Added debugging, LLM response caching, and trade visualization preparation.

LAAlalala

Nothing major

Again architecture update

Added new architecture, it seems to better describe how the project is developing

Added new architecture, seems to more accurately describe how the project has developed

feat(backtesting): Implement sequential event-based news trading system

Added new aligned data for 2week and 6days for before and after a major event, this allows me to refine the promopt easier with the specific data. Also added NewsController to framework and refactored Bakctester and NewsStrategy as sperate classes, this allows the backtester to not get bloated and just encapsulate the relevant elements in their respective places. The backtester will be able to load sets of 2 week data (1 week before and 1 after) around a n event and then do the analysis on the news to see if the Deeper analysis generates a signal and also filter out all the noise surrounding the event. TODO - Check everything functions, claude died, then design a metric for calculating how succesful the LLM is against some optimal value and then start the prompt refining process

Added an edentdataextractor that splits the major news data into 1 week before and 1 after so i can start tuning the LLM's prompts on known causations, might also split into 3 days and do more than 10 top moves, Also seperated the newsAnalyzer out of the statisticalAnalysis as that find the categorieis and isnt needed for other analysis. Next to load the 2 week aligned data into backtester and start tuning the model promots

Enhanced LLMAnalyzer for robust API integration and response handling

feat(analysis): implement statistical price move analysis with LLM integration

More House keeping, tidied aload of scripts and will be starting the major quant stuff next like batch LLM news analysis and statistical analysis of the movement caused by specific news and types of news

More house keeping, moved test scripts into their own folder and changed the import path, next i will sort out the filepath saves for aligneddata and sentiment analysis, everything needs tidying lol

Just some house keeping, responses for real time analysis now save to the reponse subfolder

Added JSON logging for prompt results and corresponding articles. Results are saved to a 'responses' subfolder. Added a 1-second delay between processing articles to prevent API rate-limit issues.