Course Name: Trading Strategies with News and Tweets, Section No: 12, Unit No: 5, Unit type: Notebook
After fetching the artciles where does it get stored?
because you need to pull these artciles to do the sentiment analysis correct? or you dont need to store them?
but the next unit it says read from the spreadsheet.
The articles you pull are used to do the sentiment analysis. You can store them in a python pickle if you want to save the data for future use. The notebooks on the portal show how to fetch the news data and perform sentiment analysis on the same.
In Section 12, Unit 9 Notebook, the description says: " We have stored the sample article for stock Apple from sample_apple_article excel file.".
This reading from the CSV is an effort to avoid repeating the code part for the article fetching (already covered in Section 12, Unit 5) and focus more on the Qualitative Features.
If you think you need to save the news articles as you fetch them, you can use the python pickle, or convert it into a datagram and use the to_csv function.
Hope this helps!
ah okay thank you - so you dont have to store it then.
I understand you need to put through keywords to pull articles to do the sentiment analysis, but what if you dont want to do that yet. I would rather let the daily articles that I pull tell me what are the main words that has been repeated to give me an idea what the talk is about for that day. Do a count let's say more than 10 repeat to be more significant.
Hi Ludovic,
The keywords do you mean "Apple" or "AAPL"? The count for the Apple or AAPL keyword is to identify the mention of the particular stock you want to find the in the news articles.
If you want to find "what the talk is about" for that day, then more complex NLP techniques would need to be used. You might count the frequency of word mentioned, but then you will have to remove stop words and other common English words so as not to find a pattern in the non-essential words. But this approach will have an added effort to map that particular piece of news back to the ticker you are trading.
In short, it's better to find out news about Apple and see if it's positive or negative.
It's more challenging to find good or bad news first, and then see how it affects Apple.
Hope this helps!
but not everyday there is google news…
so how do you come up with trading ideas if you have to think about what keywords to indentify?
you would have to manually key in a bunch of symbols
Yes!
Sentiment driven trading signals are not coming in at fixed intervals. They depend on some news-worthy event.
You definitely need to enter the ticker symbols you want to monitor. You don't have to identify keywords. just the ticker symbols. then scan the news for those ticker symbols.
I think the project/strategy you are talking about is more along the lines of NLP combined with sentiment analysis.
yeah so at the moment i dont particular want to do automated trading base on the sentiment scoring yet.
I want to be able to give me a report daily to get an idea what people are talking about and then do a deeper dive if that news can move the needle.
Hi Ludovic,
That's a great idea! Do follow the guidelines mentioned in the above comments on how to remove the stop words and then find the frequency of the "talked about" topic of the day.
Thanks!