Trading tools

Trading tools refer to utilities that can be used to process stock symbols. Stocks in any market are characterized by a symbol (ie. name) and some historical data. The following tools take historical data and builds a simple regression model to find interesting companies in the market.

There are two parts in this repository. One the one hand, scripts to run locally and process historical data of stock symbols. On the other hand, an AppEngine frontend to run this functionality automatically.

Build

Install python and protocol buffers (more information on protocol buffers can be found here https://github.com/google/protobuf).

Build protocol buffers for this project:

$ cd build
$ ./update_protos.sh

Local tools

Get some historical data

• From a list of stocks:

    $ mkdir -p data/csv
    $ ./get_historical_data.py --filename config/symbols_little --output_path data/csv --api_key XX
  

• A single symbol:

    $ mkdir -p data/csv
    $ ./get_historical_data.py --symbol GOOG --output_path data/csv --overwrite --api_key XX
  

Generating data to get polynomial models

• Multiple symbols in parallel:

    $ mkdir data/res
    $ for f in `ls data/csv/*.csv`; do echo $f; done > /tmp/list
    $ ./process_symbols.py --filename /tmp/list --num_days 40 --window 5 --output_path data/res/ --num_threads=10 --iexcloud_token XX
  

• One symbol:

    $ mkdir data/res
    $ ./process_symbol_data.py --filename data/csv/CINE.L-2010-2015-week.csv --num_days 40 --window 5 --output_path data/res  --iexcloud_token XX
  

Filtering symbols

In order to filter information of a set of stock symbols, you can do it locally too:

$ mkdir data/filtered
$ for f in data/res/*.json; do echo $f >> /tmp/res_list; done
$ ./filter_symbols.py --filename /tmp/res_list --output_path data/filtered/ --filter filters/soft.json

You can look at the mean percentage changes over the number of weeks each symbol has been processed for:

$  grep mean: data/filtered/*.res | sort -k 2

AppEngine

An AppEngine frontend can be found in the gae/ folder. First of all, please install the python tools from https://cloud.google.com/appengine/downloads#Google_App_Engine_SDK_for_Python.

You can start the server by doing the following:

$ cd gae
$ ./start_server.sh

Then you can visit http://localhost:8080 for testing locally.

You can have a look at a running instance of this AppEngine here:

http://deep-trading.appspot.com/

Note that this will only let you look at the previous reports.

TODO:

• Review data structure to minimize reads and writes.
• Have the list of symbols in the database to keep track of unknown symbols.