(Hannah Reyes Morales/Bloomberg)> By Neil Greenberg > Neil Greenberg Reporter analyzing advanced sports statistics for The Washington Post Email Bio Follow > August 14
Our fantasy football rankings are built from the ground up using a combination of player valuation by the world’s top fantasy experts and statistical modeling based on past performances by players at each position.
First, we convert each player’s consensus draft ranking from the experts surveyed at Fantasy Pros to an end-of-season point total using a point-per-reception scoring format.
For example, Los Angeles Rams running back Todd Gurley is ranked, at best, as the fourth-best running back and, at worst, the 22nd best back for the upcoming season, with more than two-thirds of those surveyed expecting him to finish the season somewhere between No. 7 and No. 15. As you would expect, the production from the seventh-best fantasy running back (approximately 269 points over the past three years) is significantly different from that of the 15th (202). With those rankings serving as the boundaries for potential outcomes, we then use those parameters to simulate the season 1,000 times, each time recording the number of fantasy points the player produces. That average of those simulated results provides the season point projection. We also calculated a high and low ranking for each player at their position, representing the 95th and 5th percentiles of their 1,000 simulated seasons.
Those estimates are then divided into weekly segments and adjusted for opponent. After all, a player facing a poor defense would be expected to score more points than he would against a stout defense. In addition to seeing how the player should perform week to week, we also get a sense for how often they will be a Top-10 option at the position. Given that fantasy football results stem from how well your players perform collectively on a given week, knowing how often a player is likely to rank among the weekly leaders at that position is valuable information. The more top-tier players you have rostered for a given week, the more likely you will be to win in that week.
The summation of the adjusted weekly projections (the average scores, not the best-case scenarios) give us a final valuation for the 2019 season. This provides the rankings for every player at their respective position. Next we determine the overall player rankings by using a modified version of the value based drafting baseline, which determines the value of a player not by how many points he will score but by how much he outscores his peers playing the same position, to rank the players. This accounts for positional scarcity and is why you’ll see some players with lesser season point projections ranked ahead of some with higher seasonal projections.
The baseline for each position is determined by how many man-games are needed in the league and how far in the draft it takes to get there. For instance, in a 12-team league using a standard roster there would need to be 192 man-games played by quarterbacks (12 teams each start one quarterback each week for 16 weeks) during the season. Over the past three years, it took the first 14 quarterbacks drafted to get that many games played, so our baseline for quarterbacks is the projection for the 14th best passer in the league (266 points). This functions as a replacement level. Any quarterback ranked below that point would reduce your chances at an above-average scoring week if you have to start them. The converse, is also true.
Once we’ve established the replacement level for each player it becomes easy to compare different players at different positions: the higher the value over a replacement player the higher you should be willing to draft him.
Read more fantasy football from The Post:
Source : https://www.washingtonpost.com/sports/2019/08/14/how-posts-fantasy-football-projections-work/1240