October 20, 2017

Measuring fantasy football performance using consistency reports. Addison Hayes uses standard deviation for weekly fantasy point performance to find consistent starters.

Last week, I introduced a new way to look at consistency in fantasy football based on standard deviation. The method was very well-received on Twitter and Reddit alike, but some had trouble drawing conclusions or making predictions from the raw standard deviations.

This week, I would like to present a new way to quantify a player’s weekly floor and ceiling using standard deviations. This adds another layer of fantasy applications to this method of consistency.

To read last week's initial consistency report article, click here.

How Do We Determine Floors and Ceilings for Fantasy Players?

To introduce how I am calculating a player’s floor and ceiling, I must dive into some math. Now before you exit this article at the sight of the word math, let me explain. Essentially, the fantasy points scored by a player, over time, follows what is called a normal distribution, or a bell-shaped curve. With the normal distribution, statisticians use the empirical rule of statistics, which explains what percentage of observations fall within one, two, or three standard deviations from the average.

The picture above depicts the empirical rule using a normal distribution. What it means is that 68% of the data will fall within one standard deviation (on both sides) of the average. 95% of the data falls within two standard deviations, and 99.7% within three. Now you may be thinking, “Addison, this is a fantasy football article not a statistics class” and you’re right! So where does this apply for us fantasy football players? As I mentioned before, this can help us quantify a player’s weekly floor and ceiling. Since 68% of the observations fall within one standard deviation of the average, all we do is add or subtract the standard deviation from a player’s average points per game to get their floors and ceilings. In short, a player is going to score points weekly within a certain range, on average, in 7/10 weeks.

We can take this a step further by finding a player’s true floor and ceiling by going two standard deviations from the average to account for 95% of the observations. This gives us a very wide range of outcomes though, and is harder to use for predictions. Additionally, we can go one final step farther and find each player’s absolute floor and ceiling by going three standard deviations from the average. This would account for 99.7% of all observations, which is essentially every week. However, this range is so wide that it is nearly impossible to draw conclusions from for fantasy purposes. Knowing this, we will stick with only one standard deviation from the average, which is a nice range to work with.

Just like last week, I will breakdown the updated Week 6 consistencies by position, but also include each player’s weekly floor and ceiling. I filtered each position by at least three “startable” games, which I define as top-24 for QBs and TEs and top-36 for RBs and WRs. I additionally filtered tight ends by having at least one top-12 performance.

Quarterbacks

Above are the 27 quarterbacks who made the cut, sorted by the highest floor. Let’s look at the first name on this list to illustrate floors and ceilings. Dak Prescott, through six weeks, has been one of the most consistent fantasy QBs this year. he is averaging nearly 22 points per game with a standard deviation of 4.92. Using this, we see that Prescotts’s floor is 16.72 and his ceiling is 26.56. This was found by subtracting 4.92 from 21.64 (floor) and adding 4.92 to 21.64 (ceiling). Now, Dak owners may be saying, “wait, Dak scored me 29 points in Week 5, so how is his ceiling only 26-27 points?” As I mentioned before, this range only covers 68% of his performances. His 29-point performance against the Packers would fall in the 32% of observations outside this range. If we were to extend the range (enough to cover 95% of his performances), his 29-point performance would be covered by the spread.

You might be asking why we don’t just make the range wide enough to cover 95% of the performances. To answer that, I give you Andy Dalton. Dalton is averaging just under 13 points per game with a standard deviation of 10.13. If we were to account for 95% of his performances, that range would be between -7.38 and 33.14. The range is too wide to be able to use on a weekly basis.

Looking more at the chart, we can see which QBs are more reliable on a weekly basis than others, while also seeing which players have the potential to win you a matchup. One such player who can win you a week (and probably already has) is Deshaun Watson. According to the chart, Watson has the highest ceiling of any QB through six weeks at 32.72. What makes Watson even more impressive is that he also has the sixth highest weekly floor among QBs due to his rushing ability. This makes for an elite passer for fantasy who has now become a weekly must start.

On the opposite side, we see Eli Manning towards the bottom of the list in terms of floor. He is currently sitting on an average of just over 15 points per game, with a floor of 7.61. There may be many explanations for Manning’s poor 2017 performance, but in reality, he is just not getting it done for fantasy owners. He may be able to have a stud performance once in a while, but more times than not he has been subpar at best.

