Photo by David Zalubowski/Associated Press
Defense-adjusted Value Over Average, or DVOA, is often referred to by fantasy analysts as a basis behind arguments. I am guilty of it; it's a statistic I have loved since I came across it over at Football Outsiders sometime last year. I've consistently utilized it to back my arguments on players such as Jonnu Smith, Ronald Jones, and Calvin Ridley without actually knowing it's correlation to fantasy football; it was irresponsible analysis, honestly. So in recent weeks, I have been searching for a research article that discussed the correlation of DVOA to fantasy football. Interestingly enough, I've come up empty with the exception of one piece using "percent finished inside the top-24," but lacking statistical significance. So I decided to see if there was basis behind my DVOA references. Could we be heavily underutilizing DVOA as an argument for our fantasy analysis? Or have we been wrongly using it up to this point?
To determine if DVOA had statistically significant correlation to fantasy performance, I first gathered all of the DVOA data over the last three years. I only used three years because the metric is often being tweaked, and any further back would render the data insignificant. After gathering the last three years of QB, RB, WR, and TE DVOA data, I then added the fantasy points per game numbers for the same year; meaning the results would indicate if DVOA is a good in-season predictor. I then ran a correlation function to see if there was a strong correlation between DVOA and FPTS/G. For those who don't know, an r-value of 0.50 or higher in traditional statistics is considered moderate to strong; an r-value of 0.70 or higher is considered extremely strong. In an environment like fantasy football, where tons of factors impact fantasy performances, it's reasonable to believe that a correlation coefficient of 0.40 would be considered statistically significant.
I went in searching for statistical significance for just DVOA; I landed on statistical significance for both DVOA and DYAR (Defense-adjusted Yards Above Replacement). Interestingly enough, there's more significance for the latter in 3 of the 4 positions.
These values indicate that DVOA and DYAR actually are very indicative of fantasy output, albeit less at the tight end position. Defense-adjusted Yards Above Replacement should definitely be used in fantasy analysis as we move into the 2021 season for RBs, WRs, and TEs. Defense-adjusted Value Over Average can be considered a secondary indicator of fantasy success for those positions, but a primary indicator of a potential QB breakout. I mentioned that this research could be a good in-season indicator of potential success, but ultimately wanted to use this research to translate into future success. I cross-referenced the players who had a positive DVOA and DYAR in 2019 with their 2020 production up to Week 13 (please note: this is not every player that had positive values in each category, but rather some of the most significant).
The great part about DVOA and DYAR is that each of them have a cutoff that is easily identified; to put it simply, a negative DVOA or DYAR is bad and a positive DVOA or DYAR is good. If you used DVOA or DYAR in your 2020 fantasy preparation, you could have identified Corey Davis and Keelan Cole as potential breakout candidates. You could have expected sustained success from Raheem Mostert. You could have seen Terry McLaurin's second-year leap coming.
Furthermore, if you used DVOA and DYAR, you could have identified probable bust candidates for 2020. If you're a Leonard Fournette truther, shield your eyes.
It's important to note that there are outliers. David Montgomery had a -13.0% DVOA and posted -46 DYAR in 2019. David Montgomery is the RB11 on the year and has flipped his negative DVOA into a positive and his DYAR to +60. Jamison Crowder posted a -11.8% DVOA in 2019; he's averaging 15 FPTS/G right now.
Players get better. Offenses change. These findings shouldn't be treated like the Bible. But in a competitive game about a game where luck plays a factor, utilizing every significant statistic you can to gain an edge is a good idea. DVOA and DYAR are indicating tools that can help you identify breakouts in 2021 and beyond.
Early 2021 Breakout & Bust Candidates
Photo by Brett Carlsen/Associated Press
Of course, we are seemingly lightyears away from the 2021 NFL season, but as we approach the end of the 2020 season, these are some players to keep your eye on.
Cam Sims - 65.60% DVOA, 95 DYAR
Mecole Hardman - 25.80% DVOA, 114 DYAR
Gabriel Davis - 22.90% DVOA, 101 DYAR
Tyler Johnson - 21.40% DVOA, 38 DYAR
Bryan Edwards - 14.10% DVOA, 27 DYAR
Tim Patrick - 11.90% DVOA, 111 DYAR
Jalen Guyton - 11.80% DVOA, 54 DYAR
Laviska Shenault - 2.10% DVOA, 54 DYAR
Diontae Johnson - -25.90% DVOA, -96 DYAR
CeeDee Lamb - -13.10% DVOA, -3 DYAR
Jamison Crowder - -6.90% DVOA, 26 DYAR
Robby Anderson - -5.10% DVOA, 61 DYAR
Jerry Jeudy - -21.60% DVOA, -55 DYAR
One thing I'd love to find out is at what percentage DVOA and what total DYAR it becomes virtually a lock that the fantasy breakout occurs. Obviously, this significance can be utilized to best predict the breakout, but increasing the likelihood of hitting is never a bad thing. I'd also love to know what contributed to outliers like Montgomery and Crowder - though, there may just be players that are good for fantasy but fairly bad at real football (see: Diontae Johnson). Are there other indicators that you can weigh more heavily (Elusive Rating for RBs, Yards per Route Run for WRs, etc.) that would help balance projections?
Overall, it's clear that DVOA and DYAR can be utilized to help arguments. The current DVOA leaders at each position:
TE (again, the TE position has less significance so grain of salt here)
Now go use DVOA to help build your playoff lineups. And follow me on Twitter @FFBaldMan. Until next time.