Candidate194m1 – EURUSD M15 – PF 1.32 – DD 16.23%
6 replies
neilrickaby
12 years ago #110529
please find attached new EA for your consideration. This EA is for education purposes only. *Attached genetic builder file has been modified to produce better result, so EA is attached also – thanks
Mark Fric
12 years ago #120109
the results look really good, have you been testing it on Dukascopy tick data again?
Mark
StrategyQuant architect
neilrickaby
12 years ago #120110
the results look really good, have you been testing it on Dukascopy tick data again?
thanks Mark, yes I do all my backtests on 99% dukascopy data. The way i create EA’s is:[list]
[*]Download the Dukascopy tick data
[*]Export the 1min TF data from MT4 and import into GB
[*]Create GB strategies from the Tick Data and re-backtest in MT4 on the tick data again (i don’t trust backtesting!) <img src='http://www.geneticbuilder.com/forum/public/style_emoticons//smile.png’ class=’bbc_emoticon’ alt=’:)’ />
[/list]
As you say in the manual there is some difference between how MT4 and GB simulate ticks so i find this the best method for me. I’m creating strategies for 15m at the moment and basically what is profitable in GB is profitable in MT4 80% of the time. I was previously using a 5min timeframe but I had too many discrepancies between GB and MT4 results – basically I think the higher the TF the more accurate it is so far, but I’m still learning! <img src='http://www.geneticbuilder.com/forum/public/style_emoticons//smile.png’ class=’bbc_emoticon’ alt=’:)’ />
tnickel
11 years ago #120262
Hi,
I testet the Strategie EuroUSD 0.194 with the actual Version of GB 2.1, (newest Version)
but the result was different on my system.
I testet on the last year period.
Ticksimulation (slowest)
Spread 2,5
TimeM15
But I don´t get this good results.
Was has happend?
I testet:
a)ActiveTrades
b)DukasCopy
c)DukasCopy realTick
d)fhdb
e)Data from forextester
All equity curves are different. I can´t see a winning System.
https://monitortool.jimdofree.com/
tnickel
11 years ago #120263
Here are the screenshoots of my backtests with GenericBuilder
thomas
https://monitortool.jimdofree.com/
neilrickaby
11 years ago #120264
Hi Tnickel – I notice you are only using 1 yrs worth of data to test – this might explain some of it – also I only use Dukascopy Tick Data so couldn’t comment on the other sources – the Dukascopy Tick Data equity curve looks positive to me also – I know it’s not a worldbeater but it’s only one years worth of data? Also I get a spread of 2 for EURUSD with my broker, so I think I might have run the backtest with this set rather than your 2.5, which will affect overall profitability. [b]Quick question for you:[/b] If you have 5 data sources for testing and 5 different results, which one will you use or trust?
tnickel
11 years ago #120265
Hi neilrickaby,
If i have 5 datasources and 5 different results, so I don´t use the strategy.
I look for a strategy who yield stable profits on all 5 datasources.
I only have realtickdata from dukascopy for the last year. So I can better compare this 5 datasources.
I think it is possible to find a pool of strategies who have good results on different datasources. (on Simul tick slowest)
I found some. But if I test this strategies with real tickdata from dukascopy or exported tickdata from metatrader the result is very bad. I don´t know why?
thomas
https://monitortool.jimdofree.com/
Viewing 6 replies - 1 through 6 (of 6 total)