The Fantasy Football Experts

Using Multiple Simulations to Identify Replacement Players

Using Multiple Simulations to Identify Replacement Players

Jun 28, 2012

In a head-to-head fantasy football league, an owner’s objective is to earn as many wins during the regular season in order to improve positioning within a tournament bracket.  In some leagues, teams can earn playoff byes, automatically moving them a step closer to the championship.

So when an owner is considering which player to acquire—either during the preseason draft/auction or during an in-season waiver pick-up—the determining factor should not be the number of points that the player in question will score in the future.  Instead, the owner should look for the player that will provide the most WINS for the rest of the season.

Granted, each additional point increases the likelihood of winning in a particular week.  But a team’s goal is to maximize wins (not points), so an owner needs to find players who will earn his team those valuable wins.

Earlier in June, I looked at my office league by performing a simulation of the league 20,000.  Using Excel, I randomly assigned players to the ten teams of the league.  In this league, we start one quarterback, two runningbacks, a pair of receivers, a tight end, a kicker, and a team defense.  For this analysis, each team was assigned two backup backs and receivers, and a backup quarterback, end, kicker, and defense.  WIth projections from a fantasy football website (applied to my league’s scoring settings), I determined an estimated number of points that each team would score and what each team’s record would be (assuming a double round-robin).

This simulation suggests that there are only a handful of worthy quarterbacks to consider.  For whatever reason, the statistical projections for Tom Brady, Aaron Rodgers, and Drew Brees were eerily similar, as each was projected to earn 489-491 points in my league’s structure (6 points per TD), and they each posted winning percentages of 77 percent.  Those 27 percentage points above .500 equates to 3.5 wins over the course of a 13 game schedule.  Matthew Stafford and Cam Newton were the only other quarterbacks who were expected to earn a winning record in this analysis.  In other words, every other quarterback is a replacement-level quarterback, according to this simulation.

Value-based drafting is a philosophy in which a team selects which player offers the most upside compared to replacement-level players at the same position.  In a ten-team league, most consider the eleventh quarterback to be the benchmark, but this data suggests that the bar should be set higher.  In this scenario, Eli Manning, Tony Romo, and Philip Rivers were the members of the third tier of quarterbacks, each projected to score between 355 and 365 points in my league’s format.  The data showed each of these players winning only 47 percent of the time; these starters are expected to lose more fantasy games than they win.

Outside of quarterback, no single player at another position is worth more than 2.5 wins.  At runningback, there are four players who are projected to be worth at least two wins: Arian Foster, LeSean McCoy, Ray Rice, and Maurice Jones-Drew.  The trio of Marshawn Lynch, Ryan Mathews, and Chris Johnson can add at least 1 win to their fantasy teams’ records, and the next eight backs in the projections each add less than one win (but had a winning record over 20,000 season simulations).  The replacement-level back is the sixteenth ranked player, not the twenty-first ball carrier.

Calvin Johnson and Larry Fitzgerald are the only two receivers who are worth at least two wins to their teams, while Wes Welker is the only player worth between one and two wins.  The next ten receivers are showing winning records across the multiple leagues simulated, but none of these pass catchers add a win to their team over the course of the season.  The replacement-level receiver is the fourteenth best at his position.

The projections used in this data set show that Rob Gronkowski is worth 2.01 wins, while Jimmy Graham is slightly behind at 1.83 wins.  Antonio Gates, Aaron Hernandez, Jason Witten, and Vernon Davis, the next tier of ends, each post a positive winning percentage, but their value is less than 1 win on the season.  The replacement-level end is the seventh best end.

As one might expect, this analysis shows that kickers and defenses are only marginally important in determining regular season success.  The replacement-level player at each of these positions is the sixth best, and none is worth more than one win on the season.

The players with the best chance of adding wins to a team should be the ones with the highest demand.  In an auction draft, they should cost the most.  In a traditional draft, these players should be acquired before any replacement level players are taken.

Allen North

Since discovering fantasy football in the mid-nineties, Allen North has been playing in and running leagues with friends and colleagues. With a background in business, North approaches fantasy football like an economist who is looking to maximize returns and limit risk on all decisions.