Collaboration Dolly + Isakas + Nina System - page 24
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
Thank you for the link on candlesticks. But how you interpret and use the DigiStar Indicator specifically?
Thank You,
CigarguyFOR LONGTIME FRAMERS say 4H, techinical traders use candlestcik formation as a tool to decide thier trade, DigiStar is one of those formation, look at page 16 in the link above, its well explanied .. for example
morning star 1_1 ( Undecisoin candle between two decisoin candles) is a sign for bullish reversal ... and so on
I have been forward testing Shinigami's EA. Yesterday it won 72 pips and there was no trade today. Last week's result was 55 pips net. Very promising
Attached are some backtest results. I tried to optimize some settings of DIN_KusKus_EA_v2.0. It is not that easy being many external variables there. So, I cannot say I resolved the optimization task. The result of the last test is slightly better than the default was though.
Modelling quality is 90%.
(I hope I used the correct database this time, if anybody is interested please check it up
)
Backtesting EAs
chrisstoff,
When I run backtest on the v2 ea I test on 1 minute timeframe. The results can be very different from testing on the 15 or 1 hr timframes but I believe should be more accurate. This has to do with the exits at stoploss or takeprofit. When the signal takes out a trade it is on a closed candle so no problem there. However, if there is a large move on the 1 hr that hits stoploss or takeprofit it is more accurate to look at the 1 minute timeframe.
Robert
MrPip,
Thank you for your comments. I agree, test with M1 data is much more reliable than it is using higher time frame's data. The thing is though, that I used M1 data during my earlier tests either, otherwise I would not have been able to reach 90% modelling quality. I used IB FX platform and Alpari data downloaded from here: http://www.alpari-idc.ru/en/dc/databank.php.
Here goes the M1 result in two variations. I do not think the difference is relevant to my earlier results just because I used M1 data and "Every tick (based on all available least timeframes with fractal interpolation of every tick)" model for those tests.
I used M30 and sometimes M15 timeframes (and once M5) because I think most traders use M30 or M15 for trading breakout systems. (It is just an assumption, I do not state that there is no trader who uses M1 for breakout systems but I do not think that using M1 is common for this trading technique.)
I also attach Hendrick's ebook about how to reach 90% modelling quality for those members (if any) who have not come across with that.
backtest
chrisstoff,
I would not suggest using Alpari data with IBFX. Why not just download Alpari MT4 and backtest with that. One concern is when the broker server time is different than the broker where you get the data. In that situation the daata will be shifted. The same is true between data from IBFX and FXDD. The data is differnt by 2 hours so 4 hour bars are completely different.
Also when I say test on 1 minute data you only need modeling quality around 25% as that is the best the formula will show. The indicators are all locked to the 30 minute, 1 hour and 4 hour timeframes for signals. The reason for using 1 minute time frame is for more accurate exits from stoploss or take profit.
The actual signals do not come from the 1 minute candles. That is why I have an input for SignalTimeFrame.
Robert
Dear ALL
Sorry for my absence, flood disaster has invaded my city,the electricity has been turned off for 5 days until now. I couldn't follow this discussion for a moment.
Anyone, please be kindly posting update files which are necessary to be put on the first page. I will attach those file ASAP,
I really glad to see Mr Pip and Mini Me helping all the traders here. I promise I will be active as soon as this disaster over.
May Pips with you all and always remember what Mini Me said
-Sharing is Caring-
chrisstoff,
I would not suggest using Alpari data with IBFX. Why not just download Alpari MT4 and backtest with that. One concern is when the broker server time is different than the broker where you get the data. In that situation the daata will be shifted. The same is true between data from IBFX and FXDD. The data is differnt by 2 hours so 4 hour bars are completely different.
Also when I say test on 1 minute data you only need modeling quality around 25% as that is the best the formula will show. The indicators are all locked to the 30 minute, 1 hour and 4 hour timeframes for signals. The reason for using 1 minute time frame is for more accurate exits from stoploss or take profit.
The actual signals do not come from the 1 minute candles. That is why I have an input for SignalTimeFrame.
RobertANCOLL , Glad to have you back, and I am sorry for the bad weather I hope your fine and no one is hurt... if it makes you feel better today it was -23C here in Montreal, and it gets worse.
Thanks Robert for the explanation, its very good point you have raised here. That’s why both back and forward tests are required b4 deciding to use an EA on a live account.
For the record I have noticed a couple of times that the data feed differ from borker to another i took a screen shots and send it to MIG and their reply was
((( Dear Sir,
There are some wrong quotes coming to the market watch. They do not affect the trades, because no orders are processed on these wrong prices, but they unfortunately affect the graphs.
Best regards,
xxxxxxxxx
MIG Investments SA)))
This does not happen a lot, but it makes the price on MT4 charts useless.
M1 Test Results - How Increase Effectiveness?
MrPip,
Thank you for your post.
Attached are the test results as per your suggestion. Of course there are differences to the earlier ones but I still do not think they are that significant (but everybody has the possibility to decide what is a significant difference for him/her).
I usually use Alpari data since their database is the most reliable in the industry according to the general opinion of most of those traders who think backtest has some importance.
I usually use IBFX platform because I use live service of this brokerage as well and I do not use Alpari service live for the time being.
I do not think that the 2 hours difference in time has much relevance unless we test H4 or Daily data. That is because H4 and D1 data are influenced by 2 hours difference in time zone used, H1 and lower bars are not, so they should be the same. But they are not the same due to other reasons, such as different filtering, datafeed, etc. that all are different from broker to broker. But trading in real life is also different from broker to broker.
I think the important task would be to analyse the results of all the tests and look for solutions to make the EA more profitable. The ratio of profit trades is good, always above 66%. Albeit the ratio of average profit trade per loss trade is poor, the average loss trade is 2-3 times bigger than the profit trade.
So, to decrease the loss trades value and increase the profit trade value could help much. I tried to play with the settings but with not much success as I said in my post #239.
PS.: Due to Alpari's lower leverage (Alpari has 1:100 while IBFX has 1:200) Test4 with Lots=8 and Shinigami's test with MMRiskFactor=0.8; was not able to reproduce. This is also a reason for the different results.
chrisstoff,
I would not suggest using Alpari data with IBFX. Why not just download Alpari MT4 and backtest with that. One concern is when the broker server time is different than the broker where you get the data. In that situation the daata will be shifted. The same is true between data from IBFX and FXDD. The data is differnt by 2 hours so 4 hour bars are completely different.
Also when I say test on 1 minute data you only need modeling quality around 25% as that is the best the formula will show. The indicators are all locked to the 30 minute, 1 hour and 4 hour timeframes for signals. The reason for using 1 minute time frame is for more accurate exits from stoploss or take profit.
The actual signals do not come from the 1 minute candles. That is why I have an input for SignalTimeFrame.
RobertDIN_KusKus_V2 any additional indicators needed?
Robert,
You are an amazing programmer, appreciate your work and I am testing V2
With this new Version do we need any extra indicators in our indicator folder compared to V1?
Tks
Berdj
DIN_Kuskus-EA
MiniMe,
I agree about trading with the trend. That is what my backtest with v2 confirmed. I tested all the variations for checking trend with many different trend indicators and in all cases the EA did better when not using the trend for entry.
However using trend for exit did help a great deal. It kept the trades open longer. This also resulting in less trades.
Robert