System Diversification

Discussions about the testing and simulation of mechanical trading systems using historical data and other methods. Trading Blox Customers should post Trading Blox specific questions in the Customer Support forum.
Post Reply
GEKKO
Contributor
Contributor
Posts: 2
Joined: Sun Apr 27, 2003 4:18 am
Location: Brisbane, Australia

System Diversification

Post by GEKKO » Sun Apr 27, 2003 9:56 pm

With hindsight on my side I’ve now discovered quite a serious flaw in my trading plan/strategy. After going through an extensive research & development process and eventually coming up with a “systemâ€

Kiwi
Roundtable Knight
Roundtable Knight
Posts: 513
Joined: Wed Apr 16, 2003 1:18 am
Location: Nowhere near

Post by Kiwi » Mon Apr 28, 2003 12:37 am

Gekko,

Can you tell us more both about your strategy and about the nature of the drawdown? On the drawdown can you also provide information about the differences between the theoretical behaviour you saw in the forward and back tests and the actual behaviour you have been experiencing (slippage etc might be relevant)?

What was the historical drawdown? Also was the 80% on actual trading?

For comments on diversification it would help to know what you are trading, what time frame, what else you would consider?

John

AbsoluteReturn
Contributing Member
Contributing Member
Posts: 8
Joined: Sun Apr 27, 2003 10:05 am

Post by AbsoluteReturn » Mon Apr 28, 2003 12:56 am

GEKKO,

Could you please tell us, if I understand you correctly: You have retrurned 80% in the first year and after this you are trading now 12 times the number of contracts you have traded at the beginning? This seems to be an enorous increase!
Could you say us something about your position sizing?

Kind regards

AbsoluteReturn

GEKKO
Contributor
Contributor
Posts: 2
Joined: Sun Apr 27, 2003 4:18 am
Location: Brisbane, Australia

Post by GEKKO » Mon Apr 28, 2003 1:17 am

Hi Kiwi,

Q1On the drawdown can you also provide information about the differences between the theoretical behaviour you saw in the forward and back tests and the actual behaviour you have been experiencing (slippage etc might be relevant)?

A1.The current behaviour is well within the historically tested range and therefore so is the drawdown. In terms of capital requirement I trade 3 times the largest historical dd per contract.

The forward testing was done live, trading on 1 lot for an entire year. The slippage I’m experiencing now is basically the same as before. I’m really not questioning the validity of my current system. My question relates to ways I can diversify into mulit systems, in different markets on various time frames in the hope of smoothening out my total equity curve and eliminate the “diversifiable riskâ€

Kiwi
Roundtable Knight
Roundtable Knight
Posts: 513
Joined: Wed Apr 16, 2003 1:18 am
Location: Nowhere near

Post by Kiwi » Mon Apr 28, 2003 2:11 am

At the risk of being simplistic I'd suggest two strategies both of which could be used if desired:

1) if your system is counter trend can you find a trend based system that is almost certain to be profitable when your counter trend system is unprofitable (or vice versa 8) )

2) have a look at the good systems discussion on this site and either buy or develop a system (or 2) that trades long term on a range of 10-20 futures contracts.

You cant guarantee that 2) wont have its worst drawdown at the same time as your current system but you may well be able to do so with 1). Does someone else want to add to this please?

John

enigma
Roundtable Fellow
Roundtable Fellow
Posts: 99
Joined: Tue Apr 29, 2003 6:56 am
Location: UK

Post by enigma » Wed Apr 30, 2003 11:35 am

Are there any significant differences in terms of performances between the following:

1. Using a single system across different markets.
2. Using mutiple systems in a single market.
3. Using multiple systems across different markets.
4. Using a single system in a single market across different time frames.
5. Using a single system across different markets and different time frames.
6. Using multiple systems in a single market across different time frames.
7. Using multiple systems across different markets and different time frames.

If there are, I would be interested to see how forum members would rank them in terms of preferences. Thanks.

Kiwi
Roundtable Knight
Roundtable Knight
Posts: 513
Joined: Wed Apr 16, 2003 1:18 am
Location: Nowhere near

Post by Kiwi » Wed Apr 30, 2003 7:27 pm

If you define performance as a smooth equity curve (high return for max/average drawdown or std dev of returns or some such) then I'd suggest the ratings would go (from best to worst):

7. Using multiple systems across different markets and different time frames.

5. Using a single system across different markets and different time frames
3. Using multiple systems across different markets.
1. Using a single system across different markets.
6. Using multiple systems in a single market across different time frames.

4. Using a single system in a single market across different time frames.
2. Using mutiple systems in a single market.

Its very hard to be precise about this so I grouped them in clusters.

Multiple markets are good because they reduce your exposure to a single negative event but as c.f. said they need to offer Power (or a minimum return) to be worth including.

Multiple systems are good because there closed trade equities are likely to be weakly correlated. They are even better if they tend to be taking money from different parts of a trend which leads to

Multiple timeframe are good because one system can be short when the other is long reducing correlation of open trade equity.

