1 (edited by hannahis 2017-08-15 06:13:50)

Topic: Hannah's Trade/Portfolio Management Tips

Dear All

I understand some of us want to have trailing be added to our EA as an additional closing condition to our own closing rules.

Here are some tips for you to consider.

1. Use another EA and install it unto another MT4 Chart (make sure to install different charts if you use multiple symbols)

Here is the link that describe each different types of EA and it's function http://forex-strategies-revealed.com/mt … g-stop-eas (FREE)

I personally use e-trailing and I have 50-99 EA attached on a MT4 terminal and this EA can handle large number of open positions.

This web link is also an excellent place for you to search for a Trading Plan/Strategy.  You can read and learn different Trading Strategies and try to create those trading rules in FSB Pro.

2. Forex Tool Shed (FREE) http://www.forextoolshed.com/

a) Trade Manager - Trade Manager, an MT4 Expert Advisor designed to manage your Forex trades for you.  TM  can set your initial Stop Loss and move the stop from 1 to  5 times to specified targets as your trade moves along in profit.  It will also Take Profit at 1 to 5 specified targets, closing a specified lot (or percentage) size at each target.  Additionally, TM offers two pip by pip Trailing Stops and a fractal trailing stop.  TM will also alert you via popup, sound, email, or cell phone when a stop is moved, when profit is taken, when the trade is stopped out, or when a specified profit level has been achieved.

With this completely customizable Expert Advisor, you can take a trade and walk away from your computer, knowing that the Trade Manager will manage all aspects of the trade for you. 

I personally also use this TM (Trade Manager), it allows you to add Trailing, different profit protection steps (I personally like this one and use it with e-trailing (from the 1st link).  It allows me to move my SL at certain pips level.  Eg Move SL to BE when reach 5 pips (50 pips for 5 digits).  Note: the setting is 1 pip (4 dig), so if you are using 5 digit do remember to make the necessary adjustment.



b) EA Monitor - The EA Monitor expert advisor was designed for monitoring the status of trades generated by robots--status information such as number of wins, losses and pips for the day, week and month, historical drawdown stats, and progress and info pertaining to current trades.  But it can be used to monitor any types of trades, not just auto-trades, including trades with and/or without Magic Numbers.  It can be set to monitor a single currency pair or up to nine pairs.  It also does some minor trade management, like moving the stop to Break-Even when a trade gets to a certain profit level, or stopping a trade when its drawdown gets to a certain pip level or percent of account level or taking profit when a specified profit level is achieved.  It also provides popup, sound, and email alerts when profits or drawdown achieves a specified level.

This EA Monitor is good to protect your trading account.  I have about 50 - 70 EA in each Portfolio.  And for different accounts, there are different % risk and different equity level whereby I would like to stop further trading.  So instead of going to FSB EA one by one to change the min account balance to stop future trading (which is tedious if I have to keep changing 50+ EA for different portfolio and each time the account begin to grow).  Hence this EA Monitor can just allow me to it once.


3) Fx Synergy ($497) http://www.pecuniasystems.com/fxsynergy

FX Synergy is the ultimate trade management solution for MT4. Designed by professional traders looking for a better way to manage their trades using the MetaTrader platform. FX Synergy has all the functionality you need to successfully execute and manage your trades, without all the usual hassles. FX Synergy is packed with powerful and unique features, and is designed from the ground up to be quick and easy to use.

Well, this is indeed the ultimate trade management tool I'm looking forward to get.  If anyone is getting, please PM me and I'll try to get group discount for us. 

The software have many features, however I'm going to review from a Algo Traders's perspective cos all our trades are placed by our EA and hence a number of the features aren't useful to Algo Traders.  However there are some features that I wish FSB would have but unfortunately not implemented as I believe Popov has different priorities.  If such features can be added to FSB, then we will all save some money smile (hint hint).

What I like most about this software:

1) The ability to move SL to different levels for profit protection.

2) Copy Trade - as some of us aspire to be trade signal provider or manage other people's account, this software allows you to copy trade from master account to slaves account without the need for you to keep the slave account open in your computer.  Imagine if you need to manage 10 clients, you have to keep 10 clients' MT4 running in your computer and that takes up your computer capacity.

3) Scaling in at targeted pips gains/profit.  This is like FSB Pro Winner features.  However, there is a significant different.

