Trading Strategy Advice

[pmn100]

When devising a trading strategy, is it reasonable not to expect the strategy to work across mulitple instruments and multiple time frames? I've just finished reading Way of the Turtle and the author believes that a true trading strategy should be able to do this.

I'd like to hear other people's thoughts on this and whether your own trading strategies work consistently with different instruments (shares/forex/futures) and across different time frames?

Similarly, does it mean that a trading strategy that only performs well on one instrument, is not a valid system? Or is a system that works for specific currency sets, or specific shares, or even just one currency set or one share, still a valid system?

 

[tomorton]

In general I would agree, and many other authors recommend a particular position should be verified by reference to the TA of the same instrument across at least one other time frame.

But bear in mind that not all markets are the same. For example, it will be difficult to scan for trades based on significant gaps between a Close and subsequent Open in some markets where these are not adequately identified (even though they might occur) - have a look at the OHLC bars for the Dow against the FTSE100 for example - the Dow charts hardly ever show a significant Close-to-Open gap.

Also, regular key points in the timeline of some trading sessions - such as 10am on the US markets, the US open on the London markets, will not be adequately represented using some time frames. And the Close is a significant time because after that point, the market is closed to further trades - can the closing point of any 10-minute period since the Open be as significant?

 

[£10kLoser]

Having devised many systems for Forex and gold, my experience is that non work for everything. Additionally, some only work for a short time while the market conditions stay the same.

There may be a few old and basic systems that always work across all instruments but if that were the case, surely we'd know about them and all be rich?

That's the basic flaw with systems - they are all devised using past data so when conditions change, they may not work. In effect, they are curve fitted. This fact makes me laugh when I read so many people (here and elsewhere) say they've bought some EA and back tested it and IT WORKS! Of course it does, that's how it was made.

To assume that market conditions will remain the same is madness. There is obviously no system that always works for everything because there is always an element of randomness in the market. It's a fallacy to believe otherwise and this fact is proven time and time again. The risk-takers become trading geniuses - until things change and they blow up. They were 'lucky' because their strategy (system) fitted the market for a while...

The boring safe traders make far less money while they always allow for possible change and play safe. But, they're still making decent money while the 'genuises' are out of a job.

The only system that works is to change with the market and the instrument. Everybody comes into trading hoping to find some sort of holy grail system (and the idiots come into trading thinking they have it) but history shows us that there is no such thing - only luck for a while.

I spent countless hours looking for THE system and I have many that looked good, until they stopped working!

 

[pmn100]

Having devised many systems for Forex and gold, my experience is that non work for everything. Additionally, some only work for a short time while the market conditions stay the same.

There may be a few old and basic systems that always work across all instruments but if that were the case, surely we'd know about them and all be rich?

That's the basic flaw with systems - they are all devised using past data so when conditions change, they may not work. In effect, they are curve fitted. This fact makes me laugh when I read so many people (here and elsewhere) say they've bought some EA and back tested it and IT WORKS! Of course it does, that's how it was made.

To assume that market conditions will remain the same is madness. There is obviously no system that always works for everything because there is always an element of randomness in the market. It's a fallacy to believe otherwise and this fact is proven time and time again. The risk-takers become trading geniuses - until things change and they blow up. They were 'lucky' because their strategy (system) fitted the market for a while...

The boring safe traders make far less money while they always allow for possible change and play safe. But, they're still making decent money while the 'genuises' are out of a job.

The only system that works is to change with the market and the instrument. Everybody comes into trading hoping to find some sort of holy grail system (and the idiots come into trading thinking they have it) but history shows us that there is no such thing - only luck for a while.

I spent countless hours looking for THE system and I have many that looked good, until they stopped working!

So assuming this, would it not be advisable to stick to trading a very small handful of instruments and find a system for each or all, and adjust those systems as the behaviour of the instruments evolve? By doing so I expect your longevity in the market will be more than that of someone who uses a broader system that flags up potential trades across multiple instruments?

