Fantasy Football: NFL Score Predictor

1
Fantasy Football NFL Score Predictor

• Matt Grecco
• Abhishek Goyal

2
Project Description

• We have designed a database that will take in
  stats for each of the NFL teams for each week.
• Based on the stats and the predict equation, the
  program will predict the outcome of any game.

3
Description of the solution

• We used three classes to implement the database
  week, team, and teamdb
• The week class contains all the stats for each
  week
• The team class contains the name,
  wins/losses/ties, and a vector of weeks
• The teamdb class contains a vector of teams

4
Description of the solution

• The database has an insert function through which
  the user can input all the stats.
• There is also a display function which displays
  the stats for any given team.
• The main part of the database is the predict
  function which predicts outcomes of games, giving
  a score for each team.

5
Predict function

• We designed an equation that predicts the outcome
  of the game based on passing yards, passing yards
  allowed, rushing yards, rushing yards allowed,
  turnovers, turnovers forced, points for and
  points against.

6
Predict function

• The equation for the predict function
• Find averages of all games, previous 8 games, and
  previous 4 games in all eight statistical
  categories
• Average those by using
• .5 avg(4) .3avg(8) .2avg(all)
• Find these for both teams

7
Predict Function

• Average the offensive parts (rush yards, pass
  yards, points, turnovers) from one team with the
  defensive parts of the other, and vice versa.
• Convert the values into scientific notation (rush
  yards / 100, etc), and multiply new rush yards,
  pass yards, points, and turnovers forced
  together.
• Finally, multiply by 3 to get score

8
Getting Data

• We got all the stats for each team from
  www.espn.com
• All the stats are in the Excel file provided with
  the presentation.
• They have also been entered in nfl.dat

9
Results

• We tested our predict function for Week 12 and
  Week 13 of the current NFL season.
• 11 out of 16 predictions were correct each week.
• We compared our predictions to that of ESPN and
  Yahoo

10
Results

• We predicted more games correctly, with a more
  accurate score, than both internet sites
• For example
• We predicted Oakland to beat the N. Y. Jets by a
  score 26 to 20.5 in Week 12.
• The final score of the game was 26 to 20.

11
Results

• Another example
• Comparing to ESPN
• We predicted New England to beat Buffalo by a
  score of 30 to 18 and ESPN predicted a score of
  24 to 21.
• The final score of the game was 27-17.

12
Group Members

• Both of us worked together on all parts of the
  program.
• It took us lot of time to get stats for each team
  for each week and also to enter it in.
• We had to spent some time to figure out the tied
  games too.

13
Improvements

• We are planning to add a search function to find
  out statistical leaders of a particular week.
• We also want to be able to display the team
  standings by division.
• We want to figure out a way to implement the home
  field advantage and loss due to injuries to
  starting players.