SFE Night Scalper EA

[Attractor]

An user question:

Dear Joel, how to custom the "Test_End_Trade_Hour" using version 5.0? New version only has one parameter, "Test_Open_Hour".

The new version only test if worth open the trades in the 'Test_Open_Hour' hour.
First of all, you have to recognize the hour 'zero' of your broker. In gmt+2/+3 brokers, like icmarkets, is just 0.

This hour, is the best for the system, but the spreads are very bad, so, you have to test this hour with 30 pippetes for example.
The pairs that in the hour 0 have profit ratios >=2.0-2.5 are Ok for trade at this hour.
In the other hours, a spread of 10 can be used. The pairs that have profit ratios >= 1-50-1.75 I like to use in these hours.

The potential hours to trade apart of the zero hour, are the hours before and the hours after, then, at final, if the pair can trade at 0 hour, the next step is see the 1,2 hour, and the 21,22,23, and depending of the results, establish the final range in live, for example from 23 to 1.

Really, don't push to much effort in the backtest of this EA, these kind of EAs in these hours are difficult to backtest properly. I'm guiding more for the real performance of the account than the backtest itself. And the current selection of pairs and hours, the default parameters, I have a big confidence, and the weight of jpy and aud currency is low, for avoid risks.

But beware, only with spreads like icmarkets or close to icmarkets the EA will be a clear winner.
Remember that the EA stores a file in the metatrader directory with the conditions of the spread in the open. If some user don't use icmarkets and want to share these files, we can see the reality of what happens....

[YoG12]

Thank you for quick fix.
Error message was gone from my terminal.

[ForexBenchmark]

An user question:

Dear Joel, how to custom the "Test_End_Trade_Hour" using version 5.0? New version only has one parameter, "Test_Open_Hour".

The new version only test if worth open the trades in the 'Test_Open_Hour' hour.
First of all, you have to recognize the hour 'zero' of your broker. In gmt+2/+3 brokers, like icmarkets, is just 0.

This hour, is the best for the system, but the spreads are very bad, so, you have to test this hour with 30 pippetes for example.
The pairs that in the hour 0 have profit ratios >=2.0-2.5 are Ok for trade at this hour.
In the other hours, a spread of 10 can be used. The pairs that have profit ratios >= 1-50-1.75 I like to use in these hours.

The potential hours to trade apart of the zero hour, are the hours before and the hours after, then, at final, if the pair can trade at 0 hour, the next step is see the 1,2 hour, and the 21,22,23, and depending of the results, establish the final range in live, for example from 23 to 1.

Really, don't push to much effort in the backtest of this EA, these kind of EAs in these hours are difficult to backtest properly. I'm guiding more for the real performance of the account than the backtest itself. And the current selection of pairs and hours, the default parameters, I have a big confidence, and the weight of jpy and aud currency is low, for avoid risks.

But beware, only with spreads like icmarkets or close to icmarkets the EA will be a clear winner.
Remember that the EA stores a file in the metatrader directory with the conditions of the spread in the open. If some user don't use icmarkets and want to share these files, we can see the reality of what happens....

Hello Joel,

in case I want to backtest the default settings of your EA, I will have to create 4 independent backtests for a single pair like CHFJPY (you set here the default settings of 21-0)?


Best regards,
Forex Benchmark Team

[Attractor]

yes, I know this is unusual, but the best way to proceed is this. Identify what pairs are usable in the 'zero' hour with high spread, and search if in the other hours some of the pairs can be used.

Then, the way to do the bactkest is as you say. But look, the backtest of this kind of systems only serves as an orientation, the backtest results will be always a lot better than the real performance. But can be still profitable as we see in this Night Scalpe. I have the sensation that at this hour, there is a battle between Night Scalpers, playing also with the initial disavantage of the artificially widened spreads of the brokers, but seem there are a lot of bad EAs, then exists the possibility to make profit for the best of them.

