Master DFS Simulations: The Proven Path to Victory

Simulations are powerful tools that can supercharge your DFS process and help you win more. And it's been a huge year for the development of new, more powerful sim technologies in DFS. There is a huge opportunity in DFS right now to learn how to use sims before they become commonplace. The average DFS players are sticking to their old process, and the casual player doesn't even know these tools exist yet.

But the pros have been using sims for years, and now is the time for you to take advantage. But, just like using any power tool, if you don't know what you're doing, it's going to end up costing you. If you don't truly understand how sims work, You'll waste money and drain your bank while using them suboptimally.

In this video, I'll give you everything you need to know about DFS sims, and make you an expert at identifying what makes a good simulation, so you can maximize your profits using them. Here's the fundamental problem. DFS is a complex game. When you build a lineup, you're balancing the fantasy point to point projections of players, how players in the same game relate to each other in the form of stacking or correlation, how much other DFS players will use a particular player in their lineups in the form of ownership, the payout structure and the size of a contest, and the changing information as the slate plays out in the elate swap.

No matter how much of you know ball. Intuitively building lineups that correctly balance all of these factors is extremely hard, and it's almost impossible when you factor in time constraints. Optimizers are the conventional DFS tools that help you build lineups faster, but they aren't that much better at parsing all of these different important factors.

You have to set up all the constraints for the optimizer to follow to factor these things in. It doesn't know anything on its own about player ranges of outcomes, correlation, or ownership. So, enter simulations. What are they, and how do they help with this problem? A simulation is a recreation of something that takes place in the real world, and in DFS, there are two different types of simulations.

The first is game simulations of the NFL, NBA, MLB, or whatever sports game that will actually be played. The second is DFS contest simulations, which are recreations of the actual DFS contest you're entering. Simulations understand the whole of a game by recreating it over and over and over again, and seeing what is most likely to happen, and then it gives you insight into how to exploit that.

Let's talk about each type of simulation in DFS and why it's valuable. First up, game simulations. Now, if you've ever played a game of Madden or NBA 2K, that's a simulation. But most DFS simulations are far more robust than that. Recreating a game, taking into account all the factors that are important for projecting the game.

Coach and player tendencies, injuries, weather, matchups, all kinds of different things. Instead of having to use generalizations about which players are correlated to each other and what is most likely to happen in the game, A game simulation goes straight to the source by seeing what actually happens when the game plays out.

Game sims calculate the full range of outcomes of a game, and how players are interacting with each other in each of those different outcomes. These sims are repeated thousands of times for accuracy and precision, and the data from these simulations is used directly to build your lineups. Instead of optimizing for average projections, lineups built with simulations are the optimal lineup for a given game script.

Now second, contest sims are the new buzz in the DFS world, and for good reason. Contest sims are similar to game sims, but now instead of simulating football, for example, we're simulating DFS. Using the payout structure of a contest, an accurate projection of what lineups your opponents are likely to play, and game simulations to know how the sports games themselves might play out, we can recreate a DFS contest thousands of times, seeing how profitable different lineups might be if we played them into that contest.

Similar to game simulations, contest sims circumvent you having to make generalizations about how much projection you need, how much ownership you need, and what lineup construction you should play. A DFS contest sim takes all of that into account by competing lineups against each other and seeing which ones make more money if the slate played out over and over and over again.

The fundamental point here is that simulations for both the sports games and DFS contests let you get. far more accurate and precise with building profitable lineups, because instead of making guesses about what might happen and what makes money, it's simply tested out in the simulation thousands of times.

Now, with all that said, not all simulations are the same. There is a spectrum of quality and effectiveness here, and while it can seem like anything with the word simulation tacked onto it is automatically good, that is not the case. But simulations are abstract and complex. It's hard to be a good skeptic and know if a particular simulation is good.

So what makes a good sim? The first thing is using play by play game simulations when you're building lineups and when you're running your contest sims. Ultimately, you want your simulations to reflect reality. If you're using an NFL sim to make decisions about NFL DFS lineups, The sims should reflect the actual way an NFL game plays out.

Most sims out there are just extrapolating on an average projection for each player. When the sims sim a football game, what is really happening is a made up range of outcomes is generated for each player, and a random spot in that range of outcomes is selected for each player when you build a lineup.

But sports are not played on spreadsheets. Player distributions don't look like bell curves. They vary based on the player, the position, the characteristics of the game, and the sport itself. And these assumptions that most contest sims use also completely ignore the correlation between players. Now, some sims out there might use historical ranges of outcomes for players, or historical correlations between players, But this still misses the mark.

