The Consistency Reports: Quarterbacks
The question of valuing players is the seemingly never-ending issue when it comes to fantasy football. Should you target a player with upside, or should it be someone who might have a lower ceiling but provides you with consistency week to week? What about if you had a way to combine both elements in one ranking? Well, here at FFStatistics we have attempted to do just that with our Consistency Rating (COR). COR is created by taking a player’s calculated ceiling and dividing it by their coefficient of variation (more information on these and the math behind them at the bottom of the article). If you do not fancy doing the math yourself, then our consistency tool provides you with all the data. As an added bonus you will also find out how often players scored within certain brackets, top-five, top-12, 13-24, 25-36 and beyond.
Using COR, we can get an easy snapshot of where players rank on a weekly basis. We can also compare their COR value to their average fantasy points total. By looking at the difference between the two values, we can identify various subsets of players. Some players posted an average which looked great but using COR we can find out whether they were reliable or not on a week-to-week basis. The same goes for players who posted average points totals which were not great. Some did it and were wildly inconsistent, others were actually pretty reliable.
This article will take a look at the consistency of the quarterback position as a whole. Then we will dive into a couple of players whose ranking by COR differed from their ranking based on their weekly average points total.
2018 Consistency Ratings
|Player||Team||G||Average||Std Dev.||Weekly Floor||Weekly Ceiling||CV||COR||COR Rank||OVR Rank||COR vs. OVR|
Seeing Patrick Mahomes atop both the COR and AVG rankings will not surprise anyone. Matt Ryan or Andrew Luck should not surprise anyone either, but Ben Rothlisberger is a different matter. Roethlisberger has been a high variance quarterback in his career but this season he ranked fourth in COR, having been outside the top 10 in each of the last three years.
If we look at which players finished among the top-12 in quarterbacks the most, then Mahomes, Luck, and Ryan are joined by Deshaun Watson and Russell Wilson as having done it ten times or more. Interestingly only two other quarterbacks than those five ranked among the top-12 in scoring in more than 50% of their games, Cam Newton and Carson Wentz. After them, only four quarterbacks who played more than eight games were top-12 fantasy quarterbacks in 50% of games in 2018. Roethlisberger, Philip Rivers and Aaron Rodgers being on that list will surprise no one. However, Josh Allen doing it more often than the likes of Tom Brady and Drew Brees would be a fairly major surprise to most fantasy players. In fact, Allen finished among the top-five fantasy quarterback options 42% of the time.
High Scoring High Variance Quarterbacks
Here we are looking at quarterbacks who finished high in the average points scoring rankings but were unreliable week to week. These options are preferable in best ball format leagues.
Ryan Fitzpatrick, TB (COR: 55.2)
I was going to leave Fitzpatrick out as he only started eight games. However, it is hard to leave out the quarterback who ranked fifth at the position on a per game basis. That is especially the case when Fitzpatrick will be a full-time starter to begin the season for the Dolphins next year. Fitzpatrick is unlikely to be as good as he was this year, especially with a worse cast of characters around him. However, there will be a temptation for people to draft him late as a flyer. The fact that he ranked 19 places lower by COR than scoring average should be all the reason you need to not draft him.
Mitchell Trubisky, CHI (COR: 48.6)
Trubisky was a fascinating quarterback last season, with his results coming out all over the place last season. He ranked among the top five at the position 36% of the time, but outside of the top-12 57% of the time. A 43% success rate inside the top-12 is interesting, but his floor was a major concern. Hopefully, a second year in his new system will decrease the variance in his performances in 2019.
Jared Goff, LAR (COR: 55.6)
When Goff was good last season he was incredible, such as the games against the Vikings and Chiefs. However, when he was bad he was disastrous, with the games against the Broncos and Bears as prime examples. In total Goff ranked among the top-12 44% of the time. The positive is that on only one occasion did he rank outside of the top-24 quarterbacks. However, finishing 10th in scoring is somewhat less convincing when it comes with a 22nd finish in COR.
Drew Brees, NO (COR: 58.9)
Brees is perhaps the most shocking name on this list. The Saints were generally considered a high flying offense in the 2018 season. Brees finished seventh in overall points scoring among the quarterback position. However, quite amazingly he actually finished outside the top-24 at the position 27% of the time. The upside was that when Brees was in the top-12 (40%), he scored among the top-five fantasy quarterback options. That Saints offense worked as much through the run game as they did through Brees at times. With it likely they operate in a similar manner in 2019, we could see these numbers again next season.
