Welcome to the third part in my series documenting my journey of getting started with daily fantasy baseball. In the first segment, I did some introductory background reading. In the second segment, I took a look at some common statistics that you might be inclined to look at but that have been proven to be misleading (Batter versus Pitcher stats and players in the midst of hot or cold streaks).
More Background Reading
The book is a tremendous resource if you’re like me and just getting started. I presume it’s also quite helpful even if you’ve been playing the daily game for a while.
If I’m being honest, you will save yourself a lot of time by buying his book instead of waiting for me to slowly process my own way through DFS and writing about it. Once you read Bales’ book, then come back to me to see how to implement the topics covered in an Excel solution. Bales has created an excellent beginner’s guide to daily fantasy baseball by giving you strategies, tools, websites, and outlining the exact information you should be looking at when generating lineups.
One topic I’ve seen mentioned in the book and repeated many times online is that to be successful at DFS you need to play the game regularly (“grind”) and as such you need to develop an efficient routine that allows you to create quality optimized lineups regularly. My hope is to use Excel (and maybe other technology tools) to do this.
My “Aha Moment”
There are two significant benefits I got from “Fantasy Baseball For Smart People”. The first is that it’s a convenient and complete package. You could probably get nearly all of the information from the book by reading through the various articles I linked to in Part 1 of this series. If you go that route you even get the information for free. But this book will save you time and puts them in a nice easy-to-use and convenient package.
As much as I love technology, I’m still a “book-in-hand” kind of guy. So I bought the paperback, which lets me dog ear pages I want to remember and reference easily in the future. There’s no easy way to do that with the 20-30 articles I linked to in Part 1.
The second big benefit I get from the book is an insight I did not see anywhere else in my prior readings.
The goal isn’t to maximize points (in your lineup), but to maximize win probability.
~ Jonathan Bales – Fantasy Baseball for Smart People
Maybe this is obvious to everyone, but it wasn’t to me. It’s a significant difference between season-long leagues and DFS.
I can’t envision a scenario in a season-long rotisserie league that you would not want to maximize your projected points in the standings. It’s a simple concept that having the most points in the standings gives you the greatest likelihood of winning the league.
But there’s an interesting wrinkle that I alluded to in Part 1 that changes things in DFS. Player selection is not mutually exclusive. I know, I know. There I go again using math words. This is a fancy way of saying that the players you choose for your team can (and inevitably will) overlap with players chosen for other teams.
Bales’ point can be illustrated by a simple example…
Imagine a world where Tanner Bell has created the perfect DFS spreadsheet (I’m imagining it right now). It calculates the most accurate player projections and spits out the best possible lineup for a given day with the simple press of a button. SmartFantasyBaseball.com has become so popular that everyone knows about this amazing Excel file.
You use this tool and enter a contest with 1,000 total participants (you and 999 others). The contest costs $10 to enter and all $10,000 of prize money will be distributed to the highest score.
If every one of the 1,000 entrants in the competition is using the amazing Excel tool, all of them are going to end up tying with the same high score, right?
That’s Where Game Theory Comes In
What would happen if all 1,000 entrants tied with the same score? We can assume the prize pool would be evenly distributed and everyone would get their $10 back, right?
So if you knew all of this, what is your best move? Is it to join this contest and enter the same lineup as everyone else? Or should you enter a slightly different lineup and hope the spreadsheet is a bit wrong on a given day?
Baseball is a funny sport. We can do a very good job of predicting what will happen over the course of 162 games. But it’s very hard to accurately project what will happen in one specific game.
So even though this infallible Excel file is very accurate, it can’t perfectly predict what will happen on a given day.
For this reason, if you know everyone else is going to submit a certain lineup, your best move is to submit a slightly different lineup. Even if you think that same lineup is the most likely to score big, your expected winnings would be less.
A 99% chance of winning your $10 back gives you an expected return of $9.99 (0.99 * $10), while a 1% chance of being the lone first place finisher gives an expected return of $100 (0.01 * $10,000).
But That’s Not Going To Happen
I know. All 1,000 people are not going to enter an identical lineup. But on a night where the Rockies are playing at home against an awful visiting starter, could 20%, 30%, or even more participants than that have a Colorado lineup stack? I honestly have no sense of what level of ownership we could be talking, but you can see that it would be difficult for your lineup to stand out among a crowded landscape of Rockies stacks.
Maybe We Should Look at Batter Vs. Pitcher Stats?
I know what you’re thinking. “Whoa! Wait a second. You just had me read 3,000 words about why I should NOT look at BvPs and now you’re backpedaling?”
Think about this. An element of game theory study is that your decision-making process is interactive between you and your competitors. You attempt to make the optimal decision in light of what you expect your opponents will do.
So if we know there are BvP disciples roaming the DFS landscape, we can use this to our advantage. Even those that don’t believe in BvP often will say, “I use it as a tie breaker to decide between otherwise closely valued players.”
