How to Most Effectively Draft a Fantasy Football Team
Authors: Trent Bellinger (Project Lead), Brian Dinh, Jonah Jung, Memphis Lau, Nathanael Nam
Fantasy football has become an extremely popular way for football fans to showcase their football knowledge in a competitive format. Almost every avid fan of the National Football League (NFL) has at least one fantasy football team full of players that participate in the NFL. In fantasy football, NFL players receive points for their performance on the field, and the goal is to have your fantasy team get the highest number of points as possible. This format poses one major question: what is the most effective way to choose which players to pick for your team? In order to answer this question, we will look at which statistics are most useful in determining success for a player in fantasy football.
The format we will base our data on is a fantasy football league where each team consists of one quarterback, two running backs, two wide receivers, one tight end, and one FLEX (running back, wide receiver, or tight end). We will use standard Points Per Reception (PPR) scoring, where players earn points based on their success in games as outlined in the table below.
Quarterbacks:
In fantasy football, there are two ways in which quarterbacks can get points: passing and rushing. The main goal of this section is to figure out which of these categories contributes most to a quarterback’s success in fantasy football.
We will start with passing statistics. The two ways that quarterbacks gain points from passing are passing yards (0.04 points per yard) and passing touchdowns (4 points per touchdown). The graphs below depict the correlation between quarterback passing yards and touchdowns and total fantasy points. The data used is collected from the top 25 quarterbacks in terms of total fantasy points in the 2021 season.
The above graph depicts a relatively strong correlation (r-squared 0.68) between total passing yards and total fantasy points. It is shown that 6 out of the top 7 quarterbacks have more passing yards than expected by the model, which is an indication that the most successful quarterbacks rely more on passing yards than the most other quarterbacks. This indicates that passing yards are a good way to predict the most successful quarterbacks in fantasy football.
This graph has an even stronger correlation (r-squared 0.724), showing that passing touchdowns and overall a better predictor for total fantasy points for quarterbacks. The graph shows that 5 of the top 8 quarterbacks got more passing touchdowns than predicted by the model, so it is less efficient to predict the top quarterbacks using this model as compared to total passing yards.
We will now take a look at rushing statistics among quarterbacks to see if this is a more efficient way to predict quarterback success.
According to the above graph, there is an extremely weak correlation between rushing yards and success in fantasy (r-squared 0.08891). 7 out of the top 8 quarterbacks underperform according to this model, so using rushing yards as a way to predict success in fantasy football for quarterbacks is extremely inefficient.
Similarly to rushing yards, rushing touchdowns and fantasy points show an extremely low correlation (r-squared 0.0646). There is clearly no distinct pattern between rushing touchdowns and fantasy points, as many outliers can be seen on the graph. This high variance shows that rushing touchdowns can not be successfully used as a way to predict fantasy football success.
Overall, passing statistics are much more relevant and provide a much better projection of fantasy football success for quarterbacks. Using this information, clearly getting a quarterback that is in a high-powered passing offense will be much more effective than getting a quarterback that tends to rush more.
Running Backs:
Predicting the next breakout running backs is very difficult. Because running backs get tackled so often and rely so much on their speed and athleticism, injuries and age affect this position exponentially more than others. Only 3 of the top 10 running backs in 2021 were also top 10 in 2020. In fact, out of the 22 players that appeared in both top 40 lists for 2021 and 2020, the average difference in rank is 11.68, meaning that predicting a player’s performance for one year does not rely on their performance in previous years, which could be argued for other positions in football. The 18 players that were not on both lists played on average only 10 games (out of a 17 game season) in 2021, compared to the 14.4 played games on average for the 22 that remained. Injuries heavily affect these players’ performances and add another element of unpredictability to the running back position.
This graphic shows there is not a strong correlation between age and Fantasy Rank, but it is worth noting that there has only been one top-20 RB in the last two years that was 28 years or older at the time.
When analyzing how stats affect Fantasy Rank for RBs, it appears that the receiving game is not a great indicator. The proportions of fantasy points from rushing (instead of receiving) is relatively similar across all RBs and cannot be used to determine which RB will succeed the most in fantasy.
These visualizations show that the amount of carries correlate strongly to a player’s fantasy performance, while the yards per attempt does not. For fantasy football, this means that volume is more important than quality.
