I built a prototype that collects information about upcoming and ongoing IPOs, displays the results, and scores each IPO according to rules I created. At the moment, I have to provide instructions manually each day, which uses a lot of tokens and makes the process inefficient.
I would like to turn this into a program or local web application that runs on my PC. Ideally, it would collect IPO data from a reliable source, show only the fields I care about, apply my predefined scoring system, and update the results automatically on a schedule. The scoring rules would come from me; I am not asking an AI system to make independent investment decisions.
My programming experience is limited. I know basic Python, including functions and object-oriented programming, but programming is not my main field. I expect to use AI to help write code, so I am looking for guidance on the concepts, tools, and overall system design I should learn first. What would be a sensible path for breaking this project into manageable parts?
4 Answers
A practical first version could be a Python script that reads a spreadsheet or JSON file, calculates the scores, and writes the results to a new file. Once that works reliably, replace the input file with an API or web scraper, add SQLite for storage, and schedule the script to run daily. Building it in that order will make debugging much easier than trying to create the data collector, database, web app, and automation all at once.
Before choosing tools, describe the process in plain language: where the IPO data comes from, which fields you need, how each score is calculated, and what the final result should look like. Then divide it into independent pieces, such as data collection, data cleaning, scoring, storage, and display. For each piece, write down its inputs and outputs. Whenever you reach something you cannot explain or implement, that identifies a specific topic to learn instead of leaving you with the vague goal of learning all of programming.
This is best understood as a small data pipeline rather than one particular framework. You would typically fetch IPO information from a reliable website or API, convert it into a consistent format, apply your scoring rules, save the results, and then display them in a simple interface. Start with a command-line Python version before building a web UI. Focus first on HTTP requests, JSON, basic data validation, SQLite, and scheduling jobs. Keep the scoring logic as a separate function so it can be tested with saved sample data. AI can help generate boilerplate, but make sure you understand the code and errors it produces, and add tests so changes do not silently alter your scoring rules.
Be careful about the data source and the financial consequences. The program may be simple, but inaccurate, delayed, or inconsistent IPO data can produce misleading scores. Record the source and update time for every result, handle missing values explicitly, and treat the output as a screening aid rather than proof that an IPO is a good investment.
That makes sense. I currently find IPOs manually, so I can begin by testing the scoring function with information I have already collected before automating the data collection.