fantasy_football_scraping

This repo has a collection of scripts made to assemble data for fantasy football analysis as well as selenium scripts to pull data from my leagues to keep track of performance of fantasy football managers.

findings so far

database

• Already well established
  • Draft a RB or WR early
    • The distribution of quarterbacks is flatter than that of running backs, wide receivers, and tight ends
    • This reinforces that the value above replacement of a non-quarterback is higher for early picks than for quarterbacks
• Less trivial
  • Top WR's are less likely to be mediocre than are top RB's, although they have greater bust potential
    • Top WR's have a higher likelihood of repeating near the top of their position than top RB's
    • Top WR's also have a higher likelihood of finishing outside of the top 30 in their position than do top RB's

The model

• The model pulls from 6 sources *

database_assembly.ipynb

Pulls from profootball focus data of each player for specified years using beautiful soup.

Can use this to run regressions and machine learning analysis * Points per game next year * Games next year * Change in points per game next year * Change in games next year

To do:

Does not require new scraping: Need a targets available, targets taken measure for WR's and TE's Base off attempts by QB of team, tally up targets of each WR on the team last year Need a RuAtt available, look at number of RuAtt on the team last year vs next year Make RF into a function Make plot of residuals Put some cleaning back into the functions Need a better validation function to measure performance Can take out the least relevant variables

Requires new scraping:

• Add actual injury info, dummy variables for different injuries in the previous year(s)
• Need some sort of analysis for rookies
  • Does the team they were drafted by have a star at the same position?
  • Where were they picked in the draft?
  • Winning pct of team picked?
  • Typical college metrics including injury history
• Need some sort of comparison with ADP eventually
  • Look at ADP this year, ADP following year type of thing. Delta ADP for risers, sinkers
  • Look at second half performance and how it affects ADP??
• New coach flag

Do-able with current data:

• Can get a flag for backup RB
• "On a good team flag," maybe it's the winning percentage of the team, or a flag on > .500 winning pct

login_pull.ipynb

Pulls from ESPN, MFL, and CBS game and season data for each fantasy manager to later be used in visualizations and analysis