andykant / narffl

NarFFL analytics

NarFFL Analytics

This is a collection of historical NarFFL data, both a summary of its format/leagues/rules over time, as well as raw data (where available).

A Brief History

• 2009: NarFFL created by StruggleBunny
  12 teams.
• 2010: First expansion.
  48 teams.
• 2011: Premier premieres in its current structure.
  192 teams (144 Premier, 48 Alt).
  Gridiron challenge used for Horn.
• 2012: NarFFL grows a little too fast, lots of abandoned alt teams.
  1488 teams (144 Premier, 1344 Alt).
  Gridiron challenge used for Horn and promotion.
• 2013: Modern Premier/Majors/Minors structure, and shrinking for stability.
  1088 teams (144 Premier, 288 Major, 576 Minor).
  Gridiron challenge used for Horn.
  Modern promotion/relegation introduced.
  Alt teams from prior season placed in Majors/Minors based on 2012 finish (needs verification).
• 2014-2019: Modern Horn, added Farm.
  1392 teams (144 Premier, 288 Major, 576 Minor, 384 Farm)
  No changes since, aside from how the Farm lottery works.
• 2020: Expanded farm, pandemic.
  1584 teams (144 Premier, 288 Major, 576 Minor, 576 Farm)
  Temporarily added 2 IR slots for COVID-19.

Data Availability

• 2009 ❌ (NFL.com)
• 2010 ✅ (MyFantasyLeague)
• 2011 ❌ (ESPN)
• 2012 ✅ Premier, ❌ Alt (Fleaflicker)
• 2013 ✅ (Fleaflicker)
• 2014 ✅ (Fleaflicker)
• 2015 ✅ (Fleaflicker)
• 2016 ✅ (Fleaflicker)
• 2017 ✅ (Fleaflicker)
• 2018 ✅ (Fleaflicker)
• 2019 ✅ (Fleaflicker)
• 2020 ✅ (Fleaflicker)
• 2021 ✅ (Fleaflicker)
• 2022 ✅ (Fleaflicker, in progress)

ELO Ratings

We're using this data to compute ELO ratings, which can give us some insight into historical performance both of individual owners, as well as leagues and divisions.

• 2012 was the first modern-ish season, the second year of a 144-team league with relegations.
• 2012 alt leagues aren't accessible anymore, but the league had grown so much that the data may have been lower quality anyways.
• 2013 was the first season where we have full data available.
• 2013 majors/minors were constructed using 2012 alt league finishes, so we can use that to estimate starting ELO.

Given that, we'll make the following assumptions:

• ELO ratings start in 2012 when possible, otherwise 2013.
• An owner's first two seasons are considered a provisional rating (with K = 32), which will fluctuate more wildly. A single season may be a good enough sample size, but everyone has bad seasons once in a while so this helps account for that.
• Experienced owners are given an official rating (with K = 20), which fluctuates less.
• Base ratings -- primarily these are based on the size of tiers, averaging out to about 1500.
  • Premier: 1900
  • Major: 1750
  • Minor: 1500
  • Farm: 1200
• Teams start with a rating according to the base rating for their tier.
• For each game played
  • The team will get a 1.0 game credit for a win, 0.5 for a tie, and 0.0 for a loss.
  • Additionally, the team will get another 1/10 game credit for all-play against the other 10 teams -- 0.1 per all-play win, 0.05 per all-play tie.
  • By factoring both of these, we're compensating a bit for luck, while still ensuring that actual wins matter. This also helps out some in cases where a team is in an especially strong division, as they can still gain rating via all-play if they're legitimately strong.
• Each offseason there is a reset, where each team gets 2/3 of their difference from the base rating.
  • This takes effect after promotion/relegation, so a promoted Farm team would get pulled towards the Minor base rating.
  • (OLD_RATING - BASE_RATING) * 2 / 3 + BASE_RATING = NEW_RATING
  • Example: A major team finishes at 1850, but doesn't get promoted. (1850 - 1750) * 2/3 + 1750 = 1817 = NEW_RATING

ELO Ideas

• Bonuses for championships (but that may happen naturally with the offseason reset).
  • This may make more sense for Premier.