Plenty of people do well trading just one stock or one currency pair. In my mind, you'd have a better chance of success doing this than trying to find a strategy that works across multiple stocks, or multiple currency pairs. Would that be a reasonable assumption, or not necessarily?

 

[Chorlton]

When devising a trading strategy, is it reasonable not to expect the strategy to work across mulitple instruments and multiple time frames? I've just finished reading Way of the Turtle and the author believes that a true trading strategy should be able to do this.

Unfortunately, I disagree. Yes, (in the ideal world) it would be great if a system could perform equally well on multiple Mrkts as it gives the trader additional confidence that the system is robust , etc, etc.

However, I would argue that different Mrkts exhibit different characteristics (due to difference in volatility for example) and as such may not be able to be traded in the same way.

As an example, I know of one very profitable mechanical strategy that has returned excellents results over the last few years on the ASX. However, when I ran this exact model on the LSE (regardless of what testing period I used) the results were nothing short of dismal.

Are we then to assume that the strategy is not robust? Of course not. The system was developed on the ASX, so why must it perform well on other Mrkts??

As another example, if one was to develop a Long-term "Long only" system, why should one be surprised when it performs poorly in a Bear Mrkt?? Afterall, it wasn't developed to trade that condition.

In my mind, trying to find a strategy that works on Multiple Mrkts is not effort well spent. If one is found, I would assume (maybe incorrectly though) that the system results would reflect the fact that it had been "optimised" (directly or indirectly) to suit the different Mrkt Conditions. Personally, instead I would rather trade multiple systems on multiple Mrkts which individually offered a greater return.

All IMO,

Chorlton

 

[£10kLoser]

Just one?

Plenty of people do well trading just one stock or one currency pair. In my mind, you'd have a better chance of success doing this than trying to find a strategy that works across multiple stocks, or multiple currency pairs. Would that be a reasonable assumption, or not necessarily?

I agree, except that there is no need to go to the other extreme and limit yourself to just one. For example, I trade the EUR/USD, USD/JYP and the USD/CHF using the same strategy and another system works well with four currency pairs.

The point being that each month a different pair will do extremely well, while some will produce nothing overall. We have to back test to confirm a system but I rarely back test over more than two or three months. I regularly modify the system(s) and am always on the lookout for it starting to fail. It is usually a mistake to carry on thinking that it will start working again!

 

[Chorlton]

I regularly modify the system(s) and am always on the lookout for it starting to fail. It is usually a mistake to carry on thinking that it will start working again!

I agree 100% !!

 

[ronblack]

I think that when people say a trading system should work across all markets they mean that the general principle employed should be subject to that restriction and not the specific variation of it.

As a result, trading systems based on technical studies work across all markets because you can find at least one system that works on each market.

In other words, the principle is too broad and meaningless. It does not concern specific variations of trading systems.

My 2 cents (as always)

Ron

 

[jayjay121]

so if the markets and TA is just random is it not best that we behave in a random way?, funny thinking about it we probably have more chance being random and reckless than actually trading a system based on TA alone

 

[Chorlton]

so if the markets and TA is just random is it not best that we behave in a random way?, funny thinking about it we probably have more chance being random and reckless than actually trading a system based on TA alone

Personally, I'm not sure if I would agree with the statement in Bold.

Afterall, isn't the whole foundation of TA based on the fact that historical events have a tendancy to repeat themselves?

 

[jayjay121]

i agree and i have a divergence pattern that does in a range condition repeat itself quite often, also it amazes me how trendlines act as s and r, and major s and r definately work imo.............But TA is still random and best used as a secondary tool imo......It can work alone but has to be played in probabilities with r/r and money management..........I think the pro's use levell 2 or fundamentals alongside TA and this is the way i wish to go now as i need further skills to get a better edge.

 

[Adamus]

So how much bad does a successful system have to get before you determine that it's not working anymore?

Referring to your account equity on an account just trading that system, would you abandon it after it gives back 25% of its profit? 50%?

Or do you have some other measure? Or is it a trade secret?

