. This actually makes me think that I made too big of an adjustment on the home screen. So I don’t need to start it from scratch, because I’m not bringing him back down to like the 10%, 15% they were originally at, but let’s just put them to 20%. And again, what we’re doing behind the scenes is sorting through this pool of 1,500 lineups to find the 240 that best match what exposures I’m asking for here.
Now, let’s look at the players. We’ll go back to the pitchers. And so for Lopez, yeah, we’re getting a lot less of him. We’re getting less of him than the field is. That’s what I expected. I could see potentially going back in and doing another round of edits here, maybe on San Diego, maybe on some of these other pictures. I didn’t look to it much, but for right now, what I’m doing is just doing a double check and seeing if this is close enough for me to just dial it in, or if I should go back in and work on some of those fundamental assumptions.
Let’s look over at the batters. And yeah, we’re getting a lot of Berti, which we’re okay with after looking into it. He’s supposed to be leading off, and that’s all right. But maybe we’re like, okay, this could still be a little bit too much. We’re three times the field on him, so maybe we just want to keep it to two times the field. We’re still saying SaberSim likes this play. I like this play, but I don’t necessarily like it three times as much as the field does.
Then who else did we have? We had Fraley. Now, we are a little bit below the field on him, which again, we weren’t trying to remove him entirely, just wanted to get something we felt was a little bit more realistic. And then Hilliard and Tremmell. Actually both these guys. Yeah. So for each of these cases, I’m fine with this. We have a little bit more than the field. We’re not going nuts. This is where I will trust it. Maybe there is more to it than I would have initially thought, and so I’ll run with these ones.
That’s really the process. Like I said, if you have a lot more time, you’re going to follow these same steps, but maybe you do another round of those initial adjustments. Maybe you start doing some research into some of the other players that are popping up. Whatever it may be, the process isn’t really changing. You’re still going through these steps. You’re just spending more time on those spots where you can add the most value, specifically reviewing and adjusting projections, but also dialing in your exposures.
And now the last part is, we just need to get these lineups into DraftKings. I will save them here and they’ll get loaded up into that. And this used to be a real pain in the ass. It required copying and pasting from different Excel files, and just kind of praying that you didn’t make a mistake. With our new Entry Editor with Late Swap, that is all a thing of the past. I already have my entries loaded in here and they just had a random lineup I put in to reserve those entries. And what I’m going to do is fill the contests.
We’re going to pick this bill that I just saved, and I’ve got videos that explain all this in more detail. But I had said I want to put a unique lineup in each of my contests, so I made sure to build enough to cover all of my entries, and I’m going to fill unique. Hit fill. You can see that this all got applied. My exposures got updated. Over here, download and open DraftKings. Click “Upload CSV.” Just find the file that just went in my download folder automatically, and we’re done. Now, when I go to my lineups, I can see that all of these lineups are in there. Really it just took seconds.
This is the process. Like I said, you’re going to adjust how much time you spend where, based on just how much time you have available, based on your experience, where you think you personally add the most value. But this is the process. There is no more spending hours and hours fighting with outdated optimizers. Instead, you can spend your limited amount of time where it makes the biggest impact, and that means you can get an edge on the field. You can get an edge on the competition without spending all day on DFS.
If you haven’t started a free trial yet, just head over to sabersim.com and you can get signed up completely free. If you don’t like it, you can cancel your account in just a couple of clicks, and you will not have to pay us a single cent. But because we’re so confident in our tools, that’s why we offer this free trial. We want you to get in there and see for yourself. And if you have any questions at all along the way, you can always contact us using this green question mark in the bottom corner of the screen, or by emailing me personally. We’re, again, always happy to help. So please do not hesitate to reach out. Thank you and good luck.