Wednesday, November 7, 2012

Trades Updated 11/7/2012

I got around to updating my trades. All of my October trades are now available for viewing. It's been a pretty bad recent few weeks. I was fortunately a little more patient than I was before, and I opened new positions further away from the top. Nevertheless, I'm glad I was not leveraged when the market started moving down. I currently do not have any more money to open any new positions and I am taking a pretty painful loss as the market drops.

I've begun trying to build my own algorithmic trading system. I thought of some ways to reduce the amount of work that I would have to put in it by making it less automated. Basically, the system will spit out trading ideas, and I would have to manually approve before the actual trade is sent out. I'm making the system trade essentially the same way that I do, so that I can effectively skip out on doing backtesting. I would adjust parameters as trades get spit out until I end up being able to approve most of them.

I've read a number of algorithmic trading textbooks recently. I figured that if I read enough of them, all the different pieces would come together in my head and it would help me build my own trading system. This is the same way that I learned accrual accounting and eventually developed my current investing strategies. I hope to be successful with the algo trading in the same manner.

On another note, while trying to transform all of my thoughts into code, I began realizing that my trade executions are not purely fundamental analysis. Apparently, based on the mathematical definitions of momentum trading, I actually incorporate a lot of momentum factors into opening and closing trades. I cannot build a high frequency trading system purely on fundamentals, since it would probably execute trades 3 times a year. So recently, I've been looking at other strategies that used by algo trading systems.

2 comments:

  1. Very honest of you to post your losses, I respect that and it adds to your credibility.

    A few questions about the post:

    -How are you getting the raw data to plug into your future algorithms? Is the acquisition of data going to be manual or automated?

    -Also what properties of momentum does the maths tell you that you are using?

    ReplyDelete
  2. I'm going to be using the API from Interactive Brokers. I've already written a service to pull data and store it into a database. Currently only storing tick data for 3 different stocks. I know I can pull daily prices easily from yahoo finance, and I plan to use that for historical data.

    The momentum formula that I found was Change in Price / Change in Time. While this is obviously meaningless without the use of my discounted cash flow analysis, I mainly enter positions after I see the market and the target stock drop for a couple of days, to make sure that I buy as the price is falling. One of the algos that I plan to build will use a multi-factor model which incorporates my discounted cash flow analysis and momentum, and possibly moving averages as factors.

    ReplyDelete