strategy to automate?

[wskanaan]

Hi everyone....

I just started with mql4 and got the hang of it...

I would like to implement a good strategy... which strategy would you advise? I tried some strategies like using the moving averages and rsi and williams trend... but backtesting did not produce promising results...

 

[£10kLoser]

Ha! You think you can automate a winning strategy and make pots of money? Dream on...

It's a common delusion that backtesting will find a winning method but why should the market behave tomorrow like it did yesterday? If you want to automate a system. try random buy or sell...

After all, you'll need lots of luck. Good traders don't try and predict where the market is going, they look at it and trade accordingly.

 

[wskanaan]

Ha! You think you can automate a winning strategy and make pots of money? Dream on...

It's a common delusion that backtesting will find a winning method but why should the market behave tomorrow like it did yesterday? If you want to automate a system. try random buy or sell...

After all, you'll need lots of luck. Good traders don't try and predict where the market is going, they look at it and trade accordingly.

I learned this the hard way... however... there are strategies that will increase the probability of your winning... therefore.... if you use proper money management with an average loss to profit ratio you should make some money...

and anyway... if you have a strategy that you follow that can be automated... why sit at the screen waiting for the exact right time to enter when the system can do it more accurately?

 

[Mathemagician]

Ha! You think you can automate a winning strategy and make pots of money? Dream on...

It's a common delusion that backtesting will find a winning method but why should the market behave tomorrow like it did yesterday? If you want to automate a system. try random buy or sell...

After all, you'll need lots of luck. Good traders don't try and predict where the market is going, they look at it and trade accordingly.

Nonsense! Many people have been making a living trading automated systems for quite some time, myself included. Remember, just because you can't do it doesn't mean it's impossible!

jj

 

[£10kLoser]

Yes, the advantage with automation is that your emotions can't interfere and ruin the trades. However, 'increasing the probability of winning' is a far cry from 'certainty of winning'. Nobody disputes that around 95% of traders lose money but most people think that doesn't apply to them - as shown by this very forum!

The role of luck in trading is much underestimated. When we lose, it's bad luck and when we win it's because we're a good trader...

 

[£10kLoser]

Nonsense! Many people have been making a living trading automated systems for quite some time, myself included. Remember, just because you can't do it doesn't mean it's impossible!

jj

Nothing to do with the fact that I can't do it. If what you say is true, how come I've never seen one bit of evidence anywhere to support that statement. After all, there would be no need to keep it secret and if other people used the exact same system it would more likely increase profits rather than decrease them.

It is well known that the vast number of automated trading systems cause sudden rises and falls at trigger points but that doesn't mean that consistent profits are made. I know that some institutional traders automate their systems and I never said that it was impossible to make a profit as such. However, many institutional trading companies lose vast sums of money and only make a profit from fees charged to clients.

So, where's the evidence, other than in people's imagination?

 

[Mathemagician]

Perhaps you'll have a better exeperience with this simple change of perspective:

Instead of approaching backtesting as a way to find strategies that are guaranteed to be successful in the future, why not approach it as a way of weeding out strategies that are unlikely to work in the future?

The more bad ideas you can weed out, the better your odds of actually deploying a good idea. To many of us, this makes a lot more sense than relying on some mystical, magical "talent", though it is admittedly less appealing to those who prefer feeling special to making money.

If try as hard as you can to prove that a strategy will not succeed and fail, is this not a good candidate for live trading? Not every system has to be successful in order for you to succeed as an automated system trader. You simply need to, on balance, produce more winning models than losing models and apply appropriate risk management.

jj

 

[Mathemagician]

Nothing to do with the fact that I can't do it. If what you say is true, how come I've never seen one bit of evidence anywhere to support that statement. After all, there would be no need to keep it secret and if other people used the exact same system it would more likely increase profits rather than decrease them.

It is well known that the vast number of automated trading systems cause sudden rises and falls at trigger points but that doesn't mean that consistent profits are made. I know that some institutional traders automate their systems and I never said that it was impossible to make a profit as such. However, many institutional trading companies lose vast sums of money and only make a profit from fees charged to clients.

So, where's the evidence, other than in people's imagination?

You can't be serious.

jj

 

[£10kLoser]

Perhaps you'll have a better exeperience with this simple change of perspective:

Instead of approaching backtesting as a way to find strategies that are guaranteed to be successful in the future, why not approach it as a way of weeding out strategies that are unlikely to work in the future?

