Considerations for Auto Trading Systems

These are the results of 5 systems we have developed and currently run. The goal of this strategy was to develop a smooth and consistent profit curve by running uncorrelated pairs and products. To prevent curve fitting they have been tested on 30, 50, 60 and 90 minutes. Each system independently has been proven profitable on all these time frames.

A key part of successful algo design is not the % return since this is a function of capital employed - one should focus on: a) profit factor b) trade profit / spread (commission) costs and c) frequency of trades. Together, these systems have an average profit factor of about 1.6 and a profit multiple on the spread of about 10 times.  They also sport profitable returns across the other time frames with between 1500 to 2500 trades for each individual product. 

The percentage returns on the results below assumed maximum capital being employed. This is not possible in real time of course. However, when all these systems run concurrently, use of capital is optimized and the drawdown across the system will be reduced considerably.

*Past returns are not indicative of future returns.

---

---

---

---