What is Variance?
The top three definitions from dictionary.com are:
- the state, quality, or fact of being variable, divergent, different, or anomalous.
- an instance of varying; difference; discrepancy.
- Also called mean square deviation. Statistics. the square of the standard deviation.
Divergent, anomalous, deviating from the mean/average, abnormally high or low scores – That is VARIANCE.
We want variance in our fantasy football team. Very few teams win in the playoffs by just scoring their season’s average (and counting on their opponent to bomb). We want our teams to ramp up and score a ton of points in the playoffs (variance). We also want variance in our individual players as well. How many teams last year picked up Damien Williams late in the season, to be rewarded with his dominance weeks 14-16? Williams broke away for big rushing yardage while breaking away from his season long average.
Variance is even more important in Best Ball because owners do not need to figure out when players will boom or bust. Owners automatically get to start their players when they boom and bench their players when they bust.
If you take one thing away from this article, let it be this: Variance is only used as a tiebreaker in the earlier rounds. When choosing between two players, if one is clearly better but has lower variance, still choose the better player. Near the end of the draft, variance is much more important than projected player point total. In the late rounds, you’ve already drafted your stars and with your final picks, you need players that can boom and crack the starting lineup.
Another way to frame variance is through risk-taking. If you go to the casino and only make minimum bets at the roulette table, after an hour, you might be up a little, but most likely you’ll be down because of the house edge. However, if you place a $100 bet on black, you’re either going to have $0 or $200 after the spin of the wheel. When applied to Best Ball (12 team league), if you’re up a little (scored 1850 points) or down a lot (scored 1550 points), you probably didn’t win anything. Best Ball is about going for the highest ceiling (2000+ points), so take bigger risks.
Game to Game Variance
What do Marvin Jones, Brandin Cooks, Robby Anderson and DeSean Jackson have in common?
Matt Harmon breaks down receivers by watching every route they run in a season and tracks their success rates. The success rates on deep routes by the receivers mentioned above are above average or excellent. Deep threats are extremely beneficial to Best Ball teams, since these types of players can get a 70 yard TD reception and blow the ceiling off the scoreboard. Aside from Jones who was hurt mid-season, the other three were common denominators on successful teams last year (https://confidencepoolpicks.com/key-components-of-winning-teams).
How about Ben Roethlisberger and Drew Brees? Aside from them both being future first ballot Hall of Famers, they both have historically large home-road splits. That’s variance.
High vs Low Variance
Let’s take a look at the highest variance WR (Amari Cooper) and the lowest variance WR (Julian Edelman) from the Top 50 scorers last year:
Both players helped their teams to productive seasons, but they got there in very different ways. Amari Cooper was a nightmare for owners in traditional fantasy football leagues since there was no way to predict whether he would explode or disappear. In Best Ball, he is a fine pick though since we don’t have to worry about predicting the good games. Tried and true Julian Edelman was a solid player in both traditional fantasy football and Best Ball.
Variance and the Impact on Total Points
In order to illustrate the importance of variance, let’s think about a theoretical example where you draft two QBs and both scored 300 points in the season. If they both had no variance and scored their average of 18.75 pts every week, you would have a terrible situation. The QBs would cancel each other out (aside from the bye week) and you would score roughly 300 points from these two QBs at the end of the season.
Let’s take the opposite scenario where both QBs have high variance. When one has a great week and one has a terrible week, that’s a big positive since you get the high score. There will be some weeks where they both have very low point totals and both have very high point totals and nullify each other, but you’re no worse off in those weeks than the first scenario with no variance.