I think that you could trade two systems, one trend and the other counter trend with reasonable comfort especially if on a short time frame. Trading one system on multiple markets usually means medium/long term trend following and the challenge is to get used to drawdowns that are likely to be in the order of 30% and last up to 8 months. I'd prefer to be trading your option 7 so that I can be philosophical about the long drawdowns while my short term systems keep me happy and in pocket money :wink:

John

Josh M.
Senior Member
Senior Member
Posts: 38
Joined: Mon Apr 21, 2003 2:22 am
Contact:

diversity

Post by Josh M. » Thu May 22, 2003 8:04 pm

I echo many others in saying system diversity is key as it is the only free lunch in the markets. I would say, though, trading a congestion and trendfollowing system at the same time may be difficult and give a net flat position much of the time.

edward kim
Roundtable Knight
Roundtable Knight
Posts: 344
Joined: Sun Apr 20, 2003 2:42 pm
Location: Silicon Valley / San Jose, CA USA
Contact:

Re: System Diversification

Post by edward kim » Sun May 25, 2003 11:51 am

GEKKO wrote:Almost five months later I’m currently in a drawdown of 20%, trading 12 times the number of contracts I was previously.
Hi Gekko,

AbsoluteReturn was asking why you are now trading twelve times as many contracts in real time vs. when you were testing, and that maybe you have a postition sizing issue. We would like to know if that solved your problem.

Edward

Kev
Contributor
Contributor
Posts: 3
Joined: Thu May 15, 2003 4:05 pm

Trading Multiple Systems.

Post by Kev » Sun Sep 28, 2003 6:13 pm

The portfolio diversification benefits of trading Aberration, Aztec. and I-master in various combinations takes up 31 of the 72 pages in the I-master manual. Three different concepts at the moment having unperfect years. As the I-master disclaimer states, "There is no guarantee that future system performance will match that of the past."
Which makes me think about my own attempts to develop
non-correlated systems. Just because they've historically been non-correlated doesn't mean they won't correlate in the future.

Kiwi
Roundtable Knight
Roundtable Knight
Posts: 513
Joined: Wed Apr 16, 2003 1:18 am
Location: Nowhere near

Post by Kiwi » Sun Sep 28, 2003 8:45 pm

You have an excellent point Kev,

To have a hope of overcoming it I think you have to go beyond statistical testing into the question of why system a) will make money while system b) loses it and then test it in those conditions. Its not enough to just look at the correlations and say "historically they didnt correlate" you need to look at the way that they take money from the markets and try to understand what would be necessary for them both to lose money at the same time.

Even after that effort you may find that you fall into the trap of solving the problem that last hurt you only to discover a different one down the road :-)

So: does anyone want to answer the question - why might all of Keith's systems be failing at the same time?

John

Sir G
Moderator
Moderator
Posts: 243
Joined: Wed Apr 16, 2003 12:21 am
Location: Salt Lake City, Utah

Post by Sir G » Mon Sep 29, 2003 12:43 am

Hi Kiwi-
So: does anyone want to answer the question - why might all of Keith's systems be failing at the same time?
For what it is worth, I know nothing about Keith and his systems beyond what I have read on the forums…But I will give you a quick answer to your question.

IMHO, the answer to why Keith’s systems are failing at the same time is…. Risk.

Aberration is a LT systems that will, unless some sort of logic is employed to stop the bleeding, be prone to bleeding like all other LT systems. And sometimes the flow of blood is worse then other times. Survival, not greed is the key. I am forever amazed that no one seems to have a plan for the worst-case scenario.

I-Master has risk by being a one sector system. And the system from what I have read, doesn’t show any signs of life in other sectors. It is a very optimistic assumption to believe that any one market or sector will always ebb & flow a certain way.

I would have to assume that it is the luck of the draw that has those systems faltering at the same time. It isn't some act of God against the vendor and his clients..

Risk is real and it is inherit in the way we design our systems.
The portfolio diversification benefits of trading Aberration, Aztec. and I-master in various combinations takes up 31 of the 72 pages in the I-master manual. Three different concepts at the moment having unperfect years. As the I-master disclaimer states, "There is no guarantee that future system performance will match that of the past."
Which makes me think about my own attempts to develop
non-correlated systems. Just because they've historically been non-correlated doesn't mean they won't correlate in the future.
Those 31 pages are making a very optimistic assumption that his systems wont break down. I remember watching a tennis match that John McEnroe was being the color commentator. He mentioned that from the view of the action he had up in the box where he was sitting… that if he was playing he wouldn’t have missed a shot.

Hindsight is perfect and it helps sells systems and it also adds to the volume of reading material… but it doesn’t do much for the questions that really matter the most. It is simply fluff that probably made the idea of purchasing all of his systems seem like a good idea.

Loyalty to the opportunities that your systems are trying to exploit should always be balanced with your loyalty to your equity.

Gordon

Post Reply