cpptrader

This is a C++ framework for implementing and testing trading systems for the stock market. It supports up to minute-level resolution price data. Data is assumed to be stored in a MySQL datatbase.

Dependencies

• MySQL connector
• boost-date-time
• boost-system
• boost-filesystem
• libcurl

Installation

Install the dependencies and then clone and compile cppTrader :

git clone https://github.com/zwinoker/cpptrader.git
make

The executable is called run_simulation.

Data

Minute level data was scraped from Google Finance using the R script getPrices.r and stored in a MySQL database in the following format: Tables are named according to symbol (EX: "AAPL_", including the underscore). Each row in a table has the columns { Date-Time, Close, High, Low, Open, Volume } for each minute. Note that data is not included in this repository. Users must collect it themselves.

The symbols we collect data on and use are stored in symbolslist.csv. The current list is made up of symbols we can reliably get minute-level data on using the Google Finance API.

Setting input parameters

Set parameters in simulation.in. These are:

file where we save portfolio holdings
start date-time
end date-time
live?
tz_offset
max_ticks
frequency
tradefrequency
MYSQL_SERVER
MYSQL_USERNAME
MYSQL_PASSWORD
MYSQL_DBNAME

live? is a boolean that is false if we're running a backtest, and true otherwise. tz_offset is your timezone offset (in hours) from New York City. max_ticks is the number of previous time steps our algorithm will consider at any given time. frequency is the number of time steps between each stored data point. tradefrequency is the number of time steps we wait between successive executions of the trading algorithm.

The date-times are in the format YYYY-MM-DD HH:MM:SS, for example 2015-03-05 15:09:00.

Usage

Modify algo.cpp to implement and execute your trading algorithm. The basic flow here is:

• Get relevant information from the Stock_Market object
• Use your trading system to generate one or more Order objects
• Give these to the Portfolio object using the order() method. These will get executed either immediately (if they are market orders) or at a certain price or date-time (if they are limit orders).

Once the input parameters are set and the trading system is implemented, run

make
./run_simulation < simulation.in

Useful methods

Stock_Market.cpp

The stock_market object contains a few methods for retrieving price data. Use these for getting input data for your trading system.

Order.cpp

Contains constructors for market and limit orders. Limit orders are also constructed using a stoploss price and a bool indicating if we want to buy/sell above or below that price.

Portfolio.cpp

• order(Order) -- submit an Order.
• show_orderbook() -- prints your outstanding orders to the terminal
• show_portfolio() -- prints current portfolio holdings to the terminal

Algo.cpp

• sell_all(Portfolio&, Stock_Market&) -- can be used to liquidate a portfolio at the current market value.
• eval(Portfolio&, Stock_Market&) -- called every frequency*tradefrequency minutes to execute trading algorithm.

Results

The running value of the portfolio is written to portfolio.value. The order history is written to portfolio.orderhistory. Portoflio holdings are saved to portfolio.save.