My story as a beginner forex trader - Is it even possible to make a steady profit with EA ? - page 2

 
Skrulle:

So far, every backtesting session using free EAs and demo bots has ended with my demo account blowing up or with a loss. I'm starting to wonder if something is wrong with my backtesting (or backtest setups) or if this is just a scam.

My knowledge level is still very low when it comes to backtesting, and I don't know how to use the optimization mode to create a better EA setup.

It's probably a combination of scams and needing to learn better backtesting.

Anyone who knows what they're doing can make an EA look great in backtesting, and some unscrupulous Market sellers do this. It's called "overfitting", and it's when you optimize all the variables to best fit the historical data. Even when you make your own EAs, and optimize, and think "Wow, I could have 10x'd my money in 3 months!"

There are several different ways to avoid this:

  1. Start with something simple. Simple strategies are harder to overfit. I'll give you a simple example that works great on indices and blue-chip stocks, pretty well on gold and oil:
    - Buy when the daily close hits a 5-day low
    - Sell when the daily close is higher than the previous day's high
    That's it. Simple rules. And it's worked for decades on the major indices. You can optimize it a bit based on the # of days — some work better at 4, even 3, others at 6 or 7. But it doesn't break.
  2. If you have something with more range of variables, make sure it's profitable over a range. For example, if your strategy is based on the 20 MA crossing the 200 MA, then it should still work with everything from 16-24 crossing anything from 180-220. Maybe 20/200 is the sweet spot. Or maybe 18/204. And that's OK to use, but only if the range is still profitable. If it's a very narrow niche that's profitable, the strategy is overfit.
  3. Test on out-of-sample data. This is pretty easy in MQL5 - just use the forward testing option. It will optimize on the date range up to the forward testing date, then test all the results from the earlier dates on the most recent dates. You want the strategies that performed reasonably well in both, which is probably not the one that performed the best in either time period. Now, the forward testing is a good start, but I like to also test in other date ranges, like, say, how does your strategy handle a bull run, a bear run, a long sideways market, a market crisis, etc.?
  4. Robust EAs (and trading strategies in general) look at multiple timeframes and at least some basic idea of "market regime". That's been one of the biggest lessons I've learned. And most MQL5 Market and CodeBase EAs don't have much in that regard. They're basically a proof of one concept that works when it works, but doesn't when it doesn't.
Hopefully that saves you a few weeks/months of painful first-hand learning the hard way.
 
Alp Duman #:

In my opinion, your chance of becoming a good trader after working for many years is lower than your chance of becoming a good doctor in the same period of time.

Some may object ... but the FX market has a lot features in common with a gambling table ... be a scalper, day trader, swing or trend trader ... it doesn't matter at all.

As a trader ... in FX or in any other market, from a technical point of view, all you can do is to look for price patterns ... and pray that past realizations will occur in the future.

But it never happens like that, not that simple.

As someone who has been actively trading in various markets for many years, my advice to you is to become an investor in the stock market.

This way, you're more likely to find yourself making more money in the long run.

... and you can be a good doctor in the remaining time.

Alp, 

Thank you for your advice. 
I am already an investor in the stock market with above average results the last 10 years. 
I have a dividend portifolio that returns a nice income yearly. 
However, I dont spend much time to study the stock market anymore since my experience is that "time in market" beats "timing the market". 

I think forex trading is like a breath of fresh air compared to the stock market and want to investigate the possibilites further. 

So far I have just invested a minimum amount of money to test and learn, but I hope that someday this creates another source of passive income.

Cheers, 
Skrulle

 

The simple answer is yes, it is possible. A bit harder for those of us in the European trading jurisdiction where it's not "trade with any money you like, even if you don't have the money !"

But as it goes, trading is simple, but not easy... the same with programming of EAs, simple products, but with a lot of potential caveats in the usage AND creation of them. Fernando put it very well in his post.