FSB Winner - this setting allows the adding of lots to an already open position but only if it is making profit at the moment of averaging.  In another words, if you happened to have a 100 pips for the 1st position, FSB will keep adding winner lots cos the average is still positive.  Hence, if your 2nd position is in negative, it will still add a 3rd position because the average is still in positive floating profit.  Such adding method doesn't take into account whether the trend continue to be in your favorable position after your 1st position.

Hence, Fx Synergy's Scaling in method is more effective.  It allows you up to 10 scaling steps (when you use auto function).

Why is it more effective?

You are truly adding position when the trend is building up, you only add when the position hit it's targeted profit.


Why is Scaling Important (to me)?

Scaling help you increase your profit exponentially (as long as your EA/strategy has a trading edge)

a) I use Winner as a means to search for "better" winning formula.  When you add Winner as part of your EA feature, it either increase your profit exponentially or increase your loss exponentially, as a result this is a good feature to eliminate strategies that do not have a significant trading edge.

b) Most of us are Breakout traders (if you are not, this scaling function, isn't going to be helpful for you).  Since I have to wait for a "long time" in between breakouts, I maximise each breakout opportunity.  If you open only 1 position in each breakout and I open 4 positions in each breakout, I earn 4x more profit than you can earn in a yr.  What I earn in 1yr would take you 4yrs to reach the same target.  So maximise each breakout opportunity. 

c) Scaling in reduce your potential loss. 

Imagine, you open 1 lot position and the trade turned against you and you exit at SL.

Now, if you use Scaling In, you could then open only 0.01 lot and if the trade turned against you, you only lose 10% of the original SL instead of the 100% of the original SL.

If the trade is going in your direction, you can then add either 0.9 for your 2nd position or you can keep adding at the interval of every 10 pips floating profit (thus, eventually reach your 1 lot allocation).

This Scaling In method is good cos you open lesser positions during ranging period (which is the main culprit that destroy lots of Breakout EA) and you open more positions in stronger and bigger trends.

When Price breakout from the ranging channels, you start adding scaling in either 1) more positions and or 2) more lot allocation.

Most of my Strategies are Breakout EA and often it is the ranging period that caused me to "return" back the previous profits I earn (not all of cos).  So if I can find a way to plug this "drain" then I would have better trade performance. 

In Forex we make money in 2 ways 1) Make profitable Trades, increase your profit and 2) Don't lose your previous profit.  Thus by increase the chances of your winning by Scaling in when you are in profitable position, increase your profit exponentially.

Re: Hannah's Trade/Portfolio Management Tips

Hi Hannahis,
if I look in this portfolio https://www.fxblue.com/users/1111729848/stats

it is loosing. What has happend ?

Are some strategies curvefitted ?

thomas

https://monitortool.jimdofree.com/

3 (edited by hannahis 2017-08-16 01:45:44)

Re: Hannah's Trade/Portfolio Management Tips

Hi Thomas,


1) This portfolio is not meant for me to see the overall results.  It's is not a trading portfolio whereby I put together to make money, it's a portfolio whereby I put all the strategies together so that I can compare their individual results and see which is better than the other.  So do click at the "strategies" tab view individual strategy/EA results so as to know which strategies you want to select and put into your real trading portfolio.  To view my successful portfolio you can go to either http://www.myfxbook.com/members/hannahis365 or https://www.fxblue.com/users/hitport/portfolio

2) Is it curve fitted?

Well as long as we are using computer software, all strategies are curve fitted, it's a matter of degree.  So the question we rather need to ask is "How is it been curve fitted".  Is these trading rules found purely based on statistical coincidence or are these rules a "repeated patterns" based on certain "sound" trading theories.

Thus I always prefer "directed" search, instead of letting the software to search randomly for me, I would input my own trading theory and based on these initial input, "enforce" the software to search for other "complementary" rules to enhance the trading results.  I often find such directed search to be more effective and profitable.

3) The more the merrier?

I understand that the common perspective is that the more rules you have, the more it is curve fitted.  Yes and No.
We often find the word "curve fitted" to be such as taboo word.  As I mentioned I would rather examine how (and not whether) a strategy is curve fitted.  It is much more easier if you ask software to find "profitable EA" with just 1 rule.  Why? cos almost any rules and millions of such rules can be found.  Now when the number of required rules increases, the probability of finding profitable systems based on such number of required rules decreases (thus more curve fitted).  So the end results is, you either get such curve fitted rules that hardly trade at all (in the real market) or you ended up with an excellent system/strategy that has high entry/exit entry that each time it enters into a market position, it enters at such high probability of winning because all other random chances are being filtered/eliminated out.

If we were to read articles of highly successful trading company's, what do we find in common?