Low Scoring But Reliable Options
As the heading suggests this section looks at quarterbacks who finished significantly higher when ranked by COR than when ranked by average points scored. These quarterbacks might be surprisingly useful in leagues where you have to make lineup decisions weekly.
Eli Manning, NYG (COR: 62.5)
Alex Smith was actually the quarterback who posted the biggest difference between weekly average ranking and COR ranking. However, as his season was cut short and he is unlikely to play in 2019, we can skip straight over him. That brings us to Eli Manning who had a 14 spot difference between the two rankings. However, Manning’s value very much depended on the size of your league. He only finished in the top-12 on three occasions, but he finished among the top-24 in 13 of the 16 weeks. Therefore, in 16-team or two-quarterback leagues, Manning was actually an interesting option on a weekly basis. Whether he can put up similar numbers in 2019 without Beckham is highly questionable.
Baker Mayfield, CLE (COR: 73.2)
Mayfield was among the top-10 in the COR rankings last season. That was thanks to finishing 93% of the weeks he was active among the top-24. When he started the game he was among the top-24 every single time. For a rookie quarterback, Mayfield’s floor was incredibly high, finishing ninth and sandwiched between Wilson and Rivers. With the addition of Beckham and a full offseason with Kitchens system, the only direction in 2019 should be up.
Case Keenum, DEN (COR: 55.4)
Keenum is an interesting study. On one had he finished 10 spots higher by COR than average points total and finished among the top-24 fantasy quarterbacks 14 times (81%). In contrast, his COR ranking was just 23rd across the season and he was only a top-12 fantasy quarterback on one single occasion. The change in location is interesting, but even when Keenum was good in Minnesota in 2017, he ranked just 19th by COR. Keenum’s COR may once again be better than his average points total in 2019, but even so, it will be unlikely to lead to a fantasy relevant quarterback on a weekly basis.
Philip Rivers, LAC (COR: 68.0)
Rivers was a strange quarterback to own in 2018. He never really excited fantasy owners with a weekly ceiling that ranked 20th among players who started more than three games. However, Rivers also had a floor that was the 10th best, which led to a COR ranking of ninth. With a fun cast of characters around him once again next season, hopefully, we can see Rivers improve on his 19th ranking by average points total.
The Math Behind the Stats
For those of you interested here is how all of the numbers that make up the consistency rating (COR) are produced. I am going to go full geek mode here so I fully understand if you just want to bypass this section and take my word for it!
Standard Deviation and Coefficient of Variation
Firstly, we need to look at the average and standard deviation for that player. The standard deviation quantifies how spread out the numbers are. However, raw standard deviation numbers are messy to look at and hard to take in. They are also biased by the average of the player. A player with a lower average will have a bigger change in standard deviation for smaller relative changes in fantasy points than a player with a larger average weekly score. Therefore, we use another factor called the coefficient of variation (C.V.). C.V. takes the standard deviation and divides it by the average, allowing you to compare two players side by side regardless of their weekly average.
The only downside of using C.V. is if a player is consistently bad he will have a low C.V., and rank highly if we simply ranked players by that metric. Therefore, we need to look at a way to incorporate a players weekly points total. To do this we can calculate a players floor and ceiling, using their average and standard deviation. Over time the fantasy points scored by a player follow a normal distribution. With normal distribution we see the percentage of observations fall within one, two or three standard deviations either side of the mean.
68% of the observations will fall within one standard deviation either side of the mean. 95% within two standard deviations and 99.7% within three standard deviations. The wider we go with the standard deviations the harder it is to accurately predict a player. Therefore, the floors and ceilings are calculated by looking at the values which fall one standard deviation either side of a players weekly average.
Finally, the COR itself. COR is calculated by taking a player’s calculated ceiling and dividing it by the C.V. The higher the number the better that player ranks as a risk/reward play for fantasy owners. COR gives weight to players who have a high ceiling, but also to those who are consistent. Players who consistently put up big points totals rank highly and players who are highly variable fall further down the rankings.