Now, let’s look at the point totals and variance from real players. We calculated that value for every QB in 2018. We went all the way down to Alex Smith, who was drafted in a fair amount of leagues. QBs are the easiest to look at since Draft (draft.com/bestballbadger) only allows the start of one each week and they don’t count toward the flex position.
|Patrick Mahomes QB KC (12)||429.08||26.82||47.17|
|Matt Ryan QB ATL (9)||361.46||22.59||73.42|
|Ben Roethlisberger QB PIT (7)||357.36||22.34||55.64|
|Andrew Luck QB IND (6)||342.42||21.40||43.70|
|Deshaun Watson QB HOU (10)||340.70||21.29||71.46|
|Jared Goff QB LAR (9)||322.32||20.15||95.96|
|Aaron Rodgers QB GB (11)||314.58||19.66||81.62|
|Drew Brees QB NO (9)||309.48||20.63||111.66|
|Russell Wilson QB SEA (11)||305.92||19.12||35.70|
|Philip Rivers QB LAC (12)||297.02||18.56||35.63|
|Cam Newton QB CAR (7)||295.60||21.11||46.16|
|Dak Prescott QB DAL (8)||293.90||18.37||57.35|
|Kirk Cousins QB MIN (12)||292.62||18.29||62.15|
|Tom Brady QB NE (10)||291.80||18.24||42.40|
|Mitchell Trubisky QB CHI (6)||275.02||19.64||132.94|
|Baker Mayfield QB CLE (7)||254.10||18.15||24.71|
|Eli Manning QB NYG (11)||250.96||15.69||24.70|
|Case Keenum QB WAS (10)||229.90||14.37||21.99|
|Derek Carr QB OAK (6)||226.26||14.14||72.54|
|Matthew Stafford QB DET (5)||223.18||13.95||35.52|
|Josh Allen QB BUF (6)||220.06||18.34||134.97|
|Jameis Winston QB TB (7)||209.78||19.07||73.40|
|Carson Wentz QB PHI (10)||200.16||18.20||37.93|
|Andy Dalton QB CIN (9)||185.54||16.87||38.87|
|Blake Bortles QB LAR (9)||184.22||14.17||100.02|
|Marcus Mariota QB TEN (11)||183.42||13.10||89.29|
|Sam Darnold QB NYJ (4)||183.40||14.11||40.55|
|Ryan Fitzpatrick QB MIA (5)||177.84||22.23||147.47|
|Lamar Jackson QB BAL (8)||162.54||10.16||79.53|
|Ryan Tannehill QB TEN (11)||151.46||13.77||58.73|
|Joe Flacco QB DEN (10)||143.10||15.90||25.96|
|Alex Smith QB WAS (10)||143.00||14.30||13.38|
What does this chart say? Points are king (and draft value). If you drafted Pat Mahomes in the 11th and he destroys everyone with points, it doesn’t matter that he had low variance. Part of the calculation is based on average (higher average, lower variance).
Let’s say we have two quarterbacks. One that scored 309 points over the season and on that score 305 points over the season. Knowing nothing else about them, we’d expect that the teams that drafted the first QB scored somewhere between 0 and 4 points more than teams that drafted the second QB. Let’s say 2 points more to the final season score (the other 2 points were tacked on to a meaningless low scoring game).
Now, let’s add on more information. The first QB was Drew Brees who had a lot of variance (high and low scoring games) and the second QB was Russell Wilson who was fairly consistent throughout the season. What would you think the difference would be? 5 points? 10 points? 20 points?!
For this analysis, we paired up Brees and Wilson with the top 20 other scoring QBs (from Mahomes to Josh Allen) and calculated the score. Brees + QB on average scored 407.84 pts and Wilson + QB on average scored 382.36. That’s a whopping 25 point difference even though Brees only outscored Wilson by 4 points on the season.
