Automatic selection of SFTC matchups with Python Data Scraping/Analysis
The primary goal of this project is to create an automated program with an impeccable ability to pick the sports matchups posed daily on ESPN's Streak for the Cash competition (Streak.ESPN.com).
About SFTC: In this free competition, users may win cash prizes up to $25,000 by either:
- Having the longest streak of correct picks out of all players for the month
- Accumulating the maximum number of correct picks for the month
There are a variety of matchups that players may choose from, spanning all common sports:
- Picking the winning team/player (this is the most common pick)
- Picking the winning "side" of a prop bet, which may only last a brief portion of the game
- (Ex. Which QB will have more yards in the first half?)
Rules:
- Players may not register multiple accounts
- Only 1 pick may be made at a time
- Note: A player may "forfeit" their pick if they wish to make another selection before their current matchup ends
Objectives of the project:
- Identify a winning strategy: Develop two programs with objectives matching the two ways to win, listed above
- Utilize the best information available, including:
- Other SFTC analytics sites (Ex. Streak Edge)
- Betting lines/Sports analytics sites publishing predictions (Ex. 538)
- Stretches:
- Internal modeling/simulations
- Live monitoring of ongoing picks
- Find patterns that make a matchup favorable
Challenges:
- Text analytics: Translating the matchup description into the length of time that the matchup will be active (ex. a soccer game lasts about 110 minutes)
- Optimization: Maximize expected wins given estimates of win probabilities and the number of picks possible
- Classification: Separating the winners from the losers
- Stretch: real-time monitoring of on-going matchups to get a better estimate of live win probabilties