Katelynn's Report

Katelynn's Report

(US Market)


Font Size:

AI (Artificial Intelligence) Sentiment

Katelynn's Report provides AI based sentiment to facilitate investors with faster and more efficient decision making. Our AI sentiment system is currently based on two independent deep learning architectures, each makes sentiment prediction that belongs to positive, negative, or neutral.

ID Abbrev Architecture Judgement Briefing
AI 1 GPT-4 Transformer based large language model Strong Positive
Strong Negative
Others (AI 2) TBiGRU-A Transformer + Bidirectional GRU + Attention Pos + Pos
Pos + Neu
(Neu + Neu)|(Pos + Neg)
Neg + Neu
Neg + Neg
Training: financial news manually annotated by professional accountant on both relevance (0,1,2) and sentiment (positive, negative, neutral)
Agreement Level: relevance(90%), sentiment(83%) on test dataset annotated by our accountant
Others (AI 3) FinBERT Bidirectional Encoder Representations from Transformers (retrained with financial data) FinBERT is a pre-trained NLP model to analyze sentiment of financial text. It is built by further training the BERT language model in the finance domain, using a large financial corpus and thereby fine-tuning it for financial sentiment classification. For more information please check https://arxiv.org/pdf/1908.10063.pdf
Agreement Level: sentiment(75%) on training + test dataset annotated by our accountant
sentiment(89.5%) on third party report FinBERT - A Large Language Model for Extracting Information from Financial Text

Agreement level evaluated as of 2022-12-20

We emphasize that sentiment is highly subjective, and there are typically only 70~80% consistency across individual annotators. Our AI sentiment system only helps to quickly narrow down your scope of search, or get a sense of overall market sentiment on your stock of interest. However, it is still your responsibility to make sure important news are not skipped.