Another problem of backtest these EAs is if the 'zero' hour of the historical prices changes along the years, because for an unknow reason, in some brokers the downloaded data is not homogeneus along the years in relation to the gmt.

[Jimmy]

I just found out that ForexBenchmark released a new review on SFE Night Scalper a few days ago: https://forexbenchmark.com/eas/reviews/sfe-night-scalper/

The live performance was 4.5 stars but it was only rated 3 stars overall because the backtest was not reliable (as Joel had pointed out) and the EA is not very customisable, and he considered the price was too steep.

I think his points are fair, but to me the live performance result is all that matters.

[Attractor]

Be aware, is easy manipulate the backtest, when I see some EAs with straight line backtest, seems very strange not see any kind of DD or stagnation there. And real action put these kind of backtest in contradiction since the start.

But it's true that for this kind of EAs, the reliability of MT4 backtest is below the desired, and don't make easy get conclusions.

I noticed this a long time ago, for this reason I have been working on improve the reliability of the backtest while I have been working on SFE Stealth.

The work has end in success and also affects the SFE Night Scalper, where form now I will be able to offer a more realistic view of the expected performance.

I will post more information this night

PD: I agree with the reviews in general, but I think don't have sense that the 'customizability' issue has this weight in the puntuation. For my experience, 95% of the SFE users wants to exactly replicate the base signal without do any other effort, then, launch an EA which is plug and use in one minute, don't deserve that bad punctuation.

And my algorithms are based in that the parameters, or work for all the pairs, or are useless. And once are correct, they are internal, don't exist the need to change the value externally.

For me, the customization in this case comes from the basket selection of EAs, and as I said, I will post useful information this night.

[Jimmy]

Yes I think “customisability” really is only required for those EAs that are not fully optimised, and people need to do their own tweaking to get better performance.

As you said, I believe if the EA developer has already perfected the settings for an EA, then there really isn’t a need for further customisation.

[Byte]

I agree. But in the end, it is also a personal preference. Some people like to optimize themselves and are really only looking for a tool that they can play with. I really like having EAs that are optimized and I just can transparently control the risk. So in a way, to me, it is important to know which kind of EA I am buying but it is neither a plus nor a minus to me.

[Attractor]

Hi there,

I'm sorry, I'm delaying the information but I'm doing the last checks before launch the new version of SFE Night Scalper.

I hope finally tomorrow post the full information and update the EA. But I will advance the main guidelines of the advances because I'm very excited with them.

The 'advances' is that I'm making compatible the code of my EAs with Metatrader 5. This is not only because I want to use or sell EAs for this platform, the main reason to use MT5 is because the backtest are a lot more reliable, and this can make the difference in EAs like SFE NS for extract conclusions.

The construction of the prices I see is better in mt5, but not only this, in mt5, in the backtest, the commission and the swap is taking into account, and what is better, the variable spread based on the bars of the broker is also stored and used.

The first consequence of this in any kind of scalper EA, happens a BIIIG donwgrade of the backtest    I think this is also one of the reasons why the developers resist to change the platform, because is a lot more difficult convince potential buyers with sexy backtests.

But if your objective is develop true EAs, this is a very good advance, because you can analyse the situation more accurate, and what happens in backtest and in real, is more close.

Then, we will see now backtest more ugly, but more realistic, and we will can develop and select better the configurations.

And other of the TOP advances of the mt5 backtest. You can backtest all the pairs at the same time, I mean, you can begin to introduce rules based on the simmultaniety of orders of different symbols, for control the top exposition or others. And with the rules extracted, you can apply these rules in live action also in mt4 code, because there obviously you have the information of all the pairs opened at any moment.

If I have messages in mql5 of any of you for reply I'm sorry, I will reply in next days, the priority is always the EAs.

[paulomaks]

... so very interesting what you tell ... and yes i am very sure you are right that this - more realistic approach - makes developers hesitate to go the route of MT5 ... looking much forward to your final news    .. and hope there will be news on Stealth EA as well ..

