The Trouble With Recency Bias
It’s human nature for people to find situations that are comfortable and present the least amount of adversity. This statement can be applied to daily fantasy sports and how most of the public goes about putting together its lineup each week.
In a novice’s eyes, the higher salary players are safer selections that should produce weekly while the lower salary players aren’t as safe. This generalization isn’t totally false but the weekly salaries are more a product of prior performance, a favorable projected matchup, or DraftKings and FanDuel adjusting a salary to a bargain bin player due to an injury to someone higher on the depth chart.
Revisiting the original thought, people want comfort and security and what’s more comfortable than selecting a player with a high salary or one that finished as one of the top performers in his respective position the prior week? The NFL season is one of ebbs and flows and from the fantasy angle, ceilings and floors. A RB or WR posting 33 FP one week isn’t an indication he will do it again the next week. Even if he posted two consecutive weeks of 30+ FP, at some point, regression is bound to hit.
Think of it from a historical perspective, specifically taking a look at LaDainian Tomlinson’s 2006 season. He accumulated 2323 scrimmage yards and 31 touchdowns on 404 total touches, an average of 29.64 FPPG in PPR scoring. There have been only 65 instances of a player totaling over 2,000 scrimmage yards in a season and only 28 instances of a player scoring 20+ touchdowns. These two metrics suggest that regression should be imminent after two straight weeks of 30+ FP. Yet, there is still a certain comfort level in taking that player the following week after posting massive numbers the week before when failure seems imminent.
Recency bias can not only wreak havoc to lineups but it can also blind someone to other options that are cheaper and have better matchups. Think of recency bias from this perspective; what goes up must come down. It’s not realistic to expect a player to put up crooked numbers each week.
Tips for Avoiding Recency Bias
Consider these 2017 figures and my Recency Bias Document for each position:
- In redraft leagues, a commonly applied strategy is to draft a quarterback in the later rounds. It’s just as much a streamable position in redraft as it is in DFS. That statement can easily be validated as 22 different quarterbacks cracked a top 3 spot at one point or another last year. DeShaun Watson was the only one to make the top 3 in three or more consecutive weeks before tearing his ACL in what was shaping up to be a historic rookie season. Of 48 quarterbacks observed, 20 of them (42%) dropped more than 10 spots from their respective top 3 spots the prior week.
- Names like Le’Veon Bell, Alvin Kamara, and Todd Gurley were mainstays in the top 5 RB list for a nice chunk of 2017. Combined, these three running backs made the top 3 at the position 19 different times. For all other running backs not named Bell, Gurley, and Kamara, 24 of the 61 (39%) saw drops of more than 16 spots from the top 3 the following week.
- Wide receiver was more volatile than running back. Of 112 evaluated receivers in 2017 that reached the top 7 for the week at the position, 67 saw drops of 15 or more spots from the week prior. Of those 67 instances, 27 receivers (40%) saw drops of 50 spots or more from the previous week.
- Tight end was as streamable as quarterback; at one point last year, choosing any tight end against the New York Giants guaranteed that tight end scoring a touchdown for the first 10 weeks. Excluding Rob Gronkowski, Travis Kelce, and Zach Ertz, 20 different tight ends were part of the top 3 list at one point in 2017. Including Gronkowksi, Kelce, and Ertz, 23 of the 48 (48%) dropped more than 10 spots from the top 3 list from the week prior.
- 41 of the 48 kickers in the top 3 won their respective games and the names ranged from Greg Zuerlein who made the list seven times in 14 weeks to Phil Dawson and Robbie Gould who were priced at $4500 on FanDuel regularly in 2017. 25 of the 48 kickers (52%) who made the top 3 dropped more than 10 spots that next week. This should come as no surprise as 24 different kickers made the weekly top 3 last year at the position.
- Unsurprisingly, nearly every defense that cracked the top 3 won their respective game. What was surprising is 24 different teams cracking that list. If the aforementioned trends are any indication, then it shouldn’t come as any shock that 26 of the 48 top 3 defenses (54%) suffered drops of more than 10 spots from the week prior.
Dig Deep to Avoid Bias
Lineup differentiation is critical to winning higher entrant, larger prize tournaments. Having Ezekiel Elliott and Julio Jones in lineups is just as critical as fitting in an Austin Ekeler and Kenny Golladay type of player; the salary cap forces those cheaper selections to be made. The challenge is finding the perfect week to utilize the bargain bin players to compliment the studs that 25-30% of lineups possess. Those bargain bin players that go off with 5-10% ownership are the ones that help win tournaments especially when the higher owned players bust.
The public will select the sexier name or the one that greatly produced last week to generate lineups. The knowledgeable DFS player is cognizant of recency bias and creates a more contrarian lineup by anticipating what the public will and will not do. It doesn’t mean the well-versed DFS player won’t use a player that was successful the week before but will evaluate each scenario on a case-by-case basis. By weaponizing the flaws that the public possesses in lineup selection, recency bias can be used to generate increased success in daily fantasy.