Beta Testing New Ideas: The Draft 2023

Here at the Science of Fantasy Football, we use the scientific method to analyze statistics for Fantasy Football. The seven scientific method steps are question, research, hypothesis, experiment, data analysis, conclusion, and communication. Over the past ten years, I have developed statistical tools for Fantasy Football using this method, and the results have been outstanding. Over the past two seasons, I have developed a drafting philosophy called “Better Than Zero,” which utilizes ADP, my Weekly Values, and Consistency Ratings to draft my teams. Through research over the years, I have learned that three things are most responsible for success in Fantasy Football. Consistency, prominent game ability, and avoiding bad games. Before the formula was developed for the MVP Index this season, I categorized those three performance levels and used them to draft new teams. Let’s look at the experiment and then at the results after 12 weeks of the Fantasy Football season. 

Hypothesis

Chasing what matters in Fantasy Football will lead to a better draft. Consistency, significant game ability, and lousy game potential are the three things that matter the most. This concept was a big part of my new drafting theory called “Better Than Zero,” which debuted last year. 

Experiment 

After I visited the Fantasy Football Expo in Canton, Ohio, this August, I was invited to join many new leagues. This opportunity would be the perfect experiment to test my new drafting theory and statistical tools. 13 different leagues were part of this experiment. They covered many formats: auction, redraft, best ball, and dynasty. 

Experimental Flaws

In any scientific experiment, you need to have the variables identical when you test one situation against the next. This process is only possible in fantasy football if all 13 leagues contain the same players and rules. By utilizing players more active in fantasy football, I aim to get more active, engaged players with above-average fantasy football knowledge involved in this test case. Using leagues of different sizes and rules while going against the scientific method of limiting variables should give a better understanding of the success or failure of the drafting methods and data used in various leagues a typical fantasy football player would encounter. 

Grading Results

Fantasy Football success should be measured every season in return on investment (ROI). For this case study, I will ignore the ROI and focus on metrics such as winning percentage and playoff opportunity. Many leagues have a weekly matchup against the median score and the head-to-head matchup. 

Data Analysis-see data table at end of article

Fantasy Football provides a complicated data set, so it’s difficult to assign a minimum percentage of success that is satisfactory for any new tool used in the competition. In this 13-league test case, the median win percentage is 75%. Sixty-nine percent of my teams are locked into the playoffs with two weeks to go in the regular season for fantasy football. Eighty-five percent of my teams are in a playoff spot, with only one out of the thirteen failing to reach the playoffs. Let’s break the data down into smaller bites by league type. 

Auction- Only one league is involved. 75%-win percentage, locked in playoffs.

Best Ball- Three leagues involved. Median 62.5%-win percentage, 2 of 3 locked in playoffs with one out. 

Dynasty- Six leagues involved. Median 66.7%-win percentage, 4 of 6 locked in playoffs with all in. 

Redraft- Three leagues are involved. Median 75% win percentage. 2 of 3 are locked in playoffs with all in. 

League Size- There is no significant difference in performance by league size. 

Number of Starters- There is no significant difference in performance by number of starters. 

Superflex Rule- There appears to be no significant difference in performance by Superflex rules. Eight leagues included a Superflex starter, one had a double Superflex, and four were single QB leagues. 

Conclusions

The experiment is incomplete for several reasons. First, this drafting method would need to be compared to a control to have a better analysis of whether it was better or worse than just using ADP, for example, or some other draft theory. Second, there needs to be more data points to make definitive statements on whether this method is better than any other in fantasy football. Third, the success in dynasty formats should be based on more than one year since a player’s drafting philosophy can alter the first-year results. 

(Win Now theory versus Build for the Long Haul) Over my 30-plus years of playing fantasy football, a success rate of getting 80% of my teams into the fantasy football playoffs has been a level that has defined a successful year since this leads to a satisfactory ROI. With two weeks to play in the regular season, my test group of 13 leagues has hit the playoffs at an 85% rate. 

The worst team in this case study was one of the Best Ball teams. This result makes logical sense because, in Best Ball, we care more about a player’s upside potential and not how many lousy games he might have in any given season. The awful game percentage is more important in set lineup leagues, regardless of whether they are a single-season or dynasty format. 

While the small sample size of data points makes it impossible to make any definitive conclusions, I am delighted with the results. I will have another report on these thirteen teams upon the conclusion of the fantasy football playoffs. 

DATA