Just to give a comparison, if we owned Brees and Wilson and then drafted a 3rd QB like Matt Stafford, it bumps our total from 396.96 to 426.18 or about a 29 point bump. Stafford was a reasonable QB drafted in the 11th round last year. Most teams who drafted a 3rd QB, drafted them at the very end of the draft (16th-18th). Guys like Flacco, Tannehill, Darnold, Eli Manning, Dalton, Bradford, and Tyrod Taylor. These QBs would have resulted in 18.24, 9.66, 5.14, 24.90, 25.78, 0, 0 point bumps respectively. It’s a bit of a crapshoot with the 3rd QB and it requires a draft pick. That means the value of a high variance mid-tier QB is roughly equivalent to a low variance mid-tier QB + late round QB that doesn’t bust.
Player to Player Variance
Let’s take a look at the two QB situation again. The one where they both scored 300 points. That combination will score roughly 380 points by the end of the season with 220 points going to waste as they cancel each other out.
Ok, now imagine kicking those two similar and boring players out the door and replacing them with a QB that scored 600 points and a QB that scored 0 points. That would be AMAZING since your team just scored 600 points and wasted 0 points from the QB position. That’s quite an extreme example, but it illustrates the benefit of owning top tier players at every position (i.e. difference makers).
Last year, would it have been better to own the QB pair of “Pat Mahomes + Case Keenum” or the QB pair of “Andrew Luck + Deshaun Watson”?
Take a few moments to think about that situation…
Even though Luck + Watson scored a combined 683 points compared to Mahomes + Keenum combined 659 points, there was much more overlap. The vaunted Luck/Watson combo would have contributed 429 points to your team, while Mahomes and the pedestrian Keenum would have yielded 449 points. The big point/little point combo worked out better than the middle/middle point combo.
Last year, the top 4 scoring QBs were all drafted around the 10th and 11th round (Mahomes, Matty Ice, Big Ben, Luck). Could that happen this year with Wilson, Newton, Winston, Goff, Prescott, and Roethlisberger going in that range? Yes, but odds are still on Mahomes, Watson, Luck, and Rodgers to outpace the field.
Grabbing two QBs in the 10th to 11th rounds is a valid strategy, but Best Ball Badger recommends trying to snag a QB earlier with a better shot at a Top 3 season (Watson, Luck, Rodgers, Ryan, Wentz – in the 6th-9th round). Mahomes draft capital is steep but might be worth a stab at with Tyreek Hill avoiding suspension.
Other Forms of Variance
We’ve already mentioned speedsters or deep threats having an edge at the WR position by creating variance with long pass plays. Same thing goes at TE with field stretchers getting opportunities for big plays. OJ Howard, Evan Engram, and Vance McDonald have shown they can score long TDs. It might be too early for Noah Fant, but watch him in future years.
Fast RBs can act as home run hitters. Matt Breida and Kalen Ballage are being drafted very late and both have shown to have breakaway speed. Matt Breida was the fastest ball-carrier in 2018 according to Next Gen Stats. Kalen Ballage ran a 4.35 40-yard dash in his pro day and had a nice 75-yard run near the end of last season. Both will be in dedicated RB committees, but can score points for your team even with limited opportunities.
Rookies have a very wide range of outcomes at this point in the offseason (before pre-season) as few know what kind of workload they will receive. Every year we have rookies (rookie RBs in particular) that come out of the woodwork and into a prominent role. Last year, Nick Chubb waded through a sea of Carlos Hyde and Duke Johnson to earn the starting role. Kerryon Johnson was starting to look like a fantasy beast, averaging 15.1 ppg in 5 weeks after the bye, until he went down with an injury. Even Saquon being drafted at 1.6 could be considered a steal. In 2017, Kamara, Hunt, Fournette, and McCaffrey all significantly outperformed their draft position.
Josh Jacobs could be a McCaffrey lite with the backfield all but locked up (Lynch retired and Crowell with an Achilles injury) and having receiving skills. This year there are many potential sleepers at the position. Alexander Mattison (MIN) looks to fill the Latavius Murray role but with more upside. Devin Singletary (BUF) only needs to carve out some time from aging and oft-injured LeSean McCoy, 36-year old Frank Gore, and pass catching specialist TJ Yeldon. Miles Sanders was picked in the second round and finds himself on a great offense. By midseason, we expect him to takeover the primary RB role from Jordan Howard who has been working on his pass-catching since 2016.
