Stock Picker

• 

  Monkey we trained with our model no cap

• 

  how our data is stored

• 

  Outputs

Inspiration

• Using ML to fix a problem which many people come into, if a stock will go up or down.

  What it does

• Uses data from news article titles, and trains a sentiment model giving each article a sentimentscore from zero to one, and takes an average score from all sentimentscores to give a final output of whether the stock goes up or down.

  How we built it

• Using ML combined with google colab allowed for a powerful ai model to grasp scores and add them to different companies

  Challenges we ran into

• Where do get the headlines, this was a hard feat, but found a github repo containing a .csv file which allowed us, not much data, but a proof of concept. As this model can be used with more recent data and more headlines and become more accurate and more up to date.

  Accomplishments that we're proud of

• Getting it to work

• Exploring and exploiting transformers which are used in powerful models as seen in open ai's gpt-3 and google's various ai models.

  What we learned

• Using many and a variety of machine learning libraries which before this was unknown to me. I didn't know any machine learning before this.

  What's next for Stock Picker

• Using more data and more accurate data to automate stock trades and hopefully make money off it.

Built With

• machine-learning
• python
• stock-market-data