Transcript

How should a new player build their DFS process?

Andy Baldacci:
How should a new player, someone who’s new to DFS, just getting started with things, how should they try to develop their DFS process? Because that’s one of the things that everyone talks about is build your process, focus on the process, and this and that, but when you’re just getting started out, it’s not really even clear what that actually means. So if someone is committed to the game and does want to get better and really build out that process, how would you recommend that they do that?

Max Steinberg:
Yeah, I can take this one. I would say there’s a couple of things that just come to mind immediately when you’re talking about this. I would say one is just reading up, watching videos. We have a lot of videos with SaberSim on our YouTube channel, reading strategy, just try to find all the content you can and just eat up that content, because there’s good information out there. There’s obviously a lot of smart people who play Daily Fantasy, and the info is out there and you can always learn more and more. I even … I’ve been playing Daily Fantasy for five or six years and I still will read articles or just when someone tweets something out who I follow, read and try to learn something.

Max Steinberg:
I think also in that same vein is looking at … especially if you’re entering contests that good players are in, maybe even single entry contests or cash games, look at those players’ lineups. Look at the players who are entering thousand dollar heads-up, thousand dollars GVPs, look at their lineups in these double ups or single etry GVPs and just see how they’re different from yours, what players they’re using that are unexpected, and think about why, because I think there’s a lot and you can just learn from those players. They’re good. They’re playing these high stakes for a reason, and I think obviously they’re going to make some plays that you might not understand initially, but you possibly can learn something from.

Danny Steinberg:
Yeah, I could follow up on that too. I think one thing that can be really a red herring is just paying attention to the results of each slate or each week. There’s a lot of variants in DFS, so a lot of times when someone does good or bad, you’re not necessarily going to know whether that was luck based or that was something that was predictable. So I think … I mean one thing that’s really helped me and Max both get a lot better is … our learning computer programming and data science is something that’s an integral part of our process. So I think really that’s a way that I’d really recommend if you really want to get good at DFS, that you have to have a data science aspect to your research.

Andy Baldacci:
And or that, Danny, do you think that is a necessity for all players? Or is it something where … I think there’s got to be a line you’ve got to draw somewhere, where if you want to compete at the absolute highest stakes, you do have to –

Danny Steinberg:
Yeah, totally.

Andy Baldacci:
– treat it like a full time job and probably look at some advanced analytics, but if you’re trying to make some money on the side playing the under $3 contest, do you think it’s … how far do you have to go?

Danny Steinberg:
Yeah, I mean I think there’s a lot of really good research out there and a lot of good people who have done a lot of good research, and I think for someone who maybe doesn’t want to do all the research themselves or doesn’t have the ability to do that research themselves, there’s a lot of people you can follow on Twitter. There’s people who have written articles everywhere that have really interesting thoughts on Daily Fantasy strategy. There’s a lot to learn from just everyone on the internet, and I think you can learn a lot and get better just by learning from other people.

Max Steinberg:
Yeah, I’ll piggyback on this as well. You don’t really have to … in order to win at Daily Fantasy, especially for using a sophisticated line of [inaudible 00:03:59] we offer, if you’re taking advantage of correlation, you’re taking advantage of with upside, all you really need to focus on is getting better projecting or getting better at doing ownership projections, and I think if you can focus on making sure that when you make an adjustment to say a person’s baseline projection or ownership projection, you’re adding value in some way that’s going to be a win, and if you have enough wins every slate when you’re adjusting projections or ownership projection, that’s going to make you a profitable player, and so if you can focus on where am I sure that I can add wins to these adjustments, I think you’re going to do a lot better, and you’re going to be profitable, especially on these low stakes contests or these single entry contests or so on and so forth.

Andy Baldacci:
And I think that actually is a good way to kind of transition into the next question, and it’s something that did come up and chat and also someone had asked this over email. It’s just come up a lot, and it’s kind of the idea of if SaberSim or whoever you’re using for projections is putting out solid numbers, when should you make adjustments here and why? What are the things that are going to add value to that? You talked about adding wins and finding ways to really kind of improve upon the projections, but what types of things specifically are you looking for there, Max?

Max Steinberg:
Yeah, so I would say anything that uses very advanced statistics or new statistics … SaberSim mostly works off of historical data. So if there’s something new that is something that’s like stats.nba, which uses cameras to get all these advanced statistics that we don’t use, something that has to do with expected points added or complete … there’s someone who is named Ben Baldwin on Twitter who tweets out this completion percentage above expectation stat that is really helpful. Things that are essentially not just plain historical data, you can use these sources and they can add something. I mean I think there’s a fine line, because I think sometimes people can get into “advanced statistics”, and you end up just using something that isn’t predictive and not helpful, but especially in NBA, I think this is just coming to mind for me a lot.