They have millions of code in their strategies.  Do you think they believe in this "keep it simple" mentality?

If it is just so simple, why would they need millions of code for simplicity?

If it is so simple, why then it is so hard for us to simply find successful EA? 

I do have a few good EA that is really "simple", just a few rules.  But it's success is not in it's simplicity but in it's ingenuity.

https://www.bloomberg.com/news/articles … ackest-box

In the article, they say

“Renaissance was started by a couple of mathematicians,” Brown said in a 2013 conference for computational linguists. “They had no idea how to program. They’re people who learned how to program by reading computer manuals, and that’s not a particularly good way of learning.” He and Mercer had learned how to build large systems—with many people working on them simultaneously—which was a skill set they used to Renaissance’s advantage. Not that their new field was without challenges. “It’s all noise in finance,” he said.

So what really was they challenges?  It's all noise in finance, it's all about eliminating the noise, that's their challenges.

In another part of the article, it wrote...

The goal of quant trading is similar: to build models that find signals hidden in the noise of the markets.


How to find signals hidden in the noise of the market?


Simply remove these noises and you will find the signals hidden.  In another words, you not only have to input rules based on your trading theory, you also need to eliminate noises that produce false breakouts/entry/exit signals.  These margin of errors are the one that make your system unprofitable, they make you lose more than you earn, plug the leak and your system will start accumulating profit, otherwise whatever is earn is lost in the next trade (you win in breakout only to give away your profit in ranging).


Thus in a nutshell, the key reasons for successful search, in my opinion is 1) rule base searching and 2) the ability to eliminate noises.  Search for strategies based on repeatable and sound trading rules and add relevant filters (via LTF) to eliminate noises (random chances).

Lastly, nice to hear from you Thomas, hope all is well with you.

Best Regards
Hannah

Re: Hannah's Trade/Portfolio Management Tips

Hello Hannahis,

First of all - excellent topic and really valuable information you share, thank you for your generosity. I'm learning a lot through posts like yours and would love the opportunity to understand your belief's on the market better.

I have some questions about your thoughts on the topic of Noise and finding true signals in the noise.   You discuss in your latest post about eliminating noise in the charts so the signals become clear - is smoothing an effective way to do this?

Kind regards,

PT

5 (edited by hannahis 2017-08-16 17:44:55)

Re: Hannah's Trade/Portfolio Management Tips

Hi DoCZero,

1) Is smoothing method an effective way to remove noises?

I personally don't like to use open/close price as my base price setting.

Based on historical data, can one predict what's the closing price going to be at the next bar?  Can you work out a mathematical formula for that, if you can, you are very rich, you will then know where the next closing bar going to be and place your trade now.

But based on historical data, one can have an estimated range of where the Mean price going to be? 

So What's the purpose of having longer historical data? So that based on the past history, we can formulate a more "predictable" repeatable pattern/formula. 

However, if one uses Open or Close price which are so random, we can't "benefit" too much from past historical data to predict what's the future going to be.  But if we know that when MA cross certain parameters, there is a high probability of a breakout.  Why? Cos we are using moving Average "mean" price.

Hence to use past open/close prices to chart out a predictable pattern is like watching a butterfly and guess which direction it's flight pattern going to be.  It's so random, you hardly going to get any meaning predictable patterns.

If we are using 1min time chart, which one you are more confident to "guess" the price "range" correctly for the next 1min, Open/Close price or Mean price?  The mean price of the previous 1min's bar aren't going to be a big difference from the mean price of the next 1min bar (unless a price spike).  Whereas, the price of closing or opening bar can be so random, that past historical data can't help you pinpoint with better accuracy than if you were to guess the next mean price.

So to use open/close price in your EA BASE price setting is to introduce "too much" variability and hence instability to your EA's predictability.

So smoothing period serve more to introduce "predictability" to the EA's formula rather than eliminating noises.  Yes, it does "numb" the price fluctuation and thus to some effect remove "noises" but not all "sound/signals" are noises.  Cos if you remove too much noises, you ended up with no "noise/sound/signal"

Thus not all "sound/signals" are noises.  You need to know what "noises' you want to eliminate, otherwise you also ended up eliminating your very own breakout signals.

Cos if you "numb" the "noises" too much, your EA becomes laggy and slow to respond to market changes.  Too fast, it triggers too many false signals, too slow it become laggy and have poor entry/exit signal.  It's a double edged sword, that's why you have to examine your trading theory to "calculate" your opportunity cost of responding too fast and the opportunity cost of responding too slow and strike a balance. 

