How to optimize an EA to success? Is it possible?

[dvc11]

Hi traders,


I read posts on WWI and around the net and came across 'Optimize' 'Optimize' 'Optimize' ... What is EA Optimization exactly?

Any advise, share experience, or opinion on how to optimize an EA to success in EA Forex Trading?

Or perhaps....is it possible to optimize EA to successful EA Forex Trading?

[kenboi2]

Optimized setting to me means nothing but just the best our of an EA. Most of the EAs lose money eventhough you optimize it. Optimization only means the most potential in profit with the least risk for a particular EA. If you think an EA has a potential, then you will spend at least 3-6 months testing it to get the best setting out of it.

So far, I don't see any good EAs on here. Sorry to say that. I hope I make some sense here

[sonik]

Optimized setting to me means nothing but just the best our of an EA. Most of the EAs lose money eventhough you optimize it. Optimization only means the most potential in profit with the least risk for a particular EA. If you think an EA has a potential, then you will spend at least 3-6 months testing it to get the best setting out of it.

So far, I don't see any good EAs on here. Sorry to say that. I hope I make some sense here

Not exactly.. EA Optimization is done in the backtester.. You select variables whose changes have strong correlations to changes in profit/loss, drawdown, etc. An optimization backtest will run a backtest for every possible combination of those variables and rank them according to how successful they are.

I.e. An EA that is mildly successful might benefit from different profit target/stop losses. Let's say that it has a Stop Loss of 10 and a Take Profit of 10.. maybe those settings were created for market conditions 2 years ago, but perhaps the market is more volatile now so it could perform stronger with different targets. An example optimization would be TPs between 5 and 20, and SLs between 5 and 20 at 1 pip intervals. This would result in 225 backtests ((5-20)*(5-20) = 15*15 = 225). If you optimize on an entire data set you are "curve fitting," which means that you are just making the best variable inputs for that data set. The most optimal way to do this is to backtest on one set of data (let's say a few months in 2011), then use the optimized settings on a completely new set of data (a few recent months, but different from first months - no overlap). This is called "out of sample" optimization.

Consider it this way; we all know that past prices don't predict future prices. However, if we're optimizing based on price data we already know, we can make any EA look good - the true test is if that EA will work the same on random prices that we have yet to encounter. So we simulate forward testing by optimizing on old data and backtesting on different data (new prices the EA has never seen)..

I hope this makes sense.

[Fmfx]

So far, I don't see any good EAs on here. Sorry to say that. I hope I make some sense here

You take an EA and you make it profitable/Good, and perhaps you have one since you can't see Anyone here?

Since you know what a good ea looks like it means you have seen one before somewhere else, so pls point to us the place mate.

Regards

[dvc11]

Is it possible to optimize an EA from a losing zigzag streak to an consistent uphill profitable results like the attached pic?

[Fmfx]

Is it possible to optimize an EA from a losing zigzag streak to an consistent uphill profitable results like the attached pic?

To do this you will have to know a little of MQL4, and also understand the strategy behind the bot.

I think it'$ high time i start learning mql4, perhaps opening a threat about that.

Regards

[dvc11]

To do this you will have to know a little of MQL4, and also understand the strategy behind the bot.

I think it'$ high time i start learning mql4, perhaps opening a threat about that.

Regards

Thanks for sharing FMfx. Hmm...what I'm also wondering is even if one can optimize the results as such, can it last a long period of time as we can't predict the future price?

[Richo]

When optimising you should also only do this over a certain period of your data then run these settings in and out of sample time, so lets say you have 10 years of data, optimise over say 8 years then run those settings over the last 2 years and see if that is profitable. It has to be profitable in those last 2 years to be at least in with a chance.

[Fmfx]

Thanks for sharing FMfx. Hmm...what I'm also wondering is even if one can optimize the results as such, can it last a long period of time as we can't predict the future price?

It depends on the strategy, it will last for ever if the strategy has to do with Currency Strenght, i.e buying the strong and selling the weak.

Regards

[dvc11]

Thanks for sharing Richo and Fmfx, I deeply appreciate.

It depends on the strategy, it will last for ever if †ђξ strategy has to do with Currency Strenght, i.e buying the strong and selling the wick.

Regards

Fmfx, any recommendations or advice on EA whose strategy is based on Currency Strength? Thanks alot!