I’m not saying you build a lineup around guys that are 5-for-10 against random pitchers. But there is a possibility to do a brief study of the BvP matchups for a given day and use that as insight into who other players are going to select for their lineups. Will this be a huge advantage? Likely not. But I don’t get the sense that winning at DFS is about easy-to-find large advantages. It’s about identifying numerous small advantages and putting as many of those in your favor as possible. The “Extra 2%“, if you will.
Another way to get a sense for players likely to be largely owned would be to check the most popular fantasy sites for their daily lineup picks. Every major site is doing this. If several sites are pushing the same player, note that.
How Do We Weigh Making Our Strongest Lineup Versus Our Creating a Lineup Most Likely to Win
Another neat feature of the book is that Bales was given access to DraftKings’ lineup data. He breaks down what strategies (at least on DraftKings) work the best and is able to demonstrate this concept that you’re often better off with a lesser owned lineup.
He did this by examining lineup stacks. The most popular team lineup to stack during the 2014 season was COL (of course). Take a look below:
|Team Stacked||Average Points||% of GPPs Won||% of Lineups Won|
I don’t want to show the whole table (it’s not my content to give away), but the book is packed with many similar tables of information you can use to plan lineups.
Think about what this shows. Yes, lineups including several COL hitters SCORED MORE points (nearly 10 more). And they even won more tournaments (5.4% of winning lineups had a COL stack). But playing a MIL stack was actually more beneficial, or more “efficient”. Individual players that chose a MIL stack WON more frequently than lineups that played a COL stack. This is simply due to the greater number of people playing COL stacks.
I must tell you that I’m overstating this some. Using a COL stack was still your second highest chance at winning, even despite the fact that everyone knows about this strategy. But it does illustrate an interesting point. It can help to know what others are doing when you’re building your own lineup.
I think this is a good time to go back to Bales’ quote from earlier: The goal isn’t to maximize points (in your lineup), but to maximize win probability.
Winning does not just come down to the number of points scored. Especially for guaranteed prize pool contests, it’s also a calculation of what you expect others to do.
How Does Contest Format Fit Into This?
This is a great question and is very important. Is our answer to the questions above the same for a 50-50 contest as it is for a GPP?
I don’t think it is. Notice how the table above presents % of GPPs won. Not 50-50 contests.
In a contest where your goal is to place in the top 50%, you don’t need to be (as) mindful of what others are doing, because half of the entries are not going to be identical to yours. You only need several subtle differences away from a commonly used stack to differentiate anything. If going with a Rockies stack in COL is obvious on a given night and seems like the best play, it’s not going to hurt in that contest format.
Additionally, in a 50-50 format you only need to finish in the top half of entries. Because the bar is so much lower, but in a contest where you need to scratch and claw to get into the top 20%, you do need more differentiation from the masses and it can help you greatly to be mindful of the “obvious” plays.
Bales borrows the phrase antifragility, from Nassim Taleb, to further delve into the concepts of volatility and using the decisions of “the crowd” to your benefit. By volatility, I’m referring to the unpredictable nature of projections when you are looking at one specific night of a baseball season.
If the topic of DFS strategy hasn’t sent you down enough of an internet reading rabbit-hole, take a few minutes to search Taleb’s background, investment approach, and other thoughts. You’ll quickly find that Taleb has made a name for himself by profiting when “things go wrong”. An overarching premise of his investing approach is that we really cannot predict the economic future and we are bound to encounter extreme situations. Whether these situations are extremely good or extremely painful, they do happen. They also happen much more frequently than anyone seems to realize and nobody is ever prepared for them, and we cannot predict when these extremes will occur.
Taleb’s concept of antifragility is to anticipate and gain from disorder. An example is the recent mortgage crisis experienced in the United States. Very few people had any idea that a problem was approaching. Nobody could have predicted the scale of the crash and the way it would ripple through the entire economy. In fact, many of us were told home buying was the safest investment you can make and declining home values couldn’t happen (getting married and buying a home in 2005, the peak of the bubble, in the metro-Detroit area, really worked out swimmingly for me). Even if you weren’t savvy enough to identify specific economic indicators changing and suggesting a crash could occur, you could have made small investments that would pay out enormously if a crash were to occur.
Bales likens this to below average pitcher going in to Coors Field. He even gives an example of how he won a tournament using this exact approach. When a poor pitcher starts at Coors, the DFS message boards light up with traffic stating the Rockies are an incredibly safe investment (just like investing in a home in the early 2000s). The majority of tournament lineups will contain a COL stack so they don’t miss out on this “safe” investment. There’s no way this pitcher could shut down the mighty Rockies.