This is the decision only you can make.  And only those who study and examine charts will able to come to a rationale decision.  Otherwise, it's just guessing game. 


NOTE: That's why I wish FSB Pro can optimise EA by changing the combination of smoothing period, signal and base price setting as a means to fine tune or search for the best fit.  How many of us keep changing our charts/indicators/parameters?  Year after year I kept using the same chart, indicators and parameters.  No matter how volatile the market is, my charts never fail to indicate to me when a breakout happened.  So what's is it so difficult to produce such repeatable patterns in my EA, why can't I put what I see in my chart into my EA? I find that the possible reason lies in fine tuning the right use of smoothing, signal and base prices, something our "nake" eyes can see but only a computer can fine tune such process. 

I personally always manually fine tune my EA by changing these settings one by one till I achieve the best optimal backtest results and I found lots of profitable EA from this tedious but rewarding process.  Yet I hope one day, I can get the software to do it for me instead of manually doing it.  That would speed up my progress tremendously (hint, hint Popov).

So for those who want to improve your EA's results, try altering your smoothing, signal and base price setting one by one till you get the optimal results.  BUT with each next best results, save it under another EA's name and then load all these variants for testing and see which combination yield the best (in demo testing).  That's why I ended up with thousands of EA cos I never know which combination may ended up with better predictive value.



2) So how to eliminate "noises'

First of all, you need to know what you are looking for, than any other sounds are "distractions/noises".

Determine what you looking for and then looking for incidents whereby other possible/similar setting would have triggered false signals.  I'm speaking in such abstract, I wonder whether you understand.

Let me go a bit more concrete...

If I use H4's MA fast cross MA slow, do you think this signal/rules is "adequate" enough?

Do you think you are getting the best lowest price? (if it's a long/buy position)?

So what can I do to ensure that when I'm entering at H4's crossing, does the lower time frames also "concur/agree" with the H4 signal?

Hence, in a nutshell, one of the best ways to eliminate "noises" is to ensure all other times also concur with the larger signal, otherwise, you are not really entering at the best optimal price.

So if you H4 is showing a higher than/or cross but your lower time frames aren't in agreement with your higher time frames, this potential crossing may be a false signal, ie. Noise, the trend may not be sustainable cos there are other "opposing factors"

I love to use Long Time Frame Features in FSB Pro, that's why in many ways FSB is superior than it's peers because it offers traders/users a very effective and necessary tools to "directed" (not random) search for good EA.

I hope I've answered your questions DoCZeor

Regards
Hannah

Re: Hannah's Trade/Portfolio Management Tips

Hannah,

Absolutely - you've answered my questions! Thank you, you have given me a lot of study to do! (lucky i love it, and I love topics that evolve my beliefs). I have only been using bar closing in all my systems - never even considered using the mean price!!!

I'll update here as I study more - still thank you again for your kindness,

Kind regards,

PT

7 (edited by hannahis 2017-08-17 09:37:43)

Re: Hannah's Trade/Portfolio Management Tips

Hi PT

I'm glad my post is of some help to you.

Just a kind reminder....

Please DON'T use a Blanket Rule and apply Mean prices in All your system.

This is what I'll do, it as I build my system

1. I use Bar Closing as my 1st rule (I just want to make sure, other the rest of the rules will have Previous Bar Value, unticked.  My own personal preference).  For the start, since I don't know what's the best combination for the 1st Opening Rule, it is also very hard to determine the closing rule.  So at this juncture, I would use "CLOSE AND REVERSE" as my Closing option. 

2. Next I need to set the appropriate SL, TP (optional), BE (I usually set between 150 - 250) - this setting "force" the software to find better entry accuracy, so that when your system opens a position, it better be near big trend movement that usually go beyond 150 pips, otherwise it will close at Break Even (BE).

3. Next I set Winner (as additional entry rules), I usually add 4 to 8 position.  This is a good setting to eliminate bad system that does not have a trading edge.  This Winner rule with do 2 things for you, 1) Eliminate bad system will resulted to exponential loses and 2) Any good system will show exponential returns.  Whether you are really going to use Winner later in your live trading, it is entirely up to you.  (This setting is very crucial to me, it makes my portfolio has tremendous pips counts that is insanely high, if you take a look at MyFxBook profile page, you can see that my portfolio has incredible pips count).

