My results 2003-2018 with starting 3K
Macroeconomic conditions can change so much that the market behaves distinctly from one period to another. Market sometimes have periods of high activity and periods of low activity. And there are times when system predictiability of maket direction is just absent, i.e. all hands are off when market is uncertain, hence the low performance. I do observe that there are about 3 times in a year that everyone will be having losses on a single day, and none are winners. Then of course, the black swan events that throw you into the deep end of pool with heavy losses.
Backtesting gives you the false illusion of confidence. You only need to backtest how your trading system behaves when it meets certain conditions, and whether it does exactly what you intend it to. i.e. Has it traded properly in the specific type of market structure that you are looking into? Can it handle spread changes during different times of days when liquidity is low?
If you optimise too much, it becomes overfitting to the old periods. New periods are ever changing.
There will be a time when you realise your trading strategy will be broken because it can no longer handle new situations. And you probably will trash it.
What is important is whether your profit generation can cover all the stop losses. Is this statistically invariant to all market conditions? How many market conditions have your trading system addressed? Is it robust?
I think it is better to do more forward testing instead, understand how the market behaves, and tweak accordingly.
It is not really about parameter optimisation. It is something else.
I understand, your reply is very profound and makes a lot of sense.
My strategy actually never loses a single move, in the 'bad' years it just earnt little as not many opportunities came for a trade according to my parameters, i set 4 years ago on a mission to find out a strategy that does not have losses. over 13 years of backtest - no loss.
the question now is how to make it take larger profits etc..
![MQL5 - Language of trade strategies built-in the MetaTrader 5 client terminal](https://c.mql5.com/i/registerlandings/logo-2.png)
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Hello everyone!
I have developed and algorithm and actually it's a 2 part question:
1. when testing it using real tick data i get some setups give me better profits 2003-2009 and some give me better profits in recent years - what should be my focus when optimizing? is it correct to assume that the market is more likely to behave as it has in the recent years or consider 2003 onwards as possible behaviour?
2. what is the FASTEST way to optimize an algorithm, considerint its written for MT4 and not MT5 so no multi-thread.. perhaps VPS or are there companies that would do it?