Max Steinberg:
It’s just stats.nba. There’s a lot of stats out there. People don’t really use it that much, but there’s a lot of really detailed, really good stats because they’re based off of actual cameras, video taping the games, and actually getting movement and player position, so on and so forth, and so if you use something like that and you find something that you think is helpful, something like [inaudible 00:06:36] field goals, drives to the basket, something, have a theory about it, try to look up and see if this theory may or may not have some weight, and use something like that. I think that is certainly going to add something to SaberSim’s model.

Danny Steinberg:
Yeah, I think that’s a good point. I think that sort of very nuanced information can be really helpful if you’re interpreting it for some something that’s a real deep dive into what’s going on with a team from a very advanced statistical level, that you can make sense out of. I mean some of the easier stuff to add is just providing the projections with context that’s useful. So for example, basically every projection system is relying on historical data to project out the future. Now has a player been playing through an injury? Did they have a game where they left halfway through the game? That’s something that a model’s not really going to know. So adjusting for that sort of context. Sometimes teams will decide to drastically change the role of someone on a team. Sometimes a team will be like, we’re committed to this guy at running back and we’re really going to try to establish the run that happened this year with David Montgomery. So I guess reading through the news and understanding where a team has decided to drastically change direction in some way, that can be really useful in understanding the context of the historical data can be really useful.

Max Steinberg:
Yeah, I think a great example of the injury example is Adam Thielen last week.

Danny Steinberg:
Yeah, a perfect one.

Max Steinberg:
Obviously, been playing through injury for four weeks, but we know historically he’s been a really, really productive player, and it ended up working out last week when he was back to his old self. This week he might be injured again, so who knows? But when you can catch those things where there’s a recent injury that’s sort of marring someone’s performance, you can end up just finding an easy way that someone’s under projected and most people are under projecting them and then they end up just being back to their old self, which is not hard to predict.

Andy Baldacci:
And Max, you touched on this a little bit, but when we get into advanced statistics, when we get into research, when we get into just kind of going over historical data, where do you think people can kind of fall into a trap and over value of the data out there? Max or Dan, either of you guys can jump in on this one.

Danny Steinberg:
I’ll handle this one. I think stuff with matchup can be really overrated. There’s just a lot of statistics where you can think that they would be predictive of the future when they’re really not, like recent performances, as an example. I remember one time a few years ago, Brett Gardner, the lead off hitter for the Yankees went on a streak where he had seven straight home runs in seven straight games, and he was the [inaudible 00:09:27] player every night, and everyone just thought, “Oh, this is never going to end. He’s going to always hit a home run. He’s just really keyed in,” but in reality, very extreme things are going to happen because of luck alone. So just because someone’s done really well in a certain situation or in a certain context doesn’t mean that that’s necessarily going to be predictive of anything in the future.

Danny Steinberg:
I mean another example is how a team performs against the spread in different contexts. That’s something that a lot of touts use in the sports betting industry. It’s just extreme things happen and those extreme things aren’t necessarily going to be predictive of something extreme happening in the future, and another obvious example would be steals in basketball. Someone has a 10 steal game, are they likely to have a lot of seals next game? No, of course not. Good steals are the most unpredictable stat in all of basketball. They’re very, very random, so don’t pay attention to stats that are really, really noisy, basically.

Max Steinberg:
Yeah, and I’ll just add one more thing. I think especially in football, I see this, is people making game script assumptions based on Vegas lines. I think when you’re … especially if you’re playing cash games, I think that can be valuable, but in tournaments, how the game script actually goes, what I found is random enough to the point that unless the spread is just giant where it’s an almost certainty that a team is just going to be leading the whole game, i.e., like a team is a 14 point favorite, but even we see that, the Dolphins versus Patriots, I think the Patriots were a 17 point favorite in week 17 and they lost.

Max Steinberg:
So anything can happen, and I think people just make these assumptions with a five point spread that this team’s going to be leading a whole game, and the game script can really go anywhere, and luckily for us, SaberSim accounts for that, so we don’t need to worry about that with our smart diversity, taking into account small [inaudible 00:11:30] of simulation, which I talked about in another video. I won’t get into it, but that’s something that a lot of people, they try to think, “Oh, how’s this game going to go? How’s this game …” they’re five-point favorites. This is just going to be a run heavy game, and it’s like, no, that is not what’s going to happen. You’re going to see it today, because there’s a lot of teams that are seven point favorites, and one of those things is probably going to lose. So making over assumptions on that I think is a huge mistake.

Andy Baldacci:
And this I think ties in pretty well to kind of the last question we have about just kind of building that new process, and it’s saying when there is so much noise out there, when there are so many statistics that aren’t truly predictive or aren’t predictive without a much larger sample size, how can you objectively judge whether or not your processes actually improving or not?

