Great EA in backtest! - page 124
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Thanks Wackena, but I think that there is a simpler way do have good backtest: 1. Download MT4 build 200 from any broker and at History Center, press the "Download" button;
BrazilianTrader
There have been alot of forum chat on the problem with this feature on build 200 platform Have you found a way to correct the data gaps in the MT4 downloaded tick data? If so, please adivse. This would make my life a lot easier not to have to manually download alpari data and use period_converter.
Wackena
BrazilianTrader
There have been alot of forum chat on the problem with this feature on build 200 platform Have you found a way to correct the data gaps in the MT4 downloaded tick data? If so, please adivse. This would make my life a lot easier not to have to manually download alpari data and use period_converter.
WackenaThat's a problem...
But there is also another thing to considerate on backtests:
Data feed from each broker and Server.
Alpari data feed is different than FXDD, IBFX, NF... etc.
There are other differences with the same data feed: The same broker have different servers for offline data feed, Demo accounts and Live accounts.
A backtest on Alpari offline data feed (which is hugely different than its Demo and Live accounts' feeds) will never have the same results that another backtest over the history data from FXDD or IBFX, for example. De difference between a Backtest over Alpari data feed and FXDD or IBFX Demo/Live accounts' performance could get even bigger. They could be a little similar perhaps, but not sufficiently reliable to me.
Alpari data feed could give an idea of Alpari Demo or Live, but definitely will not show what your EA will do on FXDD or IBFX (on Backtest, Demo or Live).
Alpari data feed is good, indeed, but considering that I am going to invest with FXDD, I prefer to test my EA on FXDD data feed with 90% modeling quality, and forward test on their Demo Server.
That's a problem...
But there is also another thing to considerate on backtests:
Data feed from each broker and Server.
Alpari data feed is different than FXDD, IBFX, NF... etc.
There are other differences with the same data feed: The same broker have different servers for offline data feed, Demo accounts and Live accounts.
A backtest on Alpari offline data feed (which is hugely different than its Demo and Live accounts' feeds) will never have the same results that another backtest over the history data from FXDD or IBFX, for example. De difference between a Backtest over Alpari data feed and FXDD or IBFX Demo/Live accounts' performance could get even bigger. They could be a little similar perhaps, but not sufficiently reliable to me.
Alpari data feed could give an idea of Alpari Demo or Live, but definitely will not show what your EA will do on FXDD or IBFX (on Backtest, Demo or Live).
Alpari data feed is good, indeed, but considering that I am going to invest with FXDD, I prefer to test my EA on FXDD data feed with 90% modeling quality, and forward test on their Demo Server.I totally agree that Demo & Live data feeds from most, if not all, brokers come from different servers. Although IBFX plans to have Live tick data feed for Demo accounts soon. How soon, last week they said a few months.
With the MT4 build 200, all backtest data comes from MetaQuote. That is the reason, that now all downloaded data from the History Center on build 200 with all MT4 brokers have data gaps. Hopefully, MetaQuote will correct this problem soon. But right now, the best source of data for backtesting in MT4 is from alpari. Unless you purchase live tick data , collect your own live tick data or download live tick data from a few websites and convert it to MT4 format, I'm not aware of any other MT4 formated databank on internet, other than alpari.
Wackena
I suggest you read the "great ea in back test" thread all the way through.
Well, it took a while, but that's exactly what I've been doing. Wow, what a lot of work you guys have been doing ! Amazing job. I've learnt a lot.
I have the new Build 200 of metatrader 4 now. I find that little squeaker noise the backtester makes amusing the first three or four times but now it's just plain annoying. I wish I could turn it off. Anyone know how?
In case you haven't already discovered it, I believe that if you go Tools/Options/Events , all the sounds are there. I think you should be able to change/delete them.
I'm a little behind you guys, but I've a couple of questions. Way, way back, FXDiva said he was getting great results with 185f. Is that still the case? I'm testing this one myself using a live demo with Alpari. Looks promising - I've just tweaked it, to exclude the news hours. With 1200+ posts, I may have missed something .. there's lots of discussions and tweaks, but is there actually a more recent "F" version?
One thing I'm just a little unclear on. When excluding trading hours, presumably, any open positions are closed. Is that necessarily a good thing? Might it be that a potential winning position actually loses by exiting prematurely? Also, is news actually a bad thing? I'm probably being naive here, but it could go either way. Some you would win - others you would lose. Won't it all even out in the end?
Anyway, I've excluded the news hours ... so I guess I'll find out for myself.
I spent quite some time on this system now and found some interesting points.
When Autostoploss is TRUE the system uses the spread calulation of the count bars to move the stop dynamic. In theory and least most of my test a dynamic stop should perform better then fix hard stops. This is becuase the stop should be a function of the volatilty in the market. The wilder the swings the bigger stop required. It stands to reason therefore you can NOT have a fixed 20 pip hard stop in volatile conditions. It will be hit every time.
Therefore i programmed the auto stop to include what i call stop bias and use the optimiser to find the best values and found a significant increase in performance by adjusting the control of the AutoStop feature.
