Software designed to identify and monitor social/historical cues for stock movement
csv
os
selenium
numpy
matplotlib
sklearn
CSV containing raw historical data for each stock
TSLA_RawData.csv
Date,Open,High,Low,Close,Volume
09/02/2015,245.3,247.88,239.78,247.69,4629174
09/03/2015,252.06,252.08,245,245.57,4194772
09/04/2015,240.89,244.09,238.2,241.93,3689153
This program will read in Raw Data and write out Training Data
Infile Path = path/to/RawData
Outfile Path = path/to/TrainingData
Training Data can then be fed directly to the classification algorithm
This program reads in Training Data and feeds it to the classification algorithm via sci-kit learn
Please visit: http://scikit-learn.org/stable/auto_examples/svm/plot_rbf_parameters.html before adjusting parameters
Part of the training data is reserved for accuracy testing (this parameter can be changed)
Note that accuracy may not be indicative of good/bad clustering in terms of finding hot spots
Left: Training (Dark) and Testing (Light) Data
Right: Training and Testing Data with gradient clustering
Blue = Positive (+) / Red = Negative (-)
DataCrawler.py is a headless web browser using selenium
This tool can be used to access daily stock indicators after the model is trained with historical data
Upon execution, the program will:
1. Travel to Fidelity
2. Access their research tools for a particular stock
3. Automatically fill credentials (requires a Fidelity account)
4. Access advanced charting mode
5. Select indicators of choice
6. Download a spreadsheet of the data
This data can then be automatically plugged into the model and predict whether the conditions are favorable to make a trade
Note: To edit this program, considering using Firebug/Firepath to identify your XPaths of interest