DFS Strategy: Why We Should Be Fading “Hot” Batters in Tournaments
From time to time, I’ll take a step back from the day-to-day grind and evaluate some of my decision-making processes. It’s easy to get caught up in the daily cycle of grinding through information and lineups. I find that challenging some of our core beliefs in DFS can prevent stagnation and expose flaws in our decision-making process. For some, this topic will be old news and for some it may provide a new way to look at your lineup building process. Either way, this is my case for why we should be swimming upstream on “hot” batters.
Before you read the title and get your pitchforks and riot gear ready, take a few minutes to hear me out. The concept of fading “hot” batters has brought me success in the past, and I think it’s a theory that can become lost because it forces us to challenge our in initial instincts.
“You want to be greedy when others are fearful. You want to be fearful when others are greedy. It’s that simple.” – Warren Buffett
For the purpose of this article, I’m going to presume that everyone understands the accepted principle that being contrarian is essential to DFS tournament success. If this statement is foreign to you, there is great material out there that can further explain this idea. In short, we make money in tournaments zigging, while everyone else is zagging. If you’re creating lineups without this process in mind, you’re likely to end up with a fairly chalky lineup and therefore, negative expected value. Think of our good pal Ricky Bobby. With top-heavy payouts, in order to profit long-term in tournaments, you need to take the stance that “if you aint first, your last.”
Variance in MLB DFS
In comparison to all other sports in DFS, MLB has the most variance. In any given NBA game, its safe to say, excluding injury or ejection, Lebron isn’t going to go 1 for 25, with no assists or rebounds. However, in baseball, it’s not uncommon for a star player to go 0 for 4. The scoring in MLB DFS highlights this variance. If a star hits a ball 6 inches short of a home run for an out, he is awarded no points. Yet, if he hits that same ball 6.1 inches further, he is given 14 points. We can say that both events were well hit balls, yet how they are rewarded is drastically different based on seemingly insignificant factors.
When analyzing players for DFS purposes, our hope is to eliminate as much variance as possible. That means that we have to dig deeper than observing surface statistics that contain variance. Theoretically if a star player has 3 back to back games where he hits the ball that extra 6.1 inches and then 3 more games where he falls short. Would we say that he was “hot” and now he is “cold”? Probably not. We’d agree that his performance in these 3 game series is virtually identical, but it can be lost in translation when simply reviewing fantasy points.
Each upcoming event is independent from the previous event.
I find it best to work in practical examples. Lets say Mike Trout goes 3 for 4 on a given night. There is no empirical evidence that leads us to believe that the odds for Trout’s next at bat have changed. Of course factors such as lefty righty splits, weather, ball park, opposing pitcher, etc. will change the probability of him getting a hit, but as far as my knowledge goes, there is nothing that proves that Mike Trout having a good game on night 1 will influence him to have a good game on night 2. It’s time to remove the old superstitions and biases in our decision-making process.
The problem lies in the lack of a significant sample size. Mike trout hitting 41 home runs last year, provides us with a large enough sample size to assume that he has power and will continue to hit above the league average in home runs. However, if Trout hits a home run in game A, the sample size is too small to assume that he is “hot” and think he’ll do it again the next night. On any given night, a batter only provides us with 4 or 5 data points. In order to create accurate predictive models, we need hundreds of these data points. This is where the sports bar talk of a “hot” batter falls flat. If Trout has three great games back to back, the sample size is too small to say that he’s “hot” and has a better chance of a good performance in game four.
Even if somehow this reasoning is proven wrong. Theoretically speaking, what sample size determines when a batter is “hot”? 3 games? 4 games? 10 games? What if a batter has a good night, but then follows it with a poor night? We don’t have the ability to predict when a batter will change from hot to cold. Therefore, even if batters do get “hot”, its of no worth to us because we have no way of predicting it and monetizing it as a strategy.
Let me address some narratives that may have validity. Keep in mind, that to there is no statistical literature out there to prove these as fact, but I give these more weight than just saying a batter is “hot”.
Batters are constantly adjusting their swing and this can be evident when a batter goes from a “slump” to performing better than average (Again, it has to be over a large enough sample size). However, how can we predict this and use this for our benefit? We cant. Unless you’re the hitting coach or somehow in the locker room, it’s going to be difficult to say with any certainty that its worth investing in a player because his swing is now improved. Another narrative that may hold some validity is that a player is now healthy or vise versa. This could happen after an All Star break, or time off. I do think that notable rest/health can increase performance over a significant period of time to cause a batter to be “hot.”
