EAs and autotrading
Posted: Mon Jul 06, 2009 11:39 am
I have been coding/testing EAs and would like to share my thoughts... this is more of a list of 'notes to self' than anything else.
I have been working quite hard on this for the past few weeks, but I am by no means an expert. I have some observations to share based on a lot of testing and a week of very carefully monitored live trading. This forum seems most appropriate as autotrading may be perceived as the route to 'easy money' for the unwary.
My summary is that EAs can be profitable, but you will need 'proper' live trading experience in order to create and manage them. Also, (for me anyway) they are a useful addition to one's trading arsenal, but cannot replace manual trading...
Please comment if you have something to add...
I'll add more in a couple of weeks when I have more...
The catch(es)
1/ an EA will only be profitable under certain market conditions, don't expect it to be profitable all the time (or even the majority of time)
2/ if you can't understand the strategy behind an EA then don't trade it, otherwise you will not know why it's started to drain your account
3/ never, ever, ever simply set an EA running and let it do its thing trading live on real money - it WILL drain your account dry as market conditions become unfavourable
4/ if you don't know the code in the EA, then don't trade it
What they are good for
1/ an educational tool for backtesting indicators and strategies - to see where and why they fail/win
2/ small 'extra' profit in addition to profits from your 'manual' trading
3/ precision trading without emotion OR judgement (other than rules written into the code) - i.e. would my strategy have worked today if I hadn't been undisciplined (yet again)?
4/ collection of hard data to support/reject theories
From my experience, for an EA to be profitable you need to have VERY VERY tight trade entry criteria. So tight in fact, that you'll be lucky to see more than two trades a week. Maybe others do better with their EAs... (and maybe they have more live trading experience than me anyway, and/or understand the market better)...
...and a profitable EA takes a lot of time and effort to create and test. I have only ONE profitable EA, which gives me less than 5% of total weekly profit. Is this still easy money?
If you want to trade with EAs
The allure is obvious: easy money for doing bugger all. The reality is that the 'effort to reward' ratio for EAs is much higher than for 'manual' trading, due to the work required to develop/run the EA and the relatively small returns they generate (compared to 'manual' trading).
You could take a shortcut and trade with someone else's EA (bought, pilfered or free), but unless you understand it (code and strategy) then it WILL lose you money.
An EA is no substitute for live trading experience - an EA is a mechanised expression of your trading strategy, if the strategy sucks then so will the EA. You will also need this live experience in order to properly manage your EAs - i.e. when/why to unleash them. Is this still easy money?
If you are really serious about exploring the potential of EAs (I am), and assuming you're using a proprietry trading platform (I am), then you will need to learn how to code.
ARE you that serious? Learning coding is not THAT difficult - it just takes time and effort. Is this still easy money? Would your time be better invested getting live trading experience, or better understanding your current strategy?
Testing with EAs
The fun part. I love watching this action replay - cheering the highs and booing the lows. Cursing as the *%&^$ thing refuses to trade, or trades like a demented bookie.
Setting the test to run overnight is no good, as you will miss the 'blow by blow' action, and never work out why it did what it did. You need to watch the screen as the action unfolds. I love the little graph I get at the end, but this is just a souvenier - the results don't mean much unless you know HOW they came about.
Anyways, that's enough musing for now. Glad to have something to contribute to this most excellent forum.
bazmo
I have been working quite hard on this for the past few weeks, but I am by no means an expert. I have some observations to share based on a lot of testing and a week of very carefully monitored live trading. This forum seems most appropriate as autotrading may be perceived as the route to 'easy money' for the unwary.
My summary is that EAs can be profitable, but you will need 'proper' live trading experience in order to create and manage them. Also, (for me anyway) they are a useful addition to one's trading arsenal, but cannot replace manual trading...
Please comment if you have something to add...
I'll add more in a couple of weeks when I have more...
The catch(es)
1/ an EA will only be profitable under certain market conditions, don't expect it to be profitable all the time (or even the majority of time)
2/ if you can't understand the strategy behind an EA then don't trade it, otherwise you will not know why it's started to drain your account
3/ never, ever, ever simply set an EA running and let it do its thing trading live on real money - it WILL drain your account dry as market conditions become unfavourable
4/ if you don't know the code in the EA, then don't trade it
What they are good for
1/ an educational tool for backtesting indicators and strategies - to see where and why they fail/win
2/ small 'extra' profit in addition to profits from your 'manual' trading
3/ precision trading without emotion OR judgement (other than rules written into the code) - i.e. would my strategy have worked today if I hadn't been undisciplined (yet again)?
4/ collection of hard data to support/reject theories
From my experience, for an EA to be profitable you need to have VERY VERY tight trade entry criteria. So tight in fact, that you'll be lucky to see more than two trades a week. Maybe others do better with their EAs... (and maybe they have more live trading experience than me anyway, and/or understand the market better)...
...and a profitable EA takes a lot of time and effort to create and test. I have only ONE profitable EA, which gives me less than 5% of total weekly profit. Is this still easy money?
If you want to trade with EAs
The allure is obvious: easy money for doing bugger all. The reality is that the 'effort to reward' ratio for EAs is much higher than for 'manual' trading, due to the work required to develop/run the EA and the relatively small returns they generate (compared to 'manual' trading).
You could take a shortcut and trade with someone else's EA (bought, pilfered or free), but unless you understand it (code and strategy) then it WILL lose you money.
An EA is no substitute for live trading experience - an EA is a mechanised expression of your trading strategy, if the strategy sucks then so will the EA. You will also need this live experience in order to properly manage your EAs - i.e. when/why to unleash them. Is this still easy money?
If you are really serious about exploring the potential of EAs (I am), and assuming you're using a proprietry trading platform (I am), then you will need to learn how to code.
ARE you that serious? Learning coding is not THAT difficult - it just takes time and effort. Is this still easy money? Would your time be better invested getting live trading experience, or better understanding your current strategy?
Testing with EAs
The fun part. I love watching this action replay - cheering the highs and booing the lows. Cursing as the *%&^$ thing refuses to trade, or trades like a demented bookie.
Setting the test to run overnight is no good, as you will miss the 'blow by blow' action, and never work out why it did what it did. You need to watch the screen as the action unfolds. I love the little graph I get at the end, but this is just a souvenier - the results don't mean much unless you know HOW they came about.
Anyways, that's enough musing for now. Glad to have something to contribute to this most excellent forum.
bazmo