This is exactly the same thing.

The more bad ideas you can weed out, the better your odds of actually deploying a good idea. To many of us, this makes a lot more sense than relying on some mystical, magical "talent", though it is admittedly less appealing to those who prefer feeling special to making money.

You are still talking about 'better odds' not certainty. Nobody's going to feel 'talented' or 'special' if they are losing money.

If try as hard as you can to prove that a strategy will not succeed and fail, is this not a good candidate for live trading? Not every system has to be successful in order for you to succeed as an automated system trader. You simply need to, on balance, produce more winning models than losing models and apply appropriate risk management.

As the quote required by law goes, 'Past results do not indicate future success.'

So, where is the evidence of profitable EA trading? Pipboxer comes close but even that makes losses on some pairs some times - often huge losses. Nobody has ever shown any proof of guaranteed profits because it cannot be guaranteed and that's why EAs need hype to sell them.

And silly Mcenroe remarks won't convince anyone.

 

[Mathemagician]

You sure you didn't mean to log into EliteTrader?

jj

 

[wskanaan]

All I asked for was a strategy.... sorry to have started this debate...

Anyway.. I have found some possibilities and I will be testing them tomorrow.....

As for the idea of forex being a constant loss... keep in mind that when you lose someone else wins... the biggest winner is the one that maintains a higher average gains than losses... I am still to become one... but I'm new to this....

 

[trendie]

Hi everyone....

I just started with mql4 and got the hang of it...

I would like to implement a good strategy... which strategy would you advise? I tried some strategies like using the moving averages and rsi and williams trend... but backtesting did not produce promising results...

what do you mean by "promising results"?

have you tested your rules on different time-frames?

are you getting good trades, but the profits are eaten away by lots of small losses?
in which case, add some kind of volatility filter.

is your system trading 24hrs a day?
are the losses spread evenly across the day, or can you determine that a large number of losing trades happen either
a: during quiet periods like between 2-am and 7-am (in which case, try a filter to trade between volatile times)
b: there are a sequence of losses on certain days. (what this means is, that day may have been choppy or non conducive to your system. try stopping trading for a day (or a session; such as if its a bad euro-session, stop trading until the US opens, etc)
c: large single losses on certains days. can you equate that with news events

the markets are not simple mathematical things. they have rhythms. there are quite basic and repeatable "sessions" of movement and times of "drift", which you need to test if your algorithm responds badly to, or does well in.

system development is great fun, and immeasurably frustrating.
my experience is to develop an alert when a set-up is "close-by", and then I can monitor the trade developing once alerted, and make visual assessments of trade viability.
ie, 80% computer-work, 20% human decision.

The very best of luck in developing a totally automated system.
Hope you succeed where I stopped.

PS: have you looked at Hoggums thread with his mechanical RSI tests?

 

[rnicoll]

Despite some people's skepticism; if you can see it, odds on it can be automated. That said, there's a reason people are drifting towards neural nets for this stuff, it's probably the fastest route to a working strategy.

That said, look for your strategy to be making _at least_ 10% or so a year over around 1,000 trades before you consider it useable. I've got strategies that perform that well, and I wouldn't trade, certainly (PM me if you want a copy, and promise not to hold me responsible if you try trading it). My best so far, on the other hand, manages 60-120% over a year on profitable currency pairs, and loses 10-20% on the (much smaller number of) pairs it loses money on.

A few things to watch out for:

1. If you're optimising parameters, you should expect them to trend towards a single "correct" value. If they don't, see if you can infer them from past data instead of brute forcing them by repeated testing. If you can't get them to a single value that's a good basis for everything, it's a warning sign.
2. Too many parameters makes overfit easy, but that doesn't make them bad. You just need to be aware of the risks.
3. Try everything. Don't write off an approach as "clearly that won't work". Maybe it'll surprise you, or at the least maybe you'll learn something.
4. Don't focus on one approach too early. Try lots, filter the best bits.
5. Don't be afraid to think different. The freaky stuff no-one else has tried is where there's real money to be made.

 

[counter_violent]

Automated trading....find the key components measure and weigh them accurately.

What are key components !!!! This should be your focus.

 

[Exorsus]

Nonsense! Many people have been making a living trading automated systems for quite some time, myself included. Remember, just because you can't do it doesn't mean it's impossible!

jj

Exactly, i must say. Everything that can human trade on FX ... same way the automated system can do. But experience and way of thinking mathers.