Max Steinberg:
Yeah, I think it depends on the sport, because you have to think about how much variance is in a sport, and I can just tell you, baseball is one of the higher variance sports. Football is a pretty high variance sport. Basketball is a low variance sport. So if it’s something like basketball, which is obviously going on, we’re in the middle of season right now, so that’s something probably people want to hear about, but with basketball, you can actually look at the box scores and say, okay, I adjusted this per person’s projection, A, because I thought it was going to do really well, and where obviously there’s some variance statistics there, if you adjusted it up, you should’ve got higher minutes than SaberSim expected and you should have had higher production in terms of fantasy points per 36 minutes or just the [inaudible 00:13:12], whatever you thought, it should basically happen.

Max Steinberg:
There could be variance in outcomes as the person gets into foul trouble, they shoot poorly, but you have an idea that okay, this actually worked out, and sometimes that’s not the case, but you can actually get a lot from what the actual results are. In baseball, you can’t do that at all. The results are so random, and we can use that to our advantage, because we can stack some of the low on teams that have high upside, and that’s going to benefit us, and I think with a sport like baseball, you have to think, okay, what is the … you have to actually theorize more. You have to say, okay, why do I think this is going to be the best stat? Think about what are sort of the under the radar stats. Maybe try to think of what pitcher might struggle, look at weather staffing. We take this all into account, but sort of think about it that way and maybe make some adjustments to our ownership projections if you’re using SaberSim and want to go that route, and then also just look at, again, what good players are doing and try to assess things that way, because if you’re doing something right, probably really good tournament players will be doing it with you. So I think those are the things you have to keep in mind.

Danny Steinberg:
All right. That’s well said. I just wanted to add just a few things. One is, I think you’re definitely right with what you’re saying about basketball, but I just want to emphasize if you’re judging your process based on the results of a slate in baseball, that’s very, very wrong. It’s so random that if you’re saying, “Oh, this team did well and I had this thought about them, so that must be … or they had this factor in their favor, that must mean that this factor really matters.” Terrible. Exactly. Whether someone performed well or badly in a baseball game is very dominated by luck. So it’s kind of like poker. You kind of have to have some theories about the game and sort of believe in yourself and your understanding of the game and whether something makes sense at a theoretical level, or you have to do some really high level research.

How do you handle downswings?

Andy Baldacci:
Assuming you do have faith in your process and long term you have results to back it up. You understand the theoretical underpinnings of it. How do you still deal with those down swings? How do you deal with getting through the burnout of yet another slate where you’re getting crushed without second-guessing yourself when you do feel like you have a solid process? How do you navigate that kind of mental side of it when things aren’t going your way?

Danny Steinberg:
I mean, it’s very difficult is really the answer. I’ve been gambling for a living for like 12 or 13 years and it’s never easy when you go on a downswing, but you have to keep things in perspective. If you’re doing a multi-entry GVB tournament’s in Daily Fantasy, you’re going to have times where you go on a downswing. It’s just going to happen every year. So you have to not get to have your emotions get too wildly swingy on winning a whole lot and getting on winner’s [inaudible 00:01:21] and thinking you can predict everything perfectly to losing a lot and thinking everything I do is wrong and I have no idea what I’m doing. I think there’s a middle ground there I guess that you have to keep perspective on.

Max Steinberg:
I mean we used to play poker professionally, it’s a lot easier than poker. Poker, you go on a downswing and you have to really be in the game and playing and make decisions on the fly. And that’s really hard. I think in the beauty of Daily Fantasy is there’s always another day. You have time to prepare, you have time to get yourself in a good mindset. And I think just make sure you’re getting yourself into a good mindset every day and you can always go on [inaudible 00:02:06] and keep being really mad then night of. But just make sure that you’re not mad two hours before a lock in and being emotional. When you need to make good lineups, right.

Andy Baldacci:
Yeah. No, that’s a really good point because in poker, if you lose a huge hand, you’re getting cards dealt immediately after that, and you have to make good decisions immediately afterwards. And if you’re playing against the same player that just stacked you on a bad beat you have to keep your emotions controlled. Whereas in DFS, there’s a lag in between results from one day and starting up the next, which doesn’t necessarily make it easy, but it does give you more opportunities to control your emotions and keep them in check there.

Max Steinberg:
Yeah.

What mistakes do people make when switching to SaberSim from a traditional optimizer?

Andy Baldacci:
I thought it’d be valuable to talk about just some of the mistakes people make when using SaberSim and specifically, when using SaberSim and coming from a traditional optimizer, what some of those common mistakes are. Because fundamentally, traditional optimizers and SaberSim function very differently, but it’s not always easy to grasp. So I wanted to ask Max, I guess, if you could kick it off, just what common mistakes do you see people make when switching to SaberSim from a more traditional optimizer.

Max Steinberg:
Hey, yeah. I get a lot of Twitter DMs about this, so I’m also acutely aware of this. I mean, everything that we do with SaberSim is based on our simulation data and it has to do with using correct slider settings. We have our Correlation Slider. We have our Ownership Fade Slider and we have our Smart Diversity Slider. That’s basically the complete key to our entire process. I see a lot of people who will DM me on Twitter and they’ll show me their sliders. And they’ll be like, “How does this look?” It’ll be something that’s outrageous, like it has nothing to do with the preset settings and it shows a lack of understanding the sliders.

