Why “backtesting” is holding you back


After reading your initial post as well as your subsequent ones, I do agree with the majority of what you have said. For manual traders, I have always advocated to skip trading on a demo account and start with a live account right from the start (with a small amount, of course). Even a small live account will allow a trader to feel the emotions, and start working on following his or her trading plan and discipline. At the same time, manually (by hand) backtesting a trading strategy is very prone to personal bias and look-ahead bias.

With that said, I think you could have phrased your opinion a bit more clearly. You seem to have used the terms “discretionary trading” and “technical strategies” interchangeably; they are more opposites if anything. A “purely technical strategy” as you stated, usually refers to a strategy that is 100% mechanical, as in algorithmic. By that logic, it wouldn’t make any sense to backtest such a strategy by hand, rather an automated backtester would be a better solution. On the other hand, a discretionary trading strategy may use technical analysis, but would also require the subjective opinions of the trader.

Every point you have addressed applies only to discretionary trading and not so much algorithmic/technical/mechanical trading. I think this should be made clear so that new traders reading this thread do not get confused or form bad habits.

I totally disagree. The one and only thing that gave me the confidence to trade live was manual back testing.

As a trader starts his trading what is more important for him is that he should be having a good trading knowledge and there will be a confidence that will back up his such trades …

[B]My contention is quite simple MARKET REPLAY used in correct way a very important tool for a trader’s development.
*In my previous post you find the points why I think Market Replay is important.



Interesting theme: it is well under category (EMOTION), which is the main reason that most falling away when account approach to ZERO

Personally, I believe that one should value and boasts of performance after the size of the account and the consequences of loss

[B]How much emotion is now really present when the account is between 1 to 10 K.[/B]

What are the consequences of loss !!!

Under 1 K is demo account emotion .

[B]That is why I am so impressed by Fxviper 7 million accounts … by loss the consequences are enormous for the man …
Example :

Take your results … based on 1-10 k account, okay result … world is not going under at a loss of acount .

50K account then we start to speak, good result.

*100-200 k account very good result …im impress

500-1000 K extremely good result , Im very impress ,you have balls of steel.

*Example one trader I know aggressive trading style, all over 50K in the account taken consistently out…
Then he controll his emotion …

Normally he has results that you refer to year after year.

Another I know have his limit at 70 K after there is the emotional burden too heavy .

I agree that manual becktesting (i.e scrolling back on the chart and manually test you strategy) has limited use - maybe just to get familiar of entries and exit procedures.
It’s tedious, prone to mistakes and ‘white lies’ you can only test one pair at the time and over a very small sample.
If you want to backtest properly you need to automate it using a proper charting package that allows you back test at portfolio level over a lot of trades.
I did hundreds of them using Amibroker and data going back years, over 10-20 pairs at a time. What I found is the the vast majority of purely mechanical systems don’t work in FX. Those that do are pretty lame. That’s what stopped me form trading fx until now. I’m still very tentative at it. It tells me that FX needs a significant level of discretion in trading to make it profitable. To that extend backtesting IS holding you back.
The good part of backtesting is that you can see very quickly what works and what doesn’t and when you find something that does, you can optimize it at a click. for example - in babypips article on how to build a mechanical system one example is using ma5/ma10 crossover in conjunction with RSI and stochastic. On back test that system works on the long side but not on the short side. winners are only 36%. Running optimization it turns out you can improve it if you use a ma5/ma12 crossover instead. it’s very odd but it seems common that some systems work on the long side but not on short and the other way around. I didn’t expect that in FX as there isn’t any bias like in the stock market.
Anyway, that’s my experience. If you want to do it you need to brush up you computer code skills as they are needed to covert your system into something the software can execute.

GIGO - Garbage in Garbage out.

It means that if you cant back test correctly then your results will not correctly reflect the system in question. People who trade technical systems, but disagree with back-testing simply have an issue and fail to think logically.

Back testing does not lie, it does not reflect unfair results, it does not tell white lies, it is not different because of your emotions, it is not different data, it is not different because real money is not on the line…and the list of myths continue.

The only difference is you as a human being. Fix your own emotions and proving you have back-tested a LARGE ENOUGH sample size to give you a LARGE DEGREE OF CONFIDENCE and your results will be as accurate as live trading.