Shocking discovery!
I know that Live and Demo accounts from the same Broker can differ in a number of ways, Spread, Commission . . . it wouldn't surprise me to find that there can be slight differences in the data also.
Testing in the Strategy Tester vs Live or Demo is different because the ST does not simulate an ECN Broker nor does it simulate slippage or delays in placing or closing orders.
It is often said that all that is needed in Forex to b successful is a small edge, I don't believe this to be true . . . maybe if you have fixed very small spreads, but that is not the reality of retail Forex trading. A small edge can soon be negated by a change in spread . . .
Being successful in Forex is one of, if not the, hardest things you will ever do . . . . the vast majority of people that try will fail.
Not shocking at all.
1) Is commission simulated on the tester?
2) Compare the Spread
3) How different is the datafeed. If its only a few pips then everything is normal. Demofeed thends to be more laggy/smoothed
4) Your EA seems to be really spread, / tick data sensitive.
5) Do not blame the broker for something you have not calculated.
6) Tester does not even simulate Gaps, Spikes, Requotes, Slippage
99.999% of the EA's performance will be: Backtest > Demo > Live.
Without more details on your takeprofit/stoploss timeframe i cannot say more
I agree with both RaptorUK and Zuegg.
To expand on the Backtest>Demo>Live, I trade a 100K lot standard account on a VPS. I don't like to do any work on a VPS other then setup my EA's and charts, so I set up duplicate charts and EA's on a local demo acount. I figured I could just check my laptop and know what is going on the VPS, wrong! The results from the first day I started this setup were not even close. I've also noticed differences between the number of trades in a backtest and number of trades in demo and live accounts. I trade on newbars so I thought it would be pretty close.
So I setup a 1K account on my VPS as well. So I do backtest > 1K live > 100K live. I tried a demo on my VPS and got the same type of differences with the demo on my local machine. So I use the micro account to do real testing before moving anything to the larger account. I would rather lose $9.00 then $900.00 on a bad trade or bug in my EA. I'll take the small loss/profit to test the EA in a real environment. Of course this only works if your broker offers a 1k account and you are trading a larger account. I also have the trading desk on speed dial and the link to open a dispute, bookmarked. Over the last few years I've had to submit 6 disputed trades, my broker has resolved 5 of them in my favor within 48 hours.
On the topic of slippage, it can be a killer to profits if your takeprofits are for small pips or your distance between stacks of pending orders is tight. I will only take trades with potenial of >20 pips. A slippage of 2 pips on that is still a 10% hit.
This is so demotivating! It is much harder to be profitable in Forex trading than I ever imagined.
By the way my Profit Take targets are averaging about 15 pips, and furthermore I place pending orders so slippage should not be a problem for me.
But my EA is very sensitive to ticks/price when placing a trade.
Danjp seems to have found the best solution to this conundrum. But what a story to have a chance to win!
Thanks guys for your input and bursting my bubble!
Be carefull pending orders can slip. Happened frequently enough to me that I tweaked my TP and the way I stacked orders to try and account for it.
There is one thing that is still unexplained to me.
I have said in my first post that the price data was IDENTICAL between the Live and the Demo account form 4 Jan - 23 Feb.
Then suddenly the data starts to differ after that.
How can this happen?
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
I use FXOpen as my broker and have had some strange results from my EA this past week. To cut a long story short I discovered that trading on
gave THREE different results!
When I checked the historical data for the period 4 Jan 2012 to 23 March 2012 I discovered that the data between the two ECN accounts are identical from 4 Jan - 23 Feb 11h40, BUT from 23 Feb 11h41 onwards the data starts to be different between the two accounts.
When I asked FXOpen about this they seem to think that there is nothing wrong with this.
This brings me to this conclusion:
Brokers van manipulate data any way they want to. So if you do all your testing of your system or EA on a demo account, back and forward testing and find it to be very profitable - as soon as you start using this same EA with the same broker on the same type of account you can lose all your money.
Even if you do all your testing on the Live account data the same will be true.
SO HOW CAN WE EVER WIN? Except this is a very robust EA with long term transactions you are bound to lose!
Has anyone had the same experience as me? And what can be done about it?