Wide Receivers:
When predicting wide receiver success in a season of fantasy football, it is common to search for attributes such as targets, receptions, yards, and touchdowns. Typically, these are attributes most common to take into consideration while choosing a wide receiver mainly because they have often produced results in the past. However, many individuals seem to overlook the consistency of a player week by week. Oftentimes players might have exceptional performances one week but have a rough week the next. The data may not reflect this oftentimes because the projections and seasonal statistics are not game by game. For a fantasy season with only 16 weeks, each week is very important and choosing a consistent wide receiver is vital for team success.
This graph is a representation of wide receiver consistency and the risk when drafting players. Each point on the graph represents an individual player in the top 30 of the 2021 season. The x-axis represents the average points per game and the y-axis represents the variance in this point production per game. When drafting a player, you want a player with high average point production per game and low variance (low risk, high reward). These are represented by the green sections on the graph. Conversely, you want to avoid the players with a low point production and a high risk (high risk, low reward). These are represented by the red portions on the graph. The players that are in neither of these categories are those that have low average point production and low variance or high point production and high variance. The players in these sections of the graph do not pose any extremely high reward or risk.
Tight Ends:
In order to predict which factors affect the number of fantasy points that a tight end accumulates, we must attempt to find the correlation between all the data values in each of the columns. Tight ends generally tend to accumulate the lowest number of points because of their role in football. Tight ends block for both the quarterback and the running backs, and run into the field to catch a pass. On most teams, the quarterback typically does not throw a high number of targets to the tight end. In fact, there is an average of 7.73 targets thrown for all of the tight ends in the NFL. Since players do not get fantasy points for blocks, which is the tight end’s main job, acquiring a high number of receptions is the best way to accumulate points for tight ends. In the data set, the number of receptions and yardage for each player has a high correlation with the number of fantasy points. Based on all of the statistics of the data set, it appears that tight ends should be picked in the later stages of the fantasy football draft due to their limited ways to get points throughout the season.
The visualizations above have a moderate, positive correlation that demonstrates that the more reception and target passes a tight end receives, the more fantasy points they will accumulate. The two graphs look almost identical, which shows that the players who get a lot of target passes also tend to get a lot of receiving yards. Travis Kelce, Mark Andrews, and Taysom Hill were filtered out of the data visualizations because they are outliers in the dataset. These players are outliers for the tight ends because their statistics and fantasy points are much higher than the other players. For this reason, these players should be the players that are drafted early on in the Fantasy Football Draft.
FLEX:
Choosing your players for each position is already tough as it is. When choosing which position to fill into your flex slot, the variety of choices you could have turns out to be a far more complex decision week by week as different matchups are played, players being rotated in and out, etc. Despite not knowing which position to start that week, having this choice adds immense flexibility to your fantasy team. Generally, when considering which position to start, one should think about their fantasy league and scoring guidelines they are in, the players they have starting, and what players they have left that they can utilize. Choosing a position that works for your team and matchups will grant you the most success and will maximize your flex slot to its fullest potential.
This graph shows the median fantasy points obtained by each position (running back, tight end, and wide receiver) during the 2021 season for four different fantasy football leagues. Due to the tight having less opportunities to gain fantasy points compared to the running back and wide receiver, it is usually not an optimal choice in many cases. However, between the different fantasy leagues the choice boils down to either the running back or wide receiver. As seen in the graph, if you are playing in a PPR (DK, FD) league, you typically should start a wide receiver over a running back most of the time. In a non PPR league, the choice could swing either way and is now dependent on the matchups and players that you have.
Conclusion:
Although fantasy football can seem extremely random at times, there are some underlying trends that have allowed us to create insights into which players are better to choose for your team. For quarterbacks, choosing a passing quarterback is the safest and best bet in terms of consistent points as opposed to choosing a rushing quarterback. For running backs, there is extremely high variance, but it is generally best to choose a young running back (under 28 years old) who does not rely on catching passes. For wide receivers, there are generally only 3–4 players who will consistently produce positive results, so having one of these players on your team is extremely important. There are some wide receivers that perform highly some weeks and poorly other weeks, so having a couple of those will be beneficial if you can choose the right weeks to put them in your team. Finally, for flex, if you are in a PPR league, it is almost always best to put a wide receiver in the flex position, but if you are in a non-PPR league, putting a running back in the flex position will produce similar results.
References:
https://fantasydata.com/nfl/fantasy-football-leaders?season=2021&seasontype=1&scope=1& ok subscope=1&startweek=1&endweek=1&aggregatescope=1&range=1
https://www.fantasypros.com/nfl/stats/qb.php
GitHub: https://github.com/trentbellinger/DataBlog-Fantasy-Football
