I am looking to build out an app for analysing football bets.
At the moment, I have a customGPT. This GPT has a bunch of CSV files in its knowledgeable filled with very granular football stats.
Currently, I will send ChatGPT a screenshot of this bet and ask 'Analyse this bet, call on your knowledge. Return a 240 character Tweet that uses statistics to make a rating. The output looks like this:
Rate My Acca breakdown:
• Lazio v Porto: Omorodion anytime scorer? Low goal tally this season. Europa League avg goals/match: 2.67 .
• Man Utd v PAOK: Bruno to score ✅ – solid pick; Utd’s form & over 1 goal aligns with Europa’s 96% over 0.5 goals rate . Corners over 8 is bold; Europa avg: 9.73 .
• Ajax v Maccabi: Brobbey over 1.5 shots on target? High, but Ajax’s full-time win justified. Europa btts 45%, corners >5 likely (avg 9.73) .
Verdict: Stats support Man Utd leg, others feel risky. Bet ambition > reliable data. Work on tightening up the research next time! 🧐 #AccaReview #BetSmarter”
Ideally, I want the app to work like this:
1 - A user send a tweet to my bot with a screenshot of there tweet and a hashtag ‘RateMyAcca’
2 - This will be downloaded by Twitter API (This is working)
3 - I then want to pass this to ChatGPT with context. For example, if the bet slip Includes teams; Manchester United, Chelsea, Manchester City and Fulham - then the CSVs with the relevant team data will be sent to ChatGPT with the prompt to analyst the bet and create a tweet similar to above.
4 - I am having real trouble understanding the best way to pass this context.
I have the .JSONL files to with the data.
The data sets are large but I have split them by team to get the size down.
The total size of the files are as such:
Player Data (Per team): 250-350KB file sizes
Team Data: 10KB file sizes
Match Data: 80KB ish
League data (Passed with every call): 5KB.
What is the best way to go about this?
Thank you!