We try to avoid Rookie QBs whenever possible unless they have solidified the starting job and they can run. Josh Allen and Lamar Jackson were draft day steals if you had the guts to draft them. This year we expect the pool of relevant rookie QBs to be Kyler Murray and no one else. Even if Dwayne Haskins and Daniel Jones play this year, they’re not projected to be fantasy relevant.
Rookie TEs and WRs are in the avoid category as well. We haven’t seen a bumper crop of Rookie WRs since 2014 which might live on as the best WR draft class of all-time. Sammy Watkins, Mike Evans, Odell Beckham, Brandin Cooks, Kelvin Benjamin, Jarvis Landry were fantasy relevant (Davante Adams and Allen Robinson were drafted in 2014 as well).
Last year, Calvin Ridley was the only WR to really help their team. Dante Pettis, DJ Moore, and Christian Kirk were mediocre picks as rookies, but we expect them to take a big step up in their 2nd year. If you want to take a shot this year, DK Metcalf is worth looking at. He has game breaking speed and is paired with one of the best deep ball throwers in Russell Wilson. Mecole Hardman has been left for dead with the no suspension announcement for Tyreek Hill. Hardman is also very fast and with all the extra attention paid to the other receivers, should connect with Mahomes on some deep shots. Finally, Andy Isabella may have a handful of long bomb opportunities with Murray scrambling around and buying time.
Backup RBs or handcuffs are all lottery tickets. Value at the position is equal parts opportunity and talent. Quite often there will be RBs that change from fantasy non-existence to RB1 with an injury or a suspension. Guys like Damien Williams and James Conner were elevated to superstar status when given the starting role. Phillip Lindsay seemed to come out of nowhere to take the job from higher drafted Royce Freeman.
Best Ball recommends drafting 6 RBs at a minimum to take advantage of this huge potential value. Every team has a backup RB, but a few potential targets would be Rashaad Penny and Royce Freeman. They’re well-rounded backs that could takeover if an injury happened to the larger share-owner in front of them in the RBBC. We recently removed D’Onta Foreman from this list as he was surprisingly cut from the Texans roster on 8/4.
For late round values, Alexander Mattison will be taking on the vacated Latavius Murray role in Minnesota and Matt Breida looks to be in a 2 way timeshare with Tevin Coleman due to the Jerrick McKinnon injury. Both of these RBs have some standalone value and have tremendous upside if there is an injury to the primary back.
Drafting multiple players from one team also increases variance as you are relying heavily on one team’s success. The key is choosing the team that significantly outperforms expectations.
Examples of fairly safe stacks for this year (should stay in the upper echelon of offenses):
- Colts – Luck (6th), Mack (3rd), Hilton (3rd), Ebron (7th), Doyle (13th) – this combo would typically be drafted at the 2-3 turn
- Falcons – Ryan (8th), Freeman (3rd), Julio (2nd), Ridley (5th)
Stacks from average offenses that are primed for a bounce back:
- Eagles – Wentz (9th), Sanders (7th), Jeffrey (7th), Ertz (3rd)
- Packers – Rodgers (6th), Jones (3rd), Adams (1st), Valdez-Scantling or Allison (8th)
Stacks from below average offenses that should make big improvements:
- 49ers – Garopollo (13th), take your pick at RB (6th-14th), Pettis (7th), Kittle (2nd)
- Cardinals – Murray (8th), David Johnson (1st), Kirk (6th), Fitzgerald (9th)
As with most things in life, moderation is key. We want to take on a decent amount of variance and risk, but not so much that we put our chances of winning similar to winning the lottery. It does us no good to break the record in points in one league and sputter out in all the rest.
Written on 7/15/19 by Winston Lee (Updated 8/4/19)