topbanner_forum
  *

avatar image

Welcome, Guest. Please login or register.
Did you miss your activation email?

Login with username, password and session length
  • Thursday December 12, 2024, 6:27 am
  • Proudly celebrating 15+ years online.
  • Donate now to become a lifetime supporting member of the site and get a non-expiring license key for all of our programs.
  • donate

Author Topic: MINI-NANY 2010 Release: World Cup Predictor FIFA 2010  (Read 8872 times)

mouser

  • First Author
  • Administrator
  • Joined in 2005
  • *****
  • Posts: 40,914
    • View Profile
    • Mouser's Software Zone on DonationCoder.com
    • Read more about this member.
    • Donate to Member
MINI-NANY 2010 Release: World Cup Predictor FIFA 2010
« on: December 26, 2009, 10:19 PM »
MINI-NANY 2010 Entry Information

Application Name World Cup Predictor FIFA 2010
Version v2010
Short Description This is an online javascript tool that uses Markov Chain Monte Carlo statistics to predict the best monetary bets to place on the FIFA 2010 Tournament
Supported OSes Online (firefox,IE)
Web Page https://www.donationcoder.com/wcp
Author mouser and crono
Screencast

This is an online tool to help predict the most profitable bets on the winner of the FIFA World Cup 2010 (soccer).  It uses Markov Chain Monte Carlo statistical methods to simulate tens of thousands of tournaments, using expert-provided pairwise game score estimates, and then calculates the expected payoffs of bets based on current odds.

Description

MCMC works by leveraging simple single-event probabilities, that you can easily estimate, into long-range predictions involving many interactions. There's no magic involved. It's basically a way of estimating the probabilities of complex multi-step outcomes through brute force simulation.

In the case of the World Cup, this amounts to simulating complete tournaments by playing each game from round one through the final, choosing winners based on simple pairwise expectations of how the teams pair up against each other.

The pairwise match-up probabilities are the key to being able to simulate entire tournaments autonomously, and must be provided by an expert or from records of previously played games. You can create your own pairwise estimations using WCP'10 or use one of the probability estimation tables generated by an expert.

Obviously, the accuracy of this pairwise match-up table is critical to getting good results from the Tournament Simulator. The key to the power of MCMC methods is that they let you convert simple known (or relatively easily estimable) probabilities, and by simulating full sequences of outcomes over and over again, arrive at an estimate of the probabilities of different long-term highly-interactive final outcomes.

You can read more about the statistical methods involved on the wcp page: https://www.donationcoder.com/wcp

History

This tool was originally written for the 2006 world cup, and DC member Crono was instrumental in helping me code it.

It is borderline inappropriate for NANY since it is not a complete rewrite -- but I am hoping it will be ok with everyone to submit it because it is a whole new tournament event.

Screencast


Screenshots
From the 2006 tournament:
sampleresults.png
« Last Edit: January 01, 2010, 07:33 PM by mouser »

mouser

  • First Author
  • Administrator
  • Joined in 2005
  • *****
  • Posts: 40,914
    • View Profile
    • Mouser's Software Zone on DonationCoder.com
    • Read more about this member.
    • Donate to Member
Re: NANY 2010 Pledge - World Cup Predictor FIFA 2010
« Reply #1 on: December 27, 2009, 09:27 PM »
Thought i would explain a little about my my rationale for making it a NANY submission even though the underlying code is not a complete rewrite.

Basically this version is built off of the 2006 version which accompanied the 2006 World Cup; after 2006 the predictor system went offline and was useless.

Today, 4 years later, i've set up all of the new tournament data for the 2010 event and re-set it to start from scratch all over from the beginning again for the 2010 event.

So that's my reasoning for thinking it could qualify as a NANY, but i'm happy to submit it as a Mini-NANY instead as an alternative.