Max Steinberg:
So I think one thing that is really important is educate yourself about what SaberSim actually does. There’s a lot of videos that Danny and I have made that explain in detail. You can ask us on the Slack channel. We’re happy to answer any questions, especially now if you’ve created this whole thing … knows it pretty well. So I think if you want to get better at using SaberSim, you have to understand what each slider does. Because once you have that understanding, I think you’ll find actually that the build process begins way easier and you’re not actually making as many adjustments post-build. Which I think is another big mistake that I think a lot of people do, is they just try to control too much as they try to set too many settings before they build. After they build, they’re adjusting exposures too much.

Max Steinberg:
The thing is is all you should be worrying about is making correct projections, adjusting projections so they’re as correct as possible and adjusting the ownership projections so they’re as correct as possible. If you do those two things well and you understand our sliders and set them correctly, you’re going to make really good Lineups, especially in sports work correlations, very importantly NFL, hockey, baseball, et cetera.

Max Steinberg:
Things where that’s really important, you’re going to make Lineups that are certainly better than Lineup optimizers out there, right? So focus on ownership projection, understand the sliders, make sure they’re set right, to the right tournament that you’re playing and you should do pretty well.

Danny Steinberg:
Yeah, I think, just adding to that or I mean, I think one mistake you can make is just not trusting the solver to do what’s optimal. You can try to hack it to do what you want. You should trust it to do the right thing based on whatever projections you have. If the solver’s not getting someone that you really want to get and you agree with his projection, but you feel like you should be in Lineups, just no, he shouldn’t be in Lineups. The solver knows better than you. You got to trust it a little bit.

Andy Baldacci:
I think that touches on a good point where it’s not as though we are saying SaberSim is perfect. It’s obviously not. No tool is going to be. Even though we are spending a ton of time and always will be improving it, you do have to ask yourself, “Okay, what is the tool telling me because we’re not functioning as a glorified spreadsheet, which optimizers basically are. We’re actually considering things upside.” So there is insight in what SaberSim will give you. Like you said, Danny, if you do agree with the projections, if you agree with the Ownership, the point projection, but you disagree with the exposure that you’re getting in your build, there’s some disconnect there.

Andy Baldacci:
Because we’re saying, based on everything you’ve given us, this is how much you should get that person and if you don’t disagree with their variables going in, then you really check that assumption.

Danny Steinberg:
That will vary on all of them.

Andy Baldacci:
Again, that’s why all of it goes back to those projections because that’s where you do want to make adjustments and you do want to improve and build on what we’ve given you. But once everything is dialed in, it’s not as though you just set it and forget it and don’t make any changes from there. But you do want to at least listen to what we might be trying to say.

Max Steinberg:
Right. I think using the post-build adjustment features that we offer should be making small changes in, we call it quality control, right? You’re making small adjustments to make sure you get a little more, maybe get a little more of this flair, a little less of this flair, but you shouldn’t be changing the entire build with just post-build. You should be adjusting projection and you should be adjusting Ownership projection or you should be adjusting the sliders because your sliders are not correct for the GPP you’re playing or the cash game you’re playing. But you don’t really need to be adjusting things and you shouldn’t be adjusting things too much, trying to basically rig your Lineups to get them right, because you don’t need to do that with us.

How do you build lineups for cash games?

Andy Baldacci:
When going from GPPs to cash games, what adjustments are you making to your approach? How are you changing the way you’re thinking about the entire process when you are approaching cash games?

Max Steinberg:
I can take this one.

Danny Steinberg:
Okay.

Max Steinberg:
Yeah, I think, again, no matter what tournament you’re playing, your focus is always on making the correct adjustments to projections and the correct adjustments to ownership projections. And, The difference between the cash games view is just what you’re valuing. Like how you’re valuing correlation, how you’re valuing upside, how you’re valuing ownership fit, right. So with cash games you actually want your players to be reverse correlated. If you don’t want to go that route because that is actually going to sacrifice some projection and that way in order to get that…

Andy Baldacci:
Can you just ask a little bit Max on what that actually means when you say you want reverse correlated players?

Max Steinberg:
What does it mean to you basically want your, I mean, so the goal of a cash game is to, so cash game is head to head,50/50 or double off, right? So, and head to head, you’re trying to just beat one person’s line up and the double up or a 50/50 you’re trying to beat the [inaudible 00:01:19]. It doesn’t matter if you’d be good at if you’ve got 1000 out of 2200 person field or are you at first, right? So maybe a lineup that you’re building to get first is really, really stupid, right? Want to line up that as much of the time as possible. Your going to be that top half or beats someone [inaudible 00:00:01:40]. Right? And so with correlation, when you want a really high positive correlation means you’re trying to build line ups that all your players are running, going to do well together. They’re correlated, their performances are correlated. That’s what it means is they’re going to do well together.

