At the end of April, Jan Roozenburg gave a presentation to the HCC Investment Study Club in Delft on the subject of ‘Radically new ideas for trading systems’.

Jan Rozenburg

It is unusual if you put several algorithms (in my case from old trading systems) together in one new system. Systems usually consist of one algorithm for buy and sell signals with some filters and exits. These are daily systems for the S&P-500 index that work with micro-futures. Nevertheless, I did some research and the new system with many algorithms (algos) is doing amazingly well, significantly better than the joint separate systems put together.HCC!beleggen: new trading system

9 Reasons
The 9 reasons for this are:

  1. Fewer trades and therefore less transaction costs.
  2. Algos together in one system do fewer trades, for example the system is long and a new long signal has no effect.
  3. The position of the new system is max. 1 Long, neutral or 1 Short and separate systems do stack. That leads to multiple positions.
  4. The overlap of algos and therefore the quality determines the reduction in the number of trades. This effect is significant.
  5. Testing the algos in one system is more reliable than a series of separate systems with the same algos.
  6. For an overall result of separate systems, the results must be added up, which leads to a loss of information. It is also very laborious, because the results have to be exported to Excel and edited there. In one system with all algos, the overall result is immediately visible.
  7. There is a better view of curve fitting: the quality of the algo’s contribution is immediately visible. After a few buy and sell signals from different individual algos, finding new reliable entries without overlap becomes difficult.
  8. If the algo is negative in itself but positive for the whole, you can see this immediately. You don’t see this with separate systems.
  9. There is less risk! With one system, the deployment of the number of algos is not limited.

Cons
A disadvantage is that there is less flexibility: the exits apply to all entries in one system and the exits play a minor role, because buy and sell signals neutralize each other. In addition, operating in multiple markets requires separate systems, as well as systems with different ‘bar lengths’.

Results
Here you will find a few results of the various setups.

Day system with 6 long and 5 short entries
• Net Avg. profit from 1/1/200: 2,776/yr (based on 1 MES future)
• MDD: 1,970 exceptionally low!
• Number of trades: 7/yr exceptionally low!
• % winners: 76% exceptionally high!
• RR: 1.3 exceptionally high!
• Return (margin + 5x MDD): 22% (or 12% with 10 x MDD)

Day system with a different character: 8 short entries and long after 4 days
• Net Avg. profit from 1/1/200: 2,085/yr (based on 1 MES future)
• MDD: 2,815
• Number of trades: 25/yr
• % winners: 58%
• RR: 0.7
• Return (margin + 5x MDD): 13% (or 7% with 10 x MDD)

Portfolio of 2 day systems together
• Net Avg. profit from 1/1/200: 4,516/yr (based on 1 MES future)
• MDD: 2,588
• Number of trades: 32/yr
• % winners: x%
• RR: 1.7
• Return (margin + 5x MDD): 25% (or 15% with 10 x MDD)

The test results are exceptionally good compared to the portfolio of autonomous systems. In practice, the systems are automated: the trading systems generate buy and sell signals (Vestics software). After that, the signals go to InteractiveBrokers’ TraderWorkstation.

invest

Look for more information invest.hcc.nl