Also i noted the Values period count and Value period count max plays a major part in system effectiveness. This value should be optimised between 1 and 30. Low values work better ie 3 to 7.
When a StopLoss values is either then 0 it will overide automatic stops and is less effective but its value can be controlled by the StopLoss Index.
The bad news is this system really suffers on both demo forward and demo testing the last few weeks. After much research the reason is becuse the daily ATR of euro is the worst or lowest its been since 2002!! Thats why live testing recently is showing flat/lossy performance while backtest earlier shows signicant gains. When volatitly dries up the system is unable to get the pull backs required to capture past the spreads. This effect is directly proportioned to the Daily ATR reaching its lowest through late August till now.
Now i worked out how to post charts this backtest runs all the way from Jan 2004 till today. The growth rate is mind blowing when everything is fully tweaked but you can see the recent right side of the chart is dropping. It dont look much but this is the last 2 months on low ATR is significant drawdown period.
One more thing before i go. This system makes use of everything it knows from Values Period Count. ie the last few bars which may only be the last 5 or 7 hours and includes information about any indicators used. Its ESSENTIAL the spread does not change to make an informed decision on the future.
In my opinion the system must not be used with any broker that varies the spread which rules out entirely using Interbank FX and FDD as both these brokers vary the spread.
There might be others but only Alpari (russian not UK) and NF that come to mind will keep the spreads fixed.
One more thing before i go. This system makes use of everything it knows from Values Period Count. ie the last few bars which may only be the last 5 or 7 hours and includes information about any indicators used. Its ESSENTIAL the spread does not change to make an informed decision on the future.
In my opinion the system must not be used with any broker that varies the spread which rules out entirely using Interbank FX and FDD as both these brokers vary the spread.
There might be others but only Alpari (russian not UK) and NF that come to mind will keep the spreads fixed.1. What does spread has to do with CT decisions? Until what I know, the spread only takes place when a trade is opened...
2. Which indicators does CT use? it has 5 variables on 3 possibilities (buy, sell, uncertainty) those are what define its logic;
3. Why the values period count tells it is in hours? I saw in the code it was in minutes, I may be wrong though;
4. Dave knows well how FXDD work, he never found a spread bigger than 3 for any longer than seconds, and I suppose it could effect not more than a brief noise on the final price to place orders (CT doesn't use TP);
Well, it took a while, but that's exactly what I've been doing. Wow, what a lot of work you guys have been doing ! Amazing job. I've learnt a lot.
In case you haven't already discovered it, I believe that if you go Tools/Options/Events , all the sounds are there. I think you should be able to change/delete them.
I'm a little behind you guys, but I've a couple of questions. Way, way back, FXDiva said he was getting great results with 185f. Is that still the case? I'm testing this one myself using a live demo with Alpari. Looks promising - I've just tweaked it, to exclude the news hours. With 1200+ posts, I may have missed something .. there's lots of discussions and tweaks, but is there actually a more recent "F" version?
One thing I'm just a little unclear on. When excluding trading hours, presumably, any open positions are closed. Is that necessarily a good thing? Might it be that a potential winning position actually loses by exiting prematurely? Also, is news actually a bad thing? I'm probably being naive here, but it could go either way. Some you would win - others you would lose. Won't it all even out in the end?
Anyway, I've excluded the news hours ... so I guess I'll find out for myself.Good guess but sadly the sound for the strategy tester is not among the user option on the Tools/Options/Events.
Kudos on reading the thread thru. It's good to know that some people acutally do that before reasking all the old questions anew.
as for your questions about this EA, I think in both cases the impacts are negliable. one order closing prematurely every so many days isn't going to change the overall outcome when it's a drop in the bucket so to speak. Too bad this doesn't trade live like it backtests. It is however a great learning tool to start one thinking and developing ideas.
One more thing before i go. This system makes use of everything it knows from Values Period Count. ie the last few bars which may only be the last 5 or 7 hours and includes information about any indicators used. Its ESSENTIAL the spread does not change to make an informed decision on the future.
In my opinion the system must not be used with any broker that varies the spread which rules out entirely using Interbank FX and FDD as both these brokers vary the spread.
There might be others but only Alpari (russian not UK) and NF that come to mind will keep the spreads fixed.I think you've got some valid insights here.
1- Who doesn't vary the spreads and supports metatrader4?. I've seen in live forward trading how varying spreads really kill this.
2- I'm curious to learn what you adjusted on the autostop and if you wouldn't mind posting the EA you used with your settings preset?
3- Do you think that an ATR filter might assist this more than non trading hours...I mean the hours filter is only an assumption that certain hours are predictably outside the Effectiveness of the program. Seems to me something more actual market indicator directed might be better than using off hours? Could it be that simple? You may be really onto something here...
I will do another data dump on this an gather some info on the optomized ATR setting for a filter.
there is evidence to suggest that inserting this...
int EnterMarket()
{
//if market volitility is outside this range we leave
if( iATR(NULL,0,12,0) < 0.002 )
{
return (0);
}
this will help performance
there is also some evidence that > 0.0025 may also be troublesome.