A Machine Learning Model to predict prices of certain stocks

AI & Machine learning
1

Hi there:



  Since many of you probably know… Machine learning is used by hedge funds to predict stock's prices, but neverteless… Retail  traders can also use ML to predict prices… 



On the other hand, there has always been bubbles in the stock market… what I mean with that is Stocks that tent to go insanely up… like for example Stock "Tango Terapeutics" in the biotechnology sector (NASDAQ:  TNGX), today (2023-08-09) has gone from 5 dollars all the way throught 9:39 (yesterday's close price was 3,83 That is, it went up more than 200 % from yesterday's close price),  So I was wondering… if maybe it is possible to make a forecast for the "low" of this stock…for tomorrow…



If it is possible… What would be the features to be used??  and how many features should we use…?? Did anybody here consider to make this kinds of forecasts??



and in general… Is it possible to make this kind of forecasting for this kind of bubbles??  What about RENB (Renovaro Biosciences Inc.)?? How do you determine the features and How many features to be used here ??



Thanks

2

Hi Ghery,



This sounds quite intriguing. However, it might be a little difficult due to the following reasons:

1. Predicting stock price movements, especially during market bubbles, can be challenging due to their complex and often unpredictable nature. The predictions might not be very accurate.

2. During bubble periods, the available data might be relatively limited due to the unique and often short-lived nature of these events. So again this can lead to inaccurate predictions.

3. Impact of bubbles can vary significantly across different sectors. Each sector has its own set of dynamics, factors, and market influencers. What drives a bubble in the biotechnology sector might be quite different from what drives a bubble in technology or finance. So the predictions may not apply to all sectors in the same manner.



Hope you find this helpful!

3

Hello:



Thanks for your answer… Neverteless I think there are some issues to be considered:



1.- I understand that price movements can be very hard to predict, but…  if your go to the OHLCV data from bubbles, they have also been bubbles in the past… (what I mean is that they have had huge price movements (20%, 40 % somethimes even more than 100 %) in a relatively short period of time (hours, days, etc.), perhaps there is some information there that can be captured with a mathematical model.  but only include ohlcv data (not financial ratios, not sales, not debt,  just ohlcv DATA…)



2.- Given what I stated previously mybe we can just rely on past ohlcv values… not other information.



3.- you also stated that "Each sector has its own set of dynamics, factors, and market influencers. " … well one thing in common that I noted is… a large  increase in volume… when this bubbles happen…( volume is  contained within ohlcv data), perhaps a variable that accounts for a sudden increase in volume ( like relative volume ) could be a "feature" to select…



What do you think ??

4

Hi Ghery,



Yes, you can focus on OHLCV data but you have to approach this as an experimental process. Financial markets are complex, and predicting short-term movements, especially during unusual events like bubbles, is quite challenging. So its better to exercise caution and consider the risks.



Thanks

Rushda