Proposal Feedback

[wangkaihong]

Hi,

The project seems interesting, but I would like to see more details as it should help clarify your objectives and define the scope of your tasks. For example, here are some details that should be helpful:

1. Briefly introduce future asset returns and risk measurements and explain how you compute them.
2. What data will you download from "yfinance"? Stock prices? News headlines? What form will these data be like? Numerical, categorical, or simply strings?
3. How did you select the model? Why did you choose clustering in the beginning? Does the data you obtained above fit in the clustering model? How do you plan to process the data? How would other models behave? Can you choose one more model for comparison?

Feel free to work on other points I did not mention above as well, but generally, clarifying your tasks will be helpful to plan your project ahead.

[thomas211738]

1) Asset returns were calculated as the profit (Final price - Initial Price). Risk measurements were calculate through the volatility of the stock prices over a year, which is the std of the prices.

2) Yfinance contains both daily stock price data and fundamental analysis data, which contains both numerical, categorical, and simple strings, such as (Price per share, Industry, Company Description).

3) Our current model is based of math, where we choose stocks in the same industry with the lowest volatility. Then, we prove that our model creates a better portfolio than the user's current portfolio by applying linear regression to calculate the potential future earnings. This will show that in the long term, our model has greater earnings than the user's current model. We are also still planning on making another model that uses a neural network and a bunch of features, in order to find which combination of them will lead to the portfolio with the highest earnings.

Let us know if we need to clarify or go into more detail on anything. Thanks!