UNIT calculation for CFDs & Turtle-System-2

Discussions about Money Management and Risk Control.
Post Reply
mardoc
Contributor
Contributor
Posts: 4
Joined: Mon Oct 01, 2007 3:14 am
Location: Munich

UNIT calculation for CFDs & Turtle-System-2

Post by mardoc » Thu Oct 04, 2007 6:14 am

Hi all,
actually, I have coded the TurtleTrading-System-2 into my program (its Tradecision), but not completed, because I can not add additional UNITs (pyramiding isn´t possible till now)

anyway: I would have another question:
how is the calculation of units done for CFDs?
As I´ve read, stocks whould have a big point from 1.
I would say, this normally whould be the same as for CFDs, with the difference, that for CFDs, you have to provide a margin. In most cases, the margin is 1% from entryprice per CFD. For CFD-stocks, the margin is 3%-5%.. this are the margin-requirements in case of my broker...

What in general you think about CFDs?

another thing:
If I simulate the system with an initial account of 10000EUR, in many cases the result of my current calculation is: unit = 1. This is, because I see CFDs equal to stocks. In this case with trading CFDs, this would mean, if the entryprice of the index = f.e. 8000 , the margin = 80. If the underlying goes down from 8000 to 7920, I´m out of the market.
This would also mean, that it´s necessary to have at least 20000EUR as a minimum of account-size to have more "space" for the CFDs but could also mean, that it´s not a good idea to trade CFDs for long-term, but only for intra-day-trading, or am I wrong?

I also realized, that the system-2, without any filter, performed after the year 2000 not really good, in comparison till the year 2000. Could this be the case, that now, it´s not really a good time for breakout-systems?

Actually, I´m working to identify some good filters for the system, because I like the idea of this breakout-system and also because it´s simple. I´ve realized, if I add f.e: Parabolic-SAR as condition for entry and exit, the system-result is more stable... let´s see.

thanx for feedback
Martin

sluggo
Roundtable Knight
Roundtable Knight
Posts: 2986
Joined: Fri Jun 11, 2004 2:50 pm

Re: UNIT calculation for CFDs & Turtle-System-2

Post by sluggo » Thu Oct 04, 2007 6:58 am

mardoc wrote:Actually, I´m working to identify some good filters for the system
I think this is a good idea. The basic channel breakout entry and exit, which Trading Blox calls "Donchian" in honor of its early advocate Richard Donchian, is often paired with filters. The book (and the Trading Blox software) explores Donchian + dual_moving_average filter, see book p.68 and p.131. A message posted (here) on this website, explores Donchian + PGO filter. There are lots of other ideas to be conceived and tested.

For example, you might have a look at the (trendfollowing entry, random exit) experiments found in (this thread). You will notice that dual_moving_average was used here as an entry (see figure "result4.png"). As an entry it managed to make money when paired with random exits. Hmmm, Trading Blox used dual_moving_average to filter the Donchian channel breakout, and dual_moving_average also made money as an entry when paired with random exits. Hmmm. Maybe some of the other trend entry ideas in that thread about trend entries plus random exits, might be used as filters too.

mardoc
Contributor
Contributor
Posts: 4
Joined: Mon Oct 01, 2007 3:14 am
Location: Munich

Post by mardoc » Mon Oct 08, 2007 2:46 am

Hi sluggo,
thanx for your answer. The dual-mov-avg as filter for entry worx pretty good, but only if there is a trend available and this would mean, that you catch it, for my understanding allmost (too) late.

Well, during reading c.f.´ book, I was thinking about a break-out system, where the high- lowbreakout are not based on a fix value. I mean, if we think about a 20-day break-out, the system should have good results, if the movement of the market reflects the 20-day intervall, but this isn´t "linked" on the volatility or on N.

Does it make sense, to have a break-out system, where the days of the breakout are linked to N? In case of a stable movement on a trend with a low N, the calculated days will be less than 20, f.e from 20 to 15, so the generated entry-signal would be closer to the market. In case of a high N, the calculated days would be f.e. from 20 to 30.

This would mean to have a breakout system, where the entry signals are aligned to the markets volatility.
Does this make sense?

Regarding CFDs:
is there anyone, who has experiences with CFDs?

thanx!

Post Reply