 

[£10kLoser]

so if the markets and TA is just random is it not best that we behave in a random way?, funny thinking about it we probably have more chance being random and reckless than actually trading a system based on TA alone

I don't believe the market is totally random - just that it often is random. There is enough regularity to cream a profit.

The main point being that a major 'random' event will wipe out those traders who do not allow for the 'impossible' to happen. Most don't believe it when it happens and even add to their positions in the belief that it 'must' turn around.

 

[£10kLoser]

So how much bad does a successful system have to get before you determine that it's not working anymore?

Referring to your account equity on an account just trading that system, would you abandon it after it gives back 25% of its profit? 50%?

Or do you have some other measure? Or is it a trade secret?

I call a system in need of serious revision if it loses its profitable edge for a month - if we are making, say, three trades a week. There are no strict rules. It just becomes apparent that the market is behaving differently and not allowing you your profit any more!

For example, it's obvious that gold and oil are not behaving as they did a few months ago and the EUR/USD has certainly changed as well. Systems that pulled in profits then would be struggling now. It's not just they are no longer bull markets because systems should show sell trades which made a profit back then, too. The behaviour is just different!

 

[pmn100]

i agree and i have a divergence pattern that does in a range condition repeat itself quite often, also it amazes me how trendlines act as s and r, and major s and r definately work imo.............But TA is still random and best used as a secondary tool imo......It can work alone but has to be played in probabilities with r/r and money management..........I think the pro's use levell 2 or fundamentals alongside TA and this is the way i wish to go now as i need further skills to get a better edge.

This is similar to the conclusion I have drawn. I'm starting to view indicators as secondary tools for confirming a trade. To use them alone to build a successful mechanical system you'd have to play probablilities and enter every trade knowing that you HAVE to take everyone of them to catch the winners, and run them, but probably take more losers, with potentially large drawdowns.

A more discretionary approach would be to monitor s/r levels, monitor pattern formations, and trade the breakouts/reversals, whilst using indicators as confirmation (divergence, etc). BUt to do this surely you'd need to follow few instruments. Following few instruments also gives you the time to be able to follow fundamentals, which you'd likely find difficult with a range of different instruments, on different markets. If you was to specilise in a handful of instruments with this approach, surely you'd have the advantage over a mechanical system trader who jumped in on the trade because the system flagged it up as a trading opportunity? As someone specilising in the stock, you'd have the experience to know what 'usually' happens with price action at this point.

What are people's views on this? What is your preference? Do you work a strategy (either discretionary or a mixture of discretionary/mechanical) that utilises only a handful of instruments, where you have a strong understanding of the way the instruments move? Or do you work a wholly or mainly mechanical system that flags up trades across a whole host of instruments where it is not necessary to know how any of them have behaved previous?

 

[TWI]

I have generic systems and some market specific. The generic ones tend to work across a variety of markets but certainly not all of them and also it is only the portfolio of markets that creates a good curve, any, on its own, is less than satisfactory. Personally I like to optimise and providing it is done in the correct way, that is, looking for stability rather than profit then it is useful.
market specific have shown themselves in the past to be less robust mainly due to lack of diversification but on fx I still run cross specific systems that I developed years ago and they still make a return. I know of one multi bil algo fund that uses a combnation of generic and highly market specific so i do not think their is really any right or wrong way to approach this. Once you go live you soon find out which are right and which are the wrong ones.

 

[£10kLoser]

What are people's views on this? What is your preference? Do you work a strategy (either discretionary or a mixture of discretionary/mechanical) that utilises only a handful of instruments, where you have a strong understanding of the way the instruments move? Or do you work a wholly or mainly mechanical system that flags up trades across a whole host of instruments where it is not necessary to know how any of them have behaved previous?

I only trade any system on instruments I'm confident about and have experience of how they move. You are dead right about the problem with discretionary trades. Ideally, we may need to exercise discretion. However, if you have a system that does really show a profit and you start second guessing (using discretion) that can be a recipe for disaster as the emotions kick in.

I try to trade every single signal from a mechanical system and I try not to close early!