I know… It’s not cool no more. But I felt I had to revisit “From Data to Trade” with ChatGPT-4, and see what could be trivially retrieved from its weights! The progress between the two versions is amazing! To a point, I had a near philosophical discussion with a former colleague on transformers, information theory, lossy vs. lossless compression, and the possibility that, at some point, all human knowledge could be stored in only a few hundred billions parameters. “Decoding the Quant Market” is definitely better than ChatGPT-3.5’s “From Data to Trade” essay produced last year, and while it won’t make you rich, this book contains lots of interesting entry points to relevant systematic and quant trading concepts that you can explore further with cornerstone books, papers, or even better, if you are in position to do so, by interacting with markets. #DecodingTheQuantMarket #QuantitativeTrading #MachineLearning #AI #ChatGPT4
I assume it also generated the references. That is incredibly cool, but I’m kind of lost with how they are directly connected to the text, e.g. I do remember the Sculley et al paper from SIGKDD ‘12, though iirc it’s about large scale classification with extreme imbalances 🤷♂️ maybe it sees something I don’t lol.
Out of curiosity, is it hallucinations cleaned up?
What would it's output be if we ask "Summarize and reference all the papers applying NLP to financial market forecasting"?
This doc offers a great overview of the broad concepts. However, it falls short of offering concrete real world use cases. Fro example, how would I use a moving average as a feature on a historical stock data with gaps, then run a deep learning to filter out weak signals. Or time series prediction on the equity curve to trade the strategy instead of the asset itself. Also, the audience is assumed to be either or both retail (can do just fine with TA or fundamentals, or portfolio optimisation) ani institutions (which would rather do high volume hft mm on order book imbalance for ex) Still, honestly a great read. Thanks chat gpt for it, Ani Gautier Marti for prompting it ☺️🙏
Gautier, as an algorithmic trader myself, this is an incredibly written document that can be comprehended by in my opinion, all backgrounds of finance. I will have to save this document for reference. Thanks for sharing!
Thanks so much for sharing. Thumbs up for Chapter 5, the source of the data is key, in particular for ML.
What prompt did you use Gautier? :)
Thanks for Sharing! 😁 Gautier Marti
Cryptocurrency Derivatives
11moCan you pls re-explain how this was produced. Tks