What can advanced statistics do for Fantasy Football? It turns out quite a bit. I’m playing in two NFL fantasy leagues this year. After five weeks I’m a combined 9-0-1 (9 wins, no losses, 1 tie). In first place in both leagues. I’m using two websites to achieve this record:
Website #1: http://www.borischen.co/
Boris Chen, a data scientist who works for the New York Times, applies a “clustering algorithm” to an aggregation of expert ranking data. The algorithm finds natural tiers and clusters within the data. The charts that result visualize the tiers and help you decide your starting lineup each week.
The intuition behind both this model is that a panel of a hundred experts can provide more accurate predictions than any one single expert can. Each expert uses his or her own sources and analysis to generate player rankings each week according to matchups, injuries, strength of offense and defense, etc. When you combine them, you get a consensus ranking of all players (weighted by expert accuracy).
But all ranked lists share a flaw. They imply a strict monotonic ordering and do not illustrate the true distance between players. A list implies QB1 > QB2 > QB3, whereas the reality might be QB1 >> QB2 = QB3.
The clustering algorithm addresses this, and the results are clear in the charts that are generated. The colors correspond to the tier of each player found by the algorithm.
An example chart:
Fantasy Football ranking chart
(1) Always favor higher tier players. Tiers are probabilistic; the numbers vary, but roughly speaking:
Player A is 1 tier higher than player B = player A has a 60% chance of scoring more
Player A is 2 tiers higher than player B = player A has a 70% chance of scoring more
Player A is 3 tiers higher than player B = player A has a 80% chance of scoring more
(2) When choosing between players of the same tier, flip a coin or go with your gut. I also recommend diversifying your starters across teams. Another good suggestion consists of going with the player in the game with a higher vegas line.
(3) To use the flex chart, fill in your RB, WR, and TE slots first. Then, refer to the flex chart to find the best of your remaining options.
(4) Boris updates these charts each week – they are a great way to pick up players off the waiver wire.
(5) Boris is a hero on the fantasy football subpage at Reddit – username u/prayes. Look him up.
Website #2: Defense Wins Championships!
What if you could pluck your starting defense each week off the waiver wire, and they would threaten to score league-leading totals every single time?
Another data scientist (and former poker player) is doing amazing things with NFL defenses. His name is Dylan Lerch and he posts his rankings weekly on Reddit, as well as an ‘Empeopled’ website. Dylan analyzes Las Vegas oddsmakers data on defenses to come up with his rankings. Las Vegas provides widely available, incredibly accurate projection models for NFL scoring. Dylan takes this data and translates them into fantasy scoring projections.
Most fantasy players stick with the same NFL defense for too long. Meanwhile each week there are usually better options available.
Follow Dylan on Reddit – username u/quickonthedrawl.