Running Backs

Moving on to running backs, we have a deep list of names to analyze. Using the same analysis as we did with QBs, we see that rookie Leonard Fournette is currently boasting the highest floor among running backs. In fact, the first eight names should come as no surprise, as they are the elite RB1s of 2017. Fournette has surprised me with his elite fantasy performance so far through 2017. He is averaging 21.7 points per game, but has been one of the most consistent rushers in the league. This makes for an elite weekly plug-and-play who you know is going to produce for you.

On the other hand, we see offseason darling Jay Ajayi sitting in the middle of the pack. He is averaging only 9.28 points per game, mainly because he has not found the end zone yet in 2017. I expect that to change as the Dolphins continue to find themselves on offense, making Ajayi a great buy-low at the moment.

Continuing down the list, we see two offseason hype players in Isaiah Crowell and Ameer Abdullah sitting next to each other. Both have been disappointing fantasy players through the first six weeks and the numbers show it. To Abdullah’s credit, he has performed better than Crowell, but their floors are almost identical. Crowell is actually the second-most consistent back on this chart in terms of standard deviation, but he has been consistently terrible. Hopefully for fantasy owners, there is nowhere to go but up for these two.

Wide Receivers

Looking at wide receivers, I managed to chart a bigger list of receivers than last week by cutting down on the filters. Going down the top of the list in terms of floor, nine of the top ten names should come at little to no surprise. The most surprising name among them is actually the first name on this list: Will Fuller. I guess this is what happens when you catch five touchdowns on eight receptions through only three games. Although Fuller has been very consistent so far, I do not expect this trend to continue. Is he a relevant fantasy asset? Definitely, but his fantasy production has been skewed by random touchdowns with little productivity outside them. The law of averages will catch up to Fuller in due time.

In fact, Fuller’s name at the top of this list highlights a flaw in my methods. So far, I have only been using data from 2017 to calculate a player’s standard deviation. Since we only have a handful of observations per player, the data can be very susceptible to trends and outliers in its young state. To remedy this in the future, I am currently working on calculating each player’s career standard deviations to go off of each player’s career average points per game. This will give us a more accurate look at a player’s true consistency throughout his career. (Bear with me, as it takes time to gather and analyze the data needed to do this).

Now, back to the wide receivers, the list is essentially the same as QBs and RBs. As we move down the list to the bottom, we see another name that really sticks out, Stefon Diggs. I highlighted Diggs last week, but I would like to highlight him again here. Diggs still remains the most inconsistent receiver on this chart (this is not including the games he missed due to injury). Because of that, his floor is extremely low (3.72), but his ceiling is actually the highest among receivers. If that is not the definition of boom-bust then I don’t know what is. He is an excellent receiver, but his knack of disappearing some weeks, coupled with injury concerns, makes for a risky fantasy asset.

Tight Ends

Finally, we look at the tight ends. This position has been a jumbled mess through six weeks after the first three or four elite players. If you don’t own Rob Gronkowski, Travis Kelce or Zach Ertz (or Charles Clay before the injury), you would be better off streaming the position every week looking for touchdowns. This is where I believe the power of standard deviations really takes hold.

Outside of the elite three (Gronk, Kelce, Ertz), I would like to highlight two other names who have come onto the scene as near every-week starters: Cameron Brate and Austin Seferian-Jenkins. Both players are at the top of the second tier of tight ends and have each been very consistent over the last few games. Because of this, they both now offer the third and fourth highest floor among tight ends, higher than Jordan Reed, Kelce, Delanie Walker, and Jimmy Graham. In terms of ceiling, Brate is near the top at over 20 points and Seferian-Jenkins’ 15.8 is nothing to sneeze at either (though it should be higher but I won’t get into that). These two players should be in the conversation for weekly top-10 options at the position, especially given the lackluster players behind them.

Moving down the list, I would like to point out one more name: Hunter Henry. Henry owners have probably been disappointed with his weekly play this season, as they should be. However, he has gotten more involved the past three weeks, catching two touchdowns and even posting a 5 reception/90-yard performance through that span. Henry will continue to be a touchdown play, making his weekly floor lower than others, but his upside puts him in TE1 territory every week.

Overall, I have presented a new way to quantify a player’s floor and ceiling using standard deviation. This adds a new layer to this method of consistency that makes the numbers easier to analyze and draw conclusions from. I will continue to add more layers to this consistency method that will make it easier to use and predict production on a weekly basis. Stay tuned!

More 2017 Fantasy Football & ADP Analysis