Gabriel Burrill

Designing a data dashboard to predict Rugby League match outcomes

Gabriel Burrill -

Although there are some existing applications which do this, there are specific meaningful data points that I like to use. I am a software engineer and very passionate about sports so I decided to make my own!

Initially, I am envisioning a fixture list for upcoming round of NRL games. Each individual match will have a dropdown to view a short summary and can be clicked on to see a detailed summary.

I will be using API's where possible to get the data and otherwise using web scrapers. I want to get the following information displayed in the detailed summary:

  • Try locations (left, right, middle) scored and conceded for each team in the current season
    • Amount of tries scored/conceded
    • Competition rank (for example, a particular team has the best right side attack in the competition and the other team has one of the worst left side defences can suggest a player on that side will be more likely to score)
    • Percentage of tries scored/conceded on the left/right/middle
  • Recent results which can be filtered by number of rounds on front end
      • Average points conceded and scored in previous matches can be displayed from this data
      • I also want short summaries for previous matches, so that a result that is an outlier can be easily understood at a glance (for example, players sent off in a game or significant players missing)
      • Filter by matches with or without a particular player in the side
  • Which players and what position they were playing in that game when they scored the try in previous games (for example, to see if a particular team often concedes to a right winger suggests a weak defence on that side)
  • Team news (players in/out) and if they are named in a different position to normal
  • Which position each player is expected to play in based on team lists released every Tuesday
  • Automation to automatically update matches to change for each round and show updated data once matches are played

Once I deploy this project into production I'll look at some further interactive features such as

  • Ability to sign up as a user and comment on matches, compare prospective bets with other punters or take other's insights into the match
  • Assigning my own value as opposed to the bookie's value
  • Using AI to make predictions
  • Betting with pretend money to test out before using real money

Data model - V1

The following is the data model required for the first few features (displaying upcoming/current round, previous match results and who scored from which position). This will be extended for future features. Below displays entities and

Match

  • Id (PK)
  • HomeTeamId (FK)
  • AwayTeam (FK)
  • Date (string)
  • Time (string)
  • HomeTeamScore
  • AwayTeamScore

Team

  • Id (PK)
  • Name

Player

  • Id (PK)
  • Name (string)
  • PreferredPosition (string)
  • TeamId (FK)

MatchPlayer

  • Id (PK)
  • MatchId (FK)
  • PlayerId (FK)
  • PositionPredicted (string)
  • PositionPlayed (string)

Tech Stack

I have opted to use a react frontend, with python on the backend using the following tools

  • SQLAlchemy for relational database
  • Selenium/Beautifulsoup for web scraping
  • FastAPI to manage the data to/from frontend
  • Celery to schedule the web scraper