NOTE: I'm still trying to persuade Popov to improve the Winner method of tabulation, instead of base on Average profit, it adds only when reach certain pips profit.  It's like copy trade (by add another position) if last position is making money.  So effectively you are only trading if the trend is your friend.  If you entered wrongly, you only make 1 mistake, so in Live trading, I will use 0.01 lot as my 1st position and then subsequently add 0.1 for the remaining 3 to 7 positions.

Once you have completed the setting for "Strategy Properties", you are ready to add in your Trading rules.


3. I add in my second rule (Add in Longer Time Frame 1st, starting with the highest time frame and work your way down to the lower time frames).  Then I start to adjust all the various combinations of smoothing, signal and base price.  With each better results, I save it under a new file/name.  Thus I can have 10 EA just with Rule 1 (bar close) & 2 but in different combination of smoothing, signal and base price.

Determine how you are going to organise your files names cos I use it as my MT4 Magic number too and I save under monthly folders so that it is easy for me to search once my library of EA grow very big. 

Here is a sample of the format 17.08.1000 (whereby it's yy.mm.symbols strategy).

So with each new combination I save it as 17.08.001, 17.08.002 etc

NOTE: FSB max number for MT4 magic number is 10 digits.

4. Once I have complete all the possible combinations, I input Rule #2 (this is the tedious part, but I assure you, it's very worth it, I think part of the reason people simply just put in their rules and don't change these setting, ended up not having very optimal EA and they keep repeating their "failures" instead of shifting gear and re-think what could be done to improve their workflow).

5. At certain point, you may then want to change your Closing Rules (no long use Close and Reverse) cos as you add more and more rules, it become more and more "stricter".  It's good to have strictier rules for opening but too strict closing rules may ended up bad exit.  So at certain point, you may then "transfer the opposite setting of your opening rules to the closing rules and leave it as it is and then continue to add in your opening rules.

The steps above are just a sample of what I would usually do.  You don't have to follow to the dot.  Most importantly, you need to understand what are you aiming to achieve in your trading plan/strategy and understanding the implication of the setting helps you knIow how to use these settings as your filters and to your advantage.

After you have completed, you may end up with 50 to 100+ EA, I'll then put them on Demo test (choose brokers that have identical data feeds in both demo and live.  I've tested JFD Brokers and Tickmill, I open both live and demo accounts and run my portfolio simultaneously and they have very insignificant differences.  Thus I can have confidence that I can replicate the same level of success from demo account to live account.  In the worst scenario, I copy trade from demo account to live account. 

This is just my personal preference.  I understand some people are adamant about using demo account and prefer nano account, to me, a nano account has different spread and trading conditions.  I can't say with confidence that with different trading environment, what makes a person sure he/she can replicate the same success.  As for me, I have been observing and using this methods for a long time and I can repeat my success from demo to live, so no one can talk me out of it.  Experience vs opinion which is stronger?



In Summary

1. Observe, each time you change the setting, how it affects the backtest results.  You will realise that, you don't have to use Mean price all the time!

When you change the combinations of the smoothing, signal and base price, you will realised that it is like a "Symphony".  NOT ALL players need to play the same notes (such as mean price).  You need some players to play low notes (low price), you need some players to play the high notes (High price) and some players to play the mid range such as typical, medium, mean etc

For example, you may want a "sensitive" indicator for Higher Time frames and less "sensitive" indicators for lower time frame, thus H4 offers "stability" due to longer period but with sensitivity, it eliminate the lag.

Once you realised the "Beauty" of combining these indicators to form a "Symphony" you also realised it is beyond any human mind to know such combination without the aid of a software. 

Thus I'm very happy FSB has this function that allows me to change this combination (though I really wish it can be automated, instead of manually adjust these combination).  I seen many software, unfortunately a lot of them use open/close price in their search and thus in my opinion, you can never find EA that can perform consistently because it's formula is based on a highly variable component that is so random that even with large past historical data, the predictability value is too low and random.



I hope the above step by step explanation of my workflow will shed some light and perspective, thus help you fine tune your own work flow.

Wishing you success in your EA development, PT.

Regards
Hannah

Re: Hannah's Trade/Portfolio Management Tips

The advantages of using Winner (Scaling In) option.

Scaling In is almost like copy trade when position is in profit. But instead of copy trade from 1 account to another, you copy trade to your own account. I normally use 0.01 lot for the 1st position and if the trade is profitable, I add 0.1 or more to the subsequent lots. Imagine you originally intended to open 1 lot size and your SL is $500. Given this case, if you are correct, you make money and if you were to be wrong, you stand to lose $500. However, if you were to "test water/market" with an initial lot size of 0.01 for your 1st trade and if you are wrong, you only lose $5 as compared to $500 and if you are correct, you will keep adding position till your eventual total lot size allocated is 1 std lot size (as you would have originally intended).


