Monday, December 3, 2012

Algorithmic Trading System Update 12/3/2012

I've just written something to pull historical tick data. I've left these in flat csv files rather than putting them into a database. I was originally not planning to do any backtesting in order to save some work and time, but as I put more thought into it, building a backtesting system doesn't seem that complex. I'm not at a stage where I want to pay for historical tick data, so I only have about a year's worth of tick data for a couple of stocks. Fortunately, they are on stocks that I want my algorithm to run against.

The next thing on the task list is to modify my existing system to feed in the historical data from the flat files instead of from the API. My change of heart came from playing with a site called Quantopian, which lets you backtest strategies written in Python against minute data for the past 10 years. Its simplicity has demystified the illusion that a backtesting system must be complex and intimidating.

No comments:

Post a Comment