EURUSD on the 1 hour chart

Hello,

The following is a strategy for the EURUSD on the 1 hour chart.

Opinions are very welcome.

my strategy was build in 2017 and traded live for 2 years, this is the major robustnest test

you couldnt compare strategy without EOF, without slippage with this kind of strategy…in real trading you will very soon find out

another major issue – its generated for EURUSD UTC0, you will be trading in this timezone?

Good for you. Congratulations if that’s working to you but I don’t buy it 😉

nobody is selling here, you already downloaded it for free…its my gift for you 🙂

MC tests are good to see worst performance that can be expected. Otherwise they have the tendency to remove huge amount great strategies. I think this using 50 % of RDD @95% is  no good. I got much better result on unseen OOS using an fixed (2-3) RDD @100% instead and it also feels better knowing that on 300 simulated runs no one was a looser. This way the High RDD strategies are not deleted by MC test and those are the ones that performs best coming OOS and also among the few I have from SQ3 that actually have performed great on Real accounts.

Marcel wrote:

hankeys wrote:

most probably because of user 🙂 from this picture i cant tell you anything – info whats wrong are in logs in metatrader you are doing something wrong and i dont totally understand why the hell many people need to do backtests in metatrader? if you dont have tick data suite, backtesting in MT is nonsense

Hello, Thank you for your answer. Enclosed I now send you the excerpt from the log file to your EA. Basically the test in the Metatrader is about what exactly your EA didn’t do, namely a test that it works at all. I would be very happy if you could tell me why your EA doesn’t work. Thanks a lot!

@Hankeys:

would you be so kind and tell me why the file you gave me does not trigger a trade in the Metatrader?

hard to say, never have problem with any strategy – i dont know why the MT is telling you thats not a EA

i dont know how are you producing MQL code from SQ, i dont know your MT build you are using, etc. etc.

try this comiled EX4

Same. No change. Your EA do not start any action in the backtest of metatrader and i can not install him in a chart…..

cant help, it works for me…try this MQL file and compile for yourself

Well, maybe we’re talking about obvious thongs but if @notch you say Genetic algorithm is not searching by improvement… well nothing to argue, look at wikipedia.

And yes, curve fitting is worst nightmare. All of us are curvefitting, if not look at this sample of 3.000 strategies. They have an average Profit Factor = 1,6 aprox on 15 years. After the first 18 months this 3.000 strategies go to PF=1,4. That’s Mr curvefitting in action

Gentlemen,

i appreciate this lively conversation and i appreciate the diversity because i can give my part of knowledge and take something out of every variance.

Please don’t get lost in a competition that is not one.
We can all profit from each other and should do so because only in the community can we achieve what we all want……earn money…..as fast and as long as possible.

Alone never one will be as strong as a community so let us work together and not against each other.

@hankeys:

do you have a custom indicator that I don’t have?

Enric wrote:

Well, maybe we’re talking about obvious thongs but if @notch you say Genetic algorithm is not searching by improvement… well nothing to argue, look at wikipedia. And yes, curve fitting is worst nightmare. All of us are curvefitting, if not look at this sample of 3.000 strategies. They have an average Profit Factor = 1,6 aprox on 15 years. After the first 18 months this 3.000 strategies go to PF=1,4. That’s Mr curvefitting in action

No it is actually just one year of data.  You have to compare it with the combined worst year during this 15 years not the average. But thanks this kind of comparison can be useful indeed to find adaptable good performing strategies using Walkforward  on another set of 15 years unseen data. This should prove the usefulness of Walkfoward  optimization.

mabi wrote:

MC tests are good to see worst performance that can be expected. Otherwise they have the tendency to remove huge amount great strategies. I think this using 50 % of RDD @95% is no good. I got much better result on unseen OOS using an fixed (2-3) RDD @100% instead and it also feels better knowing that on 300 simulated runs no one was a looser. This way the High RDD strategies are not deleted by MC test and those are the ones that performs best coming OOS and also among the few I have from SQ3 that actually have performed great on Real accounts.

Makes sense. If I can produce 100000 strategies in a week, I’m okay with losing a huge amount of great strategies due to false negatives. It comes with the research progress.

Well, you want to keep as many as possible in order to diversify to later merge them so You can cover different market conditions. Being to tuf when especially doing MC you will only have a few strategies left that takes the same trades . If you ease up on this you can find 20 strategies on the same instrument and time frame that do not take the same trades and it does not mean they are curvefitted and  MC have not been able to prove an improved performance going forward unfortunately. This is today easily tested using a taskflow having X amount of OOS years and then just Run different MC tests to see how this improved performance of the strategies going forward and You find that is has no effect at all.  Well MC spread ( alone) with very tuf settings have a >70% performance on OOS unseen.

Just want to add that i am still using MC even thought i have proved to my self that it has no effect .. 🙂

no custom indicators – its sucks for now in SQX.

idea, that some mystic custom indicator will win this fight against markets is nonsense – everything is changing very fast. Keep the strategies simple, thats the basics

i am running my first real ptfs from SQX strategies so far, taking strats from demo account where are 400 strategies for now, still adding more