Max Steinberg:
But when you’re playing a cash game you want it so your lineups basically, no matter what happens are going to barely be above that 50% threshold. And so you want players that don’t do well together. You want them to sort of one player does well but does another player does badly but they sort of like even each other out and you sort of just want to line up. That’s going to get a point enough points that’s just going to be at that field and so when I say negative correlation, I’m just mean that you want players that are just not going to perform well when they are performance while in football is going to be sometimes two opposing running backs.

Max Steinberg:
This might mean using a defense and a wide receiver on opposing teams, it’s you’re going to be doing things that just obviously would not be good if you’re trying to get a 200 point NFL score but are going to be really good if you’re trying to beat 50% of the field. But I think in the process of saying right is now all I do from week to week, if I’m adjusting projections and try and get good ownership projections, right when I’m building for a cash game, it just has to do with said again thing. This slide is settings correctly and that they’re going to be negative correlation. That’s not going to value ownership and it’s not going to value upside. Right? It’s just going to be let’s get good average rejection, negative correlation and be with the field basically. It’s fine with me.

Andy Baldacci:
Danny, do you have anything to add to that?

Danny Steinberg:
Yeah, I think your approach really should just be the same as to make the optimal projections and then change the sliders and settings based on what kind of game you’re playing in. So I think, I think the process is overall mostly the same. You’re just really, as Max is saying, you’re trying to get your projections right and trying to get the ownership projections right and then adjusting the sliders based on the type of game you’re playing.

Andy Baldacci:
And then does this apply? Like do you guys treat all cash the same? Is it? If you’re playing head-to-head, if you’re playing a 10 person 50/50, or planning, one of the big 50/50 says multi-entry, like are you playing all of those roughly the same way, like with the literal same lineup? Are you building lineups differently? Are you playing different lineups and just like how do you think about the different cash contest types?

Danny Steinberg:
I just use one lineup. I’m curious if Max would.

Max Steinberg:
I know I, I’d say the same thing. I think you know you can get some more new op of you’re doing it triple op or quintuple op or something like that, but if we’re talking about 50/50, double ups, and head to heads and you can do different lineups, but that just means you’re going to sacrifice some ROI from going off the best line. All right, fine. I mean people do that, they hedge their risk, but if you’re new to [inaudible], I probably wouldn’t recommend it because you don’t want to risk possibly putting in a lineup in that is minus CV. In order to headdress grad, you want to make sure that the labs you’re putting in are going to make you money. So I hedging in that way by paying for different lineup, six different cap lineups. You run that risk of putting in a lineup that’s just not a good one and so.

Andy Baldacci:
One thing that just came up in the chat that I want to touch on and it’s just kind of common was so change sliders and when all the monies, and I know we’re not heavily saying I think they were joking, but it is something good to touch on is that we, I guess from my perspective is I think almost too much focus is given to getting the sliders exactly right. I think as long as you’re within general ranges for the game type, that’s all you need and you can kind of dial in and beyond that to your own personal preference. But there’s not like one exact set of size that this is perfect for this contest every time it’s going to be based on risk tolerance in business. A lot of factors involved, but that is kind of the aside of it. The big thing really is dialing in the projections and we behind the scenes are always going to work on improving them.

Andy Baldacci:
But that is kind of what we are saying is the where the most value can be added is kind of spending your time improving that. And none of this is easy, but I think just because we are doing things a little bit differently by having those sliders, it can throw people for a loop where they kind of overthink that and that is the phase that it’s not that it matters the least cause it does matter, but it doesn’t require nearly as much time as the other phases do. So I just kind of wanted to just clear that up a little bit. Do you guys agree with that general sentiment?

Max Steinberg:
Yeah, of course.

Danny Steinberg:
Yeah, for sure.

Andy Baldacci:
Then the last thing on this topic I wanted to ask is do you think cash games are worth playing for the average player, and I guess that’s a bit of a blanket statement, but like do you think cash is profitable for all players not all players, but like a serious player for every single sport or how do you guys personally approach that?

Danny Steinberg:
I mean, I think one thing is one thing I really believe with SaberSim is our solver does a extremely good job taking into considerations all the factors that really matter in GPPS. So I think I’d really recommend if you’re using SabreSim to do GPP is you can definitely win at cash, but if you’re good at projecting players, you’re probably going to have a bigger edge of GPP than you are at cash. And SaberSim’s going to do a really good job building the best lineups with the projections that you have.

Max Steinberg:
Yeah. I also will piggy back on this. I think people get attracted to cash games because they think there’s less variance in them because they think oh well. Like a lot of the weeks, I’m just going to double up and some weeks I won’t. But if you’re making, you know 20 or 50 lineups and these low stakes do GPPs from getting a good spread, the variants are really, is not as high as one would think.

Max Steinberg:
And so I think the things that are attracting people to cash games actually can’t exist when you’re being GPP player and your ROIs are going to be higher in GPP is because of everything. Danny just said. SaberSim is built to make really good GPP labs. I think now people are attracted to cash games. They think it’s low variance. GPP is really aren’t that much higher variance and they’re going to be more profitable, which I think is the most important thing. Cause again, the worst thing you can do is put in a negative CV lineup into one of these tournaments. Cause variance doesn’t matter, then you’re just losing money right.

How do you pick your 3-max lineups?

Andy Baldacci:
You did have that video on single entries that covered a lot of the stuff there. We talked about more of it here. But, for those three-max contests, Max or Danny, do you think it’s… It’s still going to be a somewhat manual process at the end of it of picking the lineups because you don’t need to just pick the exact top three, you can kind of tinker a little bit from there without giving up much. But, in a three-max contest, are you going to… Would you do that? Would you just pick the top three or are you going to find three lineups that aren’t very correlated to each other?

Danny Steinberg:
Do you want to take it, Max?

Max Steinberg:
Yeah. I think it depends on how much time. Again, it’s variance time profitability, right? As is, how much time do you want to devote to this of making a bunch of lineups and sort of hand choosing the ones that you like. Which is totally fine strategy, but it’s just going to take a lot more work to make sure that you have the specific three entries you want or you can just set smart diversity to very high and just trust that the lineup builder is going to make a diverse array of lineups. Make sure that it’s high enough that you have three different QBs, things like that. I think in general, I usually don’t worry too much about really managing all my contest entries so they’re spread out a lot, and that I have the perfect three entries for three-max contests, and I have the perfect five for five-max, and the perfect 20 for 20. I trust that the SaberSim builder is building me three good lineups, five good lineups, 20 good lineups, right?

Max Steinberg:
I think a lot of people get really micromanagy in a way that doesn’t really add value. So, I think personally for me, the SaberSim builder builds the lineups. Again, I’m saying the same things over and over again, but I’ll just say it again. I’m working on projections. I’m working on ownership projections. SaberSim builder’s doing the rest. I set the sliders to the settings I want, depending on the contest, and that’s all I do. Right? I think micromanaging can feel rewarding when it works. It can feel terrible when it doesn’t work. But in general, you can’t micromanage your way to the perfect lineup. You just have to have some trust, and I think you can trust the lineup builder to make you quite good lineups, especially for these three- and five-max and 20-entry max contests.

Andy Baldacci:
Yeah. I think elaborating on that a little bit, I guess, or kind of going in different direction. It seems like we could also say that the more lineups you’re entering in a contest, the more costs there is to tinker in. Like, two. If you start with a base from a lineup optimizer or from SaberSim, whatever, your top few lineups are all going to be very solid. If you choose the number two instead of number one, number three instead of number one, whatever, you’re probably not giving up too much. As you start entering more lineups in the contest and start kind of picking from all over the board, the risk that you’re swapping out a plus CV lineup for a minus CV or just giving up a lot of EV between the lineups gets substantially higher. So, it’s just as you put in more and more lineups, you just want to be careful about the micromanaging because the costs get a lot higher and you can realize those a lot faster. Is that fair to say?

Danny Steinberg:
Yeah, that makes sense. Yeah. I think another thing is just with like, basketball for example. You should be paying attention to the last second news and if there is any, and to be prepared for that. A lot of times I make my lineups with like, five minutes before lock. So, there’s not time really to micromanaged if you’re paying attention to all the injury news.

Andy Baldacci:
That’s true.

How do you decide when to fade or play the chalk?

Andy Baldacci:
When do you know if it’s good chalk or bad chalk? When is it time to eat the chalk or fade the chalk? How do you guys evaluate when to, I guess, pay attention to ownership and when to ignore it?

Max Steinberg:
Yeah. I think we can talk about the underlying concept between why to fade the chalk or play a chalk, right? The underlying concept is, is the chalk worth it? Is the chalk is getting a certain amount of high ownership and is this player projected high enough, then will not fail enough where this is worth it. And so there’s reasons why this might happen maybe this player is higher variance than people think. They might be high projected player, but they actually fail quite a bit of the time and they have high upside and they have downside. And when players are really chalky you can kind of think about it in reverse. If the player has high enough downside, then they’re going to fail enough where it’s not worth it or if they’re not projected high enough where there’s not a big enough difference between this player and a player that’s just a couple points projected lower but is going to garner away lower ownership you’re going to want to fade the chalk.

Max Steinberg:
But if that player is just way higher projection than everyone else here and I want to play it. But with regard to say if you’re using SaberSim, all you have to do is project the players right, project the ownership right. Use the Ownership Fade slider to decide how much you want to consider fading in ownership and then let the builder to do it and it’s going to make the decision for you. But the underlying factors with that is does this player have a lot of downside? Is he not actually as good as this other low-on possibility? And that’s all basically SaberSim is considering.

Danny Steinberg:
Yeah, I just want to be clear for cash games, you shouldn’t want to avoid high ownership. In fact, you may want to lean into I ownership a little bit. So this is just specifically for GPPs because they’re top heavy. So you want to take a higher variance approach. So fading the chalk can sometimes be good and for exactly the reasons Max said. And I think SaberSim really takes us into consideration well, and if you’re playing like a high Stacks multi-entry GPP where ownerships are going to be really concentrated, you may want to fade ownership a lot. But if you’re doing a contest under $3 or something that is going to have a lot of beginners in it, then maybe don’t take Ownership Fade that much into consideration.

Max Steinberg:
Yep.

Andy Baldacci:
And one of the other fundamental things that you guys touched on a little bit, but I want to emphasize is, is the underlying variance of the player. Because in baseball, if you had a batter who was projected to be on like 60% it probably almost never makes sense to take that bat. Whereas for a pitcher, it’s a little different because just inherently the variance of those positions, of those players is going to be different. Because if you had someone who was very highly owned but knew with true certainty that this is the score that they’re going to get, that’s much different then saying this guy’s the ton of upside, but they only get it 10% of the time, 20% of time, whatever it may be. And I think it’s kind of understanding those trade offs are very different for different sports.

Andy Baldacci:
It’s like in basketball some of the starters are going to get a certain amount of minutes with a very high certainty and you can expect a certain amount of performance. There’s a range in there, but that kind of the band of uncertainty is much smaller for them. Whereas someone coming off the bench or someone who’s could be potentially injured, someone where there’s more uncertainty around the minutes, around performance, all those things. The more uncertainty there is, the more willing I guess you should be to fade and that seem like a good summary as well?

Max Steinberg:
Yeah, I would say that. I also think, and this is something SaberSim is going to take into account, but just as a general thought is, if there’s any leverage off this player doing badly so if it’s a pitcher in baseball, you can stack against that pitcher. So that’s going to add value to fading that person. If there’s a running back that people are going to really like, and you actually play the defense on the other team. Okay, that’s also another way where you can have some leverage off of that player. So I think if there is that leverage, it’s going to encourage. That’s another reason that you might want to fade a player.

When do you take a stand on lower owned players?

Andy Baldacci:
For those lower-owned players, for those cheaper guys, what do you look for when you’re taking a stand on them? And again, with SaberSim, it’s going to point you in the right direction, but fundamentally what types of qualities are you looking for in those people who might not be as popular, for taking a stand on them? Do either you guys want to touch this one?

Max Steinberg:
I thought I was going to let you take this, Danny. I was just thinking-

Danny Steinberg:
Yeah, well-

Max Steinberg:
… it’s upside. Right? You want…

Andy Baldacci:
I had no idea what he was doing. I thought [crosstalk 00:00:41].

Danny Steinberg:
Why do you want me to look higher or something?

Max Steinberg:
No. I was saying that when you’re taking a stand a low-end player it’s like, why do you like any player in any sport? If you’re talking about [inaudible 00:00:53] is the high-end side, right? You want [inaudible 00:00:57] they can have a big game. So if you’re playing a lower end player, these are going to be the guys whose average projection is not max they’re upside. People think, this person on average is going to do this, but they have a big game at them. Right? Or, I mean maybe they’re going to be underrated for other factors, but that’s what you’re looking for as you’re saying. Okay. For some specific reason I think this player has higher upsides than people think and maybe SaberSim is saying that too.

Max Steinberg:
So that’s those are going to be the players where I’m saying, okay I think this player can have a big game and that’s why I’m taking a stand by them. It’s not because I think his average projection is high, I think is his upside is really high.

Danny Steinberg:
Yeah. I think, I mean sometimes I’m just surprised. I kind of let say SaberSim build my lineups or let the builder build my lineups based on whatever objection it’s using and sometimes I’m just surprised that too I’m using. You know a lot of times the best low end plays are just going to be guys who are really practical. Like you don’t necessarily have to have the most genius idea in the world to have a good low on play. Like you just have to sometimes something someone just really practical is getting very overlooked for some reason and it may not be why. I mean someone in like a really bad matchup, for example, they could still do really well or you know there’s many reasons why someone could be low on but still be worth playing. But often I don’t know who I’m taking a stand on until I use the SaberSim lineup builder and see who’s actually showing up in my lineups.

Andy Baldacci:
And I guess on that note, what are you looking at when you do see someone surprising in there and like you’re getting a ton of a specific player. Cause that definitely does happen. Probably… Not probably it does happen more often in SaberSim than a traditional optimizer where we are just going to be out of the box super heavy on a couple of guys that does seem out of nowhere. How, what like questions are you asking yourself? What sort of things are you looking into? What are you guys doing in those situations to evaluate whether something’s wrong with the model or maybe we’re onto something.

Max Steinberg:
Right. I’m looking at their projection and their ownership projection because you’re saying, okay is this person’s… I mean it’s, it definitely you don’t want to take it at face value cause maybe their projection is too high or maybe their ownership projection is too low. Right? So you want to question whether it’s the correct play. But you don’t want to question whether it’s the correct play, if you check the projection, you check the ownership, it looks very-

Andy Baldacci:
It all checks out.

Max Steinberg:
It all checks out. You’re using the proper sliders. Like if that’s all the case, that just means that player’s a good play. It doesn’t mean that you should try to get less of them, like lowers max exposure, it just means there’s a… he’s projected right, his ownership projections right, and there’s a reason you’re getting him and he’s probably going to be a very profitable play if he is in fact projected be really lot.

Danny Steinberg:
Yeah, I think a lot of times you’re able to understand it. Like I’ll look through my projections and be like, Oh, this guy is going to be high owned, but I’m getting a lot of this guy and he’s actually just a little lower projected and is supposed to be way, way low round. So that, makes a lot of sense. Or maybe, Oh, you know what this is a leverage off of some very popular on play and that’s why it’s doing it. Sometimes when I’m using SaberSim’s builder, you get that sort of outcome where, you know it’s leveraging out of some high owned player having a bad game, in some way.

Andy Baldacci:
Yeah, and I think this, you guys both touched on this a lot more in kind of the videos you did on the quality control, the third step of the ladder building process for basketball and football, where you just talk about what to look out for, but just to emphasize some of that here. It’s just looking at those outliers where it’s looking at the people who you’re getting way more of than you expected. It’s checking to make sure their ownership and their projections are aligned, but going further, you can also click on their name in SaberSim, bring up the distribution of their outcomes, and you can see the correlations, and see if maybe we have them correlated to someone else who is in a lot of your lives positively or negatively correlated as someone who is going to be highly owned or whatever it may be.

Andy Baldacci:
It’s… There’s a lot of things you can look into, but those are types of players to check out when you’re in that kind of quality control phase and you can do the same thing for, Oh, I really thought I would have whoever it may be in my lineups, so I’m not getting him at all. You can hit the show players box and then see him and just kind of dig into why that may be. Maybe there’s someone else who’s somewhat projection cheaper, whatever it might be. There’s a lot in there, and it can be confusing because it can seem like a bit more of a black box in traditional tools. And so that’s why we always try to recommend is: digging into it, looking at all the data as best you can, but when you have those specific questions, you can always shoot us an email, reach out to us in Slack, and we’re happy to kind of help diagnose those things. But hopefully this gives you a good idea of what to look for in most situations.

When do you build your lineups?

Andy Baldacci:
Is there such a thing of setting your lineups too early? And the answer is yes. It depends on the sport, but for basketball specifically, you want to wait until it was close to lock as possible and even then you’re still going to miss some things. The later news comes out in the sport, the later you want to wait to build to your lineups. I think there’s nothing wrong with, if you don’t have a ton of time of doing a build early, just to get an idea of where things are going and who top plays look like, that sort of thing. So that when you’re in there for your final build, you’re not just coming in blind. But for sports like basketball, you want to be as close to lock as possible.

Andy Baldacci:
Football, they put out inactives pretty consistently about an hour and a half before kickoff, so you want to wait until that, but even there, that’s less dangerous. There’s usually not many surprises. You can exclude questionable players from your builds and probably not give up a ton, but that’s one where it’s on the line a little bit.

Andy Baldacci:
For baseball, lineups, it’s not as regimented when they come up, but those usually come out far enough in advance where you don’t need to wait till the last minute. But basically all of this comes down to wait until all the information is confirmed to finalize everything, but there’s no harm in getting in there early just to see where things are shaking out. Does that seem reasonable enough?

Danny Steinberg:
Yeah, no, that’s really well said. I think another advantage to doing… I think, if you’re going to prioritize time to do lineups, I’d definitely try an hour before lock or in that window. If you play DFS a lot, like I do, there’s been three or four times where I’ve had complete mistakes where I forgot to put in lineups or something or forgot to do a swap and just had totally dead lineups, so there’s no harm in doing an early swap or getting some good lineups in early just to play it safe, just in case something terrible happens and you’re not able to swap in time. Getting something in early is always a good idea, but you should also try to get a final run in within an hour of lock, at least, or as basketball 15 minutes before lock. But yeah, baseball and football, an hour out you should be fine.

Andy Baldacci:
And I’ll try not to go on too much of a rant here, but the short of it is just do not do dummy lineups. Like when you’re reserving your entries, if you want to do a single entry for that and just put it in 150 times, whatever, just get your lineups in there, that’s fine. But immediately after you do that, do a quick build with an optimizer, received some whatever, just something so that if the unit goes out, if something happens with our site, with another site, whatever it may be, you’re not completely screwed and running a train of whoever it is and leaving $20,000 in salary on the table, like get in some build early. If you are reserving lineup. That is probably the most costly thing you can do. And I think that might’ve just made Max remember some bad times.