Now imagine your EA's win/loss ratio is 50/50, Because you use only 0.01 lot instead of 1 lot size, your win/loss (in terms of $) would be 500/5, that's 100 times better return than your original trade performance.

So without having to alter your trading strategy's formula and with the correct use of Scaling In, you can improve your trading performance by 10000% that's better and far more easier than trying to search for a better strategy than your original strategy. What are your chances of search for a better strategy that can yield 10000% more than your original?

With Scaling In you can turn your ordinary EA into an Extra ordinary one (provided you know how to use it to your advantage).


NOTE: FX SYNERGY's Scaling In function only applies to MANUAL TRADING (at this point of time) but the future upgraded version will enable Scaling In for automated trades (as claimed by the company).  However, once the upgraded version is available, the price will be doubled (for your information).

9 (edited by hannahis 2017-08-22 09:58:47)

Re: Hannah's Trade/Portfolio Management Tips


FSB Pro has all the tools you need to build profitable EA (the issue lies in whether you know how to use the functions to your advantage)

For those who use FSB Pro to input your trading rules/theory/conditions...

Even if you know what parameters to use, such as MA 20 cross MA 50.

Do you know which is the best combination to use? For Eg, Base Price and Methods? 

There are 3 options (Base Price, Fast Method, Slow Method) and within these 3 options there is a total of 112 combinations. 
Do you know which is the best combination to use that yields the best trading results?


Beginner FSB users tend to just the "default" setting. 

However other the years, I've realised, simply by changing these settings, it can make a dramatic effect on your trading results, it can turn a losing EA into a winning one.  Hence, the problem lies not because we don't know what parameters to use but the problem lies in, we don't know which combination to use. 


Determining the right parameters aren't really hard.  We look at the chart and we have a good idea which crossing yield seem to identify a breakout.  We often just input our parameters and use the default setting, without changing any of the Base, Signal, Smoothing Method etc (depending on different indicators, some has more combination, some has less).
And then we put to either demo test or live and we are dismay that our trading theory isn't that profitable.  Thereafter, we go about hunting for another trading strategy, another theory and as long as we didn't realised that the problem lies not so much in the wrong use of parameter but in the wrong use of the combination (sometimes, but if you really have a strong trading theory, it's probably not your parameter issues).


Even Simple MA Trading theory works very well.

Not long ago, someone approached me to conduct a Forex Trading Course.

I was hesitant because I do not want to reveal my trading theory which I've work so many years to polish and fine tune it.

So I decided that IF I were to conduct a trading course, maybe I can use a simple MA theory to demonstrate to participants,

1) How to use charts to find the ideal parameters
2) How to use different LTF to eliminate trading noises
3) How to use FSB Pro to fine tune the results

But before I ever conduct that course to teach the MA theory, I want to make sure, it's not just a theory but this theory, can be transformed (via FSB) into a very practical and profitable EA and that I've real results to back my claim. 

A lot of courses out there teaches about concepts and theory but how many of these people teaches the participants to transform these theories into profitable EA.  It's so easy to teach Forex, but how many coaches not only teaches but really impart real practical skills that can help participants create truly profitable EA? 

So I started out doing an experimentation a couple months ago.  To test out a simple MA theory.  (In the past, I upload some of these Simple MA EA in my post but the previous version FSB made these EA execute "wrongly", now with the updated version, a number of my "time sensitive" EA works better.  So I decided to try out this MA theory experimentation again in the new updated version.

For every input I made, I will keep adjusting the various combination one by one, till I get the best optimal back test results (yet saving every new high as a EA file).  I've explain in great details this step in my previous post.

To my pleasant surprise, even a simple MA has really good results with pf 2 and above.  Since it was meant to be an experiment and I didn't know whether even a simple theory such as MA can be profitable or not and hence I didn't really put too much efforts into adding different LTF to eliminate any noise.

Despite such simple rules (and simply but changing the various input), 1 to 3 conditions, I developed nearly 50 profitable EA (PF 1.3 - 2.09 after 3mth of demo testing) from the various combinations of Base Price and Fast/Slow Methods. 

I belief I would have improve the results further, if I were to put in more efforts.  Since I already have a strong trading theory I've been working on over the years, I'll put this Simple MA theory aside till I completed my original theory. 

Whether I conduct the course ultimately or not. is immaterial, this experiment concluded for me that I can create profitable