Another popular theory is that a batter can become confident. Again, I agree with this idea, but how do we, as investors, determine when a player is confident and when he’s no longer confident. What if his confidence level changes from at bat to at bat? Also, who’s to say that performance has a direct correlation to confidence? I’m sure when David Ortiz is rehabbing an injury in AAA and facing a scrub righty, he has confidence, but does that mean he’ll homer in every at bat? Ultimately, it’s up to you to determine when a sample size is considered significant. If a batter struggles for a month or more, maybe he is getting worse, just err on the side caution when evaluating a player off of a limited sample size.
In summation, the thought of saying a player has had a string of two good games and is therefore a sufficient predictive indicator of his performance in the upcoming game is a false presumption. We need to be conscious of the lack of sample sizes that a batters provide us with and it’s crucial to know how to effectively use them in order to not fall into traps of outliers and meaningless analysis.
Playing off of the strategy of owning a “hot” batter, is the compliment of recency bias. Recency bias is in our nature as humans, and more specifically sports fans. If you still say, “I think the whole independent events thing is BS, I’m going to play the hot batter,” let me show you a different example of how this is also another negative expected value strategy. More people than not subscribe to the notion of a “hot” batter theory. I’m not sure if lazy would be the right word, but humans are inherently going to take the path of least resistance.
So if you can turn on Sports Center and hear them say a batter is “hot” and you should play him, you may settle and determine that their analysis can suffice. While it may not be this exact example, only 10% of people are making money in DFS and its safe to assume that the other 90% is incorporating a negative EV strategy such as playing a hot batter. So while you may think you may have the best play in the world because a guy has had 10 hits in his last 5 games, so do a million other people. I’ll implore you to take a look at ownership levels in tournaments and then check to see if the batter is deemed as, “hot.” You’ll find a strong correlation. If a high-profile batter hits two home runs on national TV, I guarantee you, he’ll have a noticeably higher ownership level in his next game. Go ahead. Check what Max Kepler’s ownership level is tonight after his three homer barrage last night.
Be the lion among the sheep.
If you’re still calling fugazi, then I’m not sure how else to help you, but I’ll make my last-ditch effort to convince you. Lets expand on our first example and say Mike Trout has a terrific three game series. DraftKings pricing algorithm takes recent performance into account and makes adjustments accordingly. It’s safe to say, in the given example, that Trout would be priced higher than he was previously. If the points above are true, then we would be paying up for a player in which we have no legitimate predictive ability, while everyone else is also doing the same. That sounds like a fairly unprofitable strategy to me.
Want your voice heard? Join the Fantasy CPR team!Write for us!
Instead, what if we target players who we know are of adequate skill level (Large sample of recent date, possibly a year to two years of data) but have struggled of late. One example of this could be Giancarlo Stanton. Stanton struggles throughout most of the month of May and early June.
More from FanSided
- Ohio State football’s fatal flaw exposed by Indiana Hoosiers
- Dez Bryant promoted to active roster by Ravens
- Alabama’s Najee Harris stiff-armed Kentucky defender into 2021 on long TD run (Video)
- Chiefs bring back former Super Bowl champ to help protect Patrick Mahomes
- Denver Nuggets avoid offseason disaster by retaining Paul Millsap
We want to buy shares of Stanton just before, and as he begins to bounce back. His struggles not only decrease his ownership but also his price. Taking a financial approach, the old adage of, “buy low and sell high” is fairly comparable. If we timed this process out correctly, we would be getting value when Stanton got back to normal. Over a large enough sample size, a player is almost always going to regress (positively or negativity) back to the mean. If in the first two weeks of the season Josh Donaldson goes 8 for 54, hitting .148%, with no home runs, would we assume that he’ll continue to hit sub .200% and never hit a home run? No, because the sample size of two weeks is too small to make an accurate judgement.
It’s easy to fall into the trap of playing a batter because of his recent form. With our ever-expanding access to information, its crucial to comb through the noise of misguided strategies. Be mindful of the next time you’re encouraged to play the “hot” batter.
Note: Cash games are a whole different ball game and this logic shifts drastically as contest type change. They also change as we change the stake levels. Higher stakes contests are going to be more immune to these simple tactics as the competition is sharper. Every slate, every player, every contest is situational. Most of the times, these are helpful strategies to keep in mind, but everything has an outlier and unforeseen variables.