Every game is different, and to beat DFS, you need to focus on the game tonight, not the averages from over the course of the season. Plus, most contest sims still require you to build the lineups to be simulated using an outdated optimizer, meaning you're giving up a ton of value in using the game simulations to build your lineups in the first place.

SaberSim uses a true play by play simulation for all of our sports. When you build lineups or run your contest sims, we are playing every game on the slate out play by play thousands of times. We don't use generated normal distributions, and we don't base things off historical averages. This gives you a more accurate assessment of what a player's upside looks like, and what the games look like when they get there.

Now, this can be a hard thing to visualize, which is why we make it clear what our player distributions and correlations look like in our app, so you can see exactly how our sims are thinking about a game. If you can't see this information visually presented to you with the simulation you're using, You should be skeptical about what assumptions are being used to build out that sim.

It's very likely bad assumptions about what is going to happen in the game are being used to generate your data. The second factor that makes a good simulation, particularly a good contest simulation, is precise contest information. Just like average projections and rules of thumb are overgeneralizations we want to avoid when building lineups, we also want to avoid overgeneralizations about the contests that we're playing when we're running contest simulations.

Not all DFS contests are the same, they have different payout structures, and the types of lineups people play are different in different contests. Now, these changing factors have a huge impact on the optimal strategy when you're running a contest sim. How can you know how to beat your opponents if you don't know what they're playing, and how can you calculate how profitable a lineup is without knowing what it wins when it wins?

A good contest sim will update dynamically to account for different field types and payout structures. Generalizing this information is going to make your simulations much less precise and lead to you playing suboptimal lineups. in your DFS contests. On SaberSim, we have 13 different representative contest buckets, all with different projected fields.

Our expected fields understand that people play shockier lineups in single entry, and that higher stakes contests are going to be sharper. And you can automatically import the exact payout structure from the contest you're playing to use in your contest sims. This means that when you run contest sims on SaberSim, you're simulating your lineups in the contests you're actually going to be playing them into, for razor sharp precision into the optimal lineups to play.

The last thing that makes a simulation accurate is the most important one, adjusting for live information as the slate is playing out. Now, if you're watching this video, you probably already know that late swap is a huge part of DFS strategy. Reacting to changing information during the slate can be the difference between winning or losing, And you'd never willingly put yourself in a position where you ignore a late scratch in the middle of a slate that changes everything.

However, there is a ton of information that we get throughout the slate that most DFS contest sims and most players completely ignore. How players are performing in the games that have already started, and the ownership of players in games that have already started. So why do these matter? Well, as the games play out, the possible ranges of outcomes for players change significantly.

If a player projected for 25 points has 20 in the first quarter, that player's new live projection and possible range of outcomes is a lot different than it was at lock. And the expected ROI of lineups that contain that player, and the ones that don't, have also changed significantly. This compounds as you start to think about every player that is currently in game.

The ROI expectation of every lineup you played is constantly shifting as the slate plays out. And we should be reacting to this information. If a lineup is crushing, we might not need the low owned play that we stuck in in the late games that we originally had in our lineup. And if a lineup is underperforming, we might need to get off of some of the chalk to get some of our min cash equity back.

Ownership works the same way. If a player projected to be in 20 percent of lineups is only in 5%, well, the field you're playing against is likely very different than originally thought, and the ownership projections for the late game players should also shift accordingly. Each late swap window we have on a slate is not only an opportunity to update for breaking But it's also an opportunity to update our lineups to react to what we now know that we didn't know before.

SaberSim updates in near real time, accounting for how the games in progress are going and the ownership of players in games that have lost. We run new sims every couple minutes from the current point of the game on, constantly generating new ranges of possible outcomes for all players. And we update our field lineups for DraftKings Contests Live, updating our fields to adjust for what your opponents are actually playing.

Your late swap contest sims on SaberSim take this information into account automatically, meaning that each time you late swap, You're maximizing your possible ROI for every lineup you played with the information you have at that time. Simulations are not just useful for when you're building your lineups at lock or late swapping during the slate, they're also the most powerful tool we have for analyzing our results.

the pros. Reviewing your DFS process is challenging because it's hard to know when we played bad or just got unlucky on a slate. It can take playing for months or even years to test out a process to see if it works. But simulations can help answer that question for us immediately. If you're not using sims to critically analyze your play, you are never going to get better.

So to learn more about how to use powerful DFS simulations to backtest your DFS process, check out this video. And to learn more about DFS sims and how to get ahead of the competition, subscribe to our YouTube channel for more DFS tips and insights. Thank you for watching. I'll see you in the next video.

‍