Text representation for direction prediction of share market

[KunalAnand57]

Title

Text representation for direction prediction of share market

Team Name

Stocks Savvy

Email

202318035@daiict.ac.in

Team Member 1 Name

Ashutosh Anand

Team Member 1 Id

202318035

Team Member 2 Name

Kunal Anand

Team Member 2 Id

202318057

Team Member 3 Name

Anjali Singh

Team Member 3 Id

202318050

Team Member 4 Name

NaN

Team Member 4 Id

NaN

Category

Reproducibility

Problem Statement

The problem this study addresses is predicting the direction of stock market close price movement using financial news articles and historical price data. The study aims to investigate the impact of news articles on stock price movement and compare the effectiveness of using only historical prices, only news articles, and a combination of both.

Evaluation Strategy

The evaluation strategy involves using accuracy, ROC-AUC, and normalized cross-correlation (NCC) to assess the performance of the prediction models. While accuracy measures the correct predictions, ROC-AUC captures the discriminative power, and NCC evaluates the correlation between predicted and actual time series with respect to time lags or leads.

Dataset

https://drive.google.com/file/d/1BKMU6zvg_IDrxCC5a0tK2llc8I8J34qu/view

Resources

https://www.sciencedirect.com/science/article/pii/S0957417422015561

[parth126]

• The project would boil down to making API calls, which is not sufficient
• Suggested a different topic. a.One possibility is stock prediction using textual data (e.g. news). This is under the condition that there exists a dataset that can be used for evaluation.

[parth126]

Advised to do this as a reproducibility experiment by implementing and building upon the paper by Ganguly et al.

[KunalAnand57]

Dear Sir, We have updated our proposal idea as advised in the GitHub issue. Kindly review the changes at your convenience. Looking forward to your feedback .

[parth126]

Looks good