This extension makes the latest developments in the field of Natural Language Processing (NLP) available for Wealth-Lab strategies and lets you access thousands of Large Language Models.
This extension allows the backtest of automated news trading strategies and fully automated news trading.
- News Browser: Show all available news articles with all their details.
- Configure up to four Large Language Models
- Choose among two LLM services: Hugging Face and OpenAI.
- Support for four different tasks: Text Classification, Zero Shot Classification, Text Generation/Completion and Conversation/Chat
- Testbed for Large Language Models: Run a single or up to four models for a single or all available news events and inspect detailed responses, results, sentiment values, etc.
- Contains six news related indicators: min. sentiment, avg. sentiment, max. sentiment, number of articles and number of articles with useful sentiment values.
After installation there is a detailed tutorial, verbose descriptions of all functions, a FAQ page and a glossary in WL's help files.
A source of news articles for Wealth-lab. This could be the finantic.News extension or a similar extension.
A paid subscription to the large language model service. (The free plan is limited beyond reason.) The cost of these services is NOT included in the Premium plan fee or the extension price.
The third-party data is subject to the availability of the respective provider (website) and may be delayed or inaccessible periodically due to network or technical reasons. As the response of LLMs is not guaranteed to be accurate, it is your responsibility to confirm that it does not contain errors before utilizing it for any type of backtesting or trading activities. Quantacula LLC and finantic are not to be held liable for any errors in model responses or their unavailability.
Support for two large language model services:
- Hugging Face
Support for four tasks:
- Text Classification: 20'000 models
- Zero Shot Classification: 135 models
- Text Generation/Text Completion: 12'000 models
- Chat/Conversation: 2'000 models
- Test entry.