Automation and backtesting

The market changes direction, from monetary policy, to world events, NFP or just the market changes directions or the your EA doesn't recognize the change in direction.

So all backtestings are doomed to fail then?

Show me a consistent profitable EA, and Ill give you a million bucks

You mean automation fails always?

Yes Almost always its doom to fail unless you know what you are trying to do. i.e Those who understand how markets works [which i believe very few of them knows it] knows that 99.9 percent of trading systems and whatever expert advisors or robots will fail in the long run.

Again, most traders use back-testing for the wrong reason. I agree with Hdhorda4, if you trade based on back-testing results ONLY, you are doing yourself a disservice and you must know how the markets work in general. You can not rely on back-testing other than to see if your idea is viable. Only forward testing in live market conditions can you have a proof of concept.

Trading is a business and you must understand how the business works before trying your hand at it. Look at it like this. Back-testing is when your product is in Alpha stage. During this stage you see what SEEMS to work and what does not. Once you have something that looks like it works based on what it is you are trying to do, you need to test it in the real world. Beta stage is that world. Forward test on real time demo to work out any kinks not shown or experienced in Alpha testing.

The key to understand the difference between Demo and live is in demo, its not real money. The things you try in demo do not always carry over to live. There is a psychological barrier that needs to be addressed when you go live. So be prepared for that.

In part 2 of the Beta test, you will need to go live in a very small account, look at it as a focus group. Pick 1 or 2 pairs, the smallest amount of real money and the smallest amount of lots. If you had the results you wanted in Alpha and the demo Beta stage yet in live beta it does not work the same, your idea should be scraped and try again. It wont get any better.

- Ric

Why "understanding the market (or similar concepts)" cannot be programmed in the backtesting?

New analysis, this time adding a comparison of yesterday (for example, buy yesterday and buy today, sell yesterday and sell today, buy or sell yesterday and the opposite today)


Y B T B 0.1139
Y S T S 0.0521
Y B T S || Y S T B 0.0521
ALL 0.1481

Again same problem, random periods give random (negative) results.

Why?

Why pro traders say test 100 trades to see if a system has positive expectancy? But when it has, it fails for 200 trades or 1000 trades?

Does it matter if 100 trades are from X to Y or from S to T? Seems yes

PIPS (blue) vs 1% Risk (red) 10000 acc

1% accumulated (blue) , pip accumulated (red)

My understanding is backtesting period means nothing! Because you can extend or shorten the period and you get completely opposite results.


So the line graph above, the period could have from peak 125000 to 9000 so you end up with 10% of your account.

Having saying that if we extend 15 years of backtest to 20 or 30 years the up move (positive expectancy) in 15 years might be a correction in a bigger down move (negative expectancy).


Do you agree?

In other words

No, they are not. It just shows that your strategy/method has some flaws. Most probably the flaws are built within the very rules you created.

Example: when I started trading back in 2010, I was crazy enough to use a pure Martingale system. At some point I risked $10k in order to earn anything between $5 and $500 in a trade. It worked for almost one year. What happened? Huge volatility in August 2011 killed my account. Since no one can predict the market volatility, you are doomed to kill your account with Martingale, sooner or later. Would any variation of Martingale survive the SNB crash in Jan 2015 and continue to give a good profit, for example?

My advice: from all those backtests, take some of the unprofitable ones and analyze them, find the reason for the failure. Most probably that will be the reason for all the other losses. Then check if your rules can be fixed and/or adapted.

You have to prove to yourself, through the backtest, that you have an edge. Without any edge, how can you feel psychologically secure to enter a series of trades* ?

* series of trades: that's a terminology used by Mark Douglas, the author of "Trading in the Zone". If you don't know him, take a look on Youtube - you will find some of his seminars on trading psychology. By the way, he already passed away...