Hello:
"I would like to fine-tune a model so that it could read candlestick charts from an exchange and make investment recommendations based on the Weiss and Waves theory. I took the fine-tuning LLM course, but I’m unsure if, in this case, I should use an LLM model or if I should use another model specialized in image reading. I’m a bit lost. What does the group recommend?
Thank you
The information in a candlestick : low, high, open, and close prices.
You can get that data in a CSV file from many websites. You don’t need a LLM or computer vision model to analyze it.
Many LLM models have the capability to analyze images. You can capture a screen shot of a section of candlestick chart, attach to the LLM prompt and pose the appropriate prompt to seek the required information. I tried the following:
That’s not really any analysis beyond what you can see with your own eye.
Perhaps the best thing to do is to upload a dictionary in Python with the graph data. It will certainly be more efficient in terms of tokens.
Thank you
You need to define how you want to approach the problem.
If you just want to use the 4 data points of the candlestick for prediction, then just use the csv file.
If you want to run a computer vision model, then you would need screen shots of the charts.