[Attractor]

Good. Attached here is the backtest made in metatrader 5 of the SFE Night Scalper 6.0 for MT4 that I will update. The backtest is the html generated by mt5. I have not attached the quantanalyzer report because at this moment quantanalyser is not loading well these statements. Is a problem with the partial closes, the report shown after load don't have any sense...

While the backtest in mt4 is impressive, the backtest in mt5 is ... disappointing at first sight. But is not as bad, with mt5 you have  a more realistic information about the possibilite of the EA, and in some pairs and some situations, you receive a clear view of what is pure garbage.

SFE Night Scalper has been demonstrating that can develop a stable growth, and with this new version using the mt5 platform as develop platform, I expect that the current changes and the current selection of pairs and hours will be a good advance in the potential of the EA.

I have been working with 2 bactkest periods, from 2015 and from 2017. This kind of EA create trading opportunities almost every day, go more far in years don't have sense. Was a period more years ago where the big action happened at alll the hours, and obviously this kind of EA would suffer there. But at this moment, and for years already, find big action is the exception, as an example the FOMC today !! The eurusd price has not moved anything....

Then, for the moment seems very safe use this kind of EA. In 2015-2016, still happened more movement, and the performace is worst, but since middle of 2017, the EA shows an impressive and stable line of growth. The current selection of pairs and hours has been based in this  period since middel of 2017.

The next image shows the current selection.



As mt5 backtes let to use all the pairs at the same time, I have been testing the concurrence of different trades at the same time, and I established some limitations for not been caught with a really big number of trades or exposition in one currency.

The recommended size of the trades for the version 6.0, is 0.01 lots for each 500$. This is a low risk level, with maximal DD expected about 20%, and at the same time, with this size ratio, the EA can be used without problems in accounts with leverages of 1:20 - 1:30.

And I expect that the average yearly profit can be about x2 - x3 times the theoric maximal DD, enough for keep the account in healthy situation.

The maximium risk recommended is use 0.01 lot each 300$. And in the current signal in mt4, I will use 0.01 lots each 400$, for each 360 € because the account is in Euros.

[Attractor]

In the chf  signal, I'm using the 4  chf symbols, with the same hours of the image of the previous message. And with a size of 0.01 lots per 240$, 220€ in the case of my account.

[Attractor]

I forgot one thing. I added a function that download online if I manually disable some currency. I mean, for example in the AUD rate decision day, is not the best open AUD symbols, then I put this information in the smartforexexpert server, and the EA will no open trades on AUD pairs until the restriction is removed again.
The add and remove of the currencies I do it manual, but the EA load this information automatically.

If someone want to follow also my manual 'disable', can do it adding the adress:   https://smartforexexpert.com to the list of 'allow web request'

When a symbol is disabled appears in light yellow like the image, where USD pairs has been disabled as an example.



I want to automatize this 100% in future, for the moment I'm using this semi manual method.

[CristianMax]

Hello,
this is my first message in this forum.
I´ve a doubt:
So far, what is the best way to deactivate a currency? I see that tonight there is fundamental in Nzd and I would like to turn off EurNzd.
By the way, thank you very much for the effort you are making to update and innovate with this great EA

[Attractor]

Hello,
this is my first message in this forum.
I´ve a doubt:
So far, what is the best way to deactivate a currency? I see that tonight there is fundamental in Nzd and I would like to turn off EurNzd.
By the way, thank you very much for the effort you are making to update and innovate with this great EA

Thank you for your participation Cristian.

Just now I added NZD to the list of disabled, and tomorrow I will add the currency again. If anyone don't want to use a symbol in a punctual moment and the symbol is not disabled by the 'server' in yellow, the way is remove it from the list.
I'm working on automatize this at 100 in the near future.

I see 2 limit orders are opened now before the disabling, I'm gonna remove the orders before the event. Sometimes the event can make a good money, but in general is risky. And the impact of the events, also in the near future I will have conclusions, because backtesting in mt5 I can take into consideration the calendar. This takes a good work, so will take some time, but I will do it.