Well, just like the market crashed horribly, the Rockies can indeed be shutdown by inferior pitching on any given night. Is it particularly likely to happen? Not necessarily. But it does inevitably happen. And you can profit from these scenarios, that people think are impossible, by structuring your investments to take advantage of the problem, should it arise. Because people underestimate the likelihood of these events, the cost to acquire these alternative investments is usually pretty low.
Another Example of DFS Being Like Investing
Stick with me for a minute as I move away from the book we’ve been discussing and into a podcast I listened to about the strategy behind playing in the DFS Masters Golf Tournament for a $1 million prize.
Yes, a huge gear shift there. And I don’t even know exactly how I stumbled upon this podcast, but it had something to do with the fact that the guest was Drew Dinkmeyer, another DFS pro that also does quite a bit of teaching the trade in his writing and podcast appearances.
The topic that came up on the podcast (at about the 18 minute and 45 second mark) was how to strategically enter multiple lineups into one tournament. Granted, this was for a huge golf tournament, but I think the principles still apply to the discussion we’ve been having about antifragility and game theory. You can use the link above or the embedded video below to get the full interview, if you wish.
During the interview, Dinkmeyer gives the example of viewing your multiple lineups as a portfolio of investments. He uses the example of playing 100 lineups in a contest. I want to focus in on a couple of snippets from Dinkmeyer:
If you have a really strong feeling on a player, you want to own more than the field (‘s ownership percentage), even if it’s a player that will be highly owned… Similarly, if you like a player a little bit but you’re really not sure of them… you want to try owning less than the field…
If you really don’t know at all and they’re going to be super high owned, it makes sense to have a piece of them, but the odds are on your side to take a fade route. So if you think a player is going to be 30% owned and you’re really not sure their ability to outperform their price tag is greater than 30%, then you might want to only own 5-10% and take an active position against them. There’s more upside to fading higher owned players than there is downside to fading lower owned players.
I want you to focus on the suggestion that you want to own players where you believe the likelihood they exceed expectations is greater than their ownership percentage and you want to avoid (fade) players where the likelihood they reach their projection is less than their ownership percentage.
Let’s Put All of This Together
Going back to Bales’ writing on Taleb, we know that people underestimate the likelihood of extreme events happening. If we combine that with Drew Dinkmeyer’s advice, we would now have a very strong strategy to use while playing GPP tournaments (where the prize money is concentrated to those that finish very high in the standings).
Let’s go through a very simple example to illustrate these points. To borrow from recent events, lets assume a power hitting and healthy Toronto Blue Jays lineup, full of right-handed batters like Edwin Encarnacion and Jose Bautista, is facing a struggling young left-handed pitcher like T.J. House. That game happens to be one of the more obvious plays of the day and you have seen it recommended on multiple websites.
On that same night the Detroit Tigers, one of the best offenses in baseball, are going against Chris Sale of the White Sox. On those same websites you’ve seen allusions that you should avoid the Tigers due to the matchup with Sale.
You consider the likelihood of House shutting down the Blue Jays, and you put the likelihood at 10% that he’ll have a good outing (remember, this is only an example). You also consider that there is a 20% chance that the Tigers can chase Sale from the game early and still put up a strong offensive performance.
A little refresher of my college statistics class tells me that the probability of two independent events happening is the product of the two probabilities. So in this hypothetical exampler there is a 2% chance that House shuts down the Blue Jays AND that the Tigers perform well offensively against the White Sox (10% chance of House * 20% chance against Sale = 2%).
Combining Dinkmeyer’s thoughts with Taleb’s, it is very unlikely that this combination (fading the Jays and buying the Tigers) will be played in 2% or more of lineups. This seems like just the antifragile approach Bales suggests as a viable strategy in tournament play.
Operate At The Extremes
For all intents and purposes, a strategy like that mentioned above has two outcomes… You’ll finish last or you’ll finish near the top of the tournament. But this is not a bad thing in a GPP format. There is no benefit from finishing in the middle of the pack. Finishing in the 51st percentile is no better than finishing in the 1st percentile. If you plug away at a spreadsheet for hours to find the optimal lineup, like many others are going to do, it’s unlikely you will separate yourself from the pack. But embracing the variability in a risky play that is clearly not “optimal” in terms of raw projected points can improve your expected outcomes.
One Last Thought
All this talk of probability and trying to separate yourself from the crowd reminds me of some lessons I took away from the book “Thinking Fast and Slow” by Daniel Kahneman. There was an entire segment of the book devoted to the fact that humans are terrible at comprehending probability.
Given that we are bad at understanding probability and that there are a lot of incentives or shortcuts that point people to always developing the “optimal” lineup, it really would seem that there is an opportunity to be exploited in “optimizing your chances of winning”.
Want More DFS Talk?
Be sure to follow me on Twitter. Again, I’m not an expert. But I’d like to think we can come at the DFS game from a slightly different angle than most of the other information that’s out there.
What is the most informative DFS strategy article you’ve come across? Please let me know in the comments below or on Twitter.