Hey, hey - we're here - just too busy trading & coding :)
This is not a new discovery but it is a timely reminder to all that back-testing and demo-trading develop optimistic results on most EA's
Live feeds also vary considerably between brokers, so an EA that is good on one may be useless on another
Recent changes on many brokers to be more straight-through has altered their feeds sufficiently to need a change in some EA tactics...
My impression is that MT5 is supposed to be improved for back-testing over MT4
-BB-
No idea what a Polish stock exchange index should look like, but this doesn't look like a case of different conditions (ie chart) between Demo and Live.
Both charts look rather similar (to me) but the trigger points are vastly different. Sure it isn't the EA doing something different between the 2 accounts?
No idea what a Polish stock exchange index should look like, but this doesn't look like a case of different conditions (ie chart) between Demo and Live.
Both charts look rather similar (to me) but the trigger points are vastly different. Sure it isn't the EA doing something different between the 2 accounts?
Actually the charts should be identical, they are not but I'm not discussing that at the moment, they are close enough for my system not to notice it.
The chart shows that behavior of the same system varies whether it's a DEMO or REAL account. The day closed somewhere around 2370, the SL was set to 2381 on both (SELL order). The next day in the morning, session started at 2421 and DEMO account had an order close exactly at the SL price, REAL account had the order close at the Open price. The problem is that I had the DEMO running for 3 months before I finally decided that it's good enough for real trading. Only when I started trading live, this problem arose.
Hey, hey - we're here - just too busy trading & coding :)
This is not a new discovery but it is a timely reminder to all that back-testing and demo-trading develop optimistic results on most EA's
Live feeds also vary considerably between brokers, so an EA that is good on one may be useless on another
Recent changes on many brokers to be more straight-through has altered their feeds sufficiently to need a change in some EA tactics...
My impression is that MT5 is supposed to be improved for back-testing over MT4
-BB-
Are you saying that this is up to the broker to fix that problem and others may have this worked out well?
Are you saying that this is up to the broker to fix that problem and others may have this worked out well?
The brokers dont see this as a problem !
Others have worked out how to handle this, in various ways
Some really clever chaps have added in-line spread/spike variation to their backtest data
Others (including myself) have projections to estimate performance on live and... we make much use of micro live accounts...
FWIW
-BB-
PLS TELL ME WHICH CODE U USE FOR BUY & SELL ?
THANKS
![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
Here is the thing.
I have 2 systems running at the same time. One of them is LIVE DEMO Account (right picture) with fake money and the other LIVE REAL Account (left picture) with real money. Take a look at today morning. It is a chart of Polish stock exchange index WIG20. Both had order open with a stop loss at the 2381 and Buy Stop order in the same place. Demo closed SELL and opened BUY exactly at the point of SL/BuyStop, Real account opened it at the stock open price.
Now my comment. There is no point in testing/optimizing the system in MT4 optimizer for any symbol where open/close gap is possible because the results will completely differ from the real results. I had my system running for 3 months on demo live account giving me really good profit (around 250%). After I started it on the Real account the equity is 10% less. In the same time Demo equity rose 30%.
Of course... there are some workarounds to make the results more realistic (virtual SL/TP control) but shouldn't the tester be a little bit smarter?
Comments welcome!!!