Making Money with Automated Trading – My “Smart” Trade-bot 7/19/15

Is it possible to have an automated trading program running on an algorithm would make me money consistently? After 9 months of development work on weekends, I think I have come one step closer to achieving that. Using Ninja Trader software with Bloodhound I have created an algorithm that will run less than 2 hours a day and will trade Nasdaq e-mini futures (/NQ) generating 100%+ returns in a year per the back test. I tested the system from end of June last year to end of June this year. The test shows that it would have generated a little over 120% return. I prefer to have an exit based on the trading logic but also using a wide take profit of 160 ticks and a trailing stop loss of 120 ticks. I figured in trading commissions and a 1 tick slippage. If my assumptions are optimistic, and we had 2 ticks slippage it should still be very positive.

Here is the equity curve generated by the program based on trading 1 contract of NQ… it looks good but the drawdowns can be 25% and maybe worse in real-time. The win rate is about 55% and that means this will generate 8 losing trades in a row of about 3% loss on average per trade, some time in a year. Money management is a key and one cannot be over leveraged. The equity curve is based on 1 contract. Multiply it by a larger number of contracts and the returns start looking even better. I believe it is possible to trade this with an account of $10,000 per 1 contract traded.

The big question is how useful is a back test going forward? I ask that all the time. Yes, the future is unlikely to be like the past. But when the markets behave in a more volatile fashion is that not a time when randomness is left behind and you have bursts of more predictable behavior? And if I take out my emotional participation and let the trade-bot run on its own then I should have a more objective repeatable result.

Next steps is to run the program using a paper account using real-time live data and see what the results are over 3 months. Using real-time data and forward testing the program I will have another piece of information before going to real money. The equity curve needs to show similar behavior over time in forward testing. There are many pitfalls to doing automated trading. I wish I had a giant hedge fund who would allow me to move this fast forward but I intend to make this happen either ways and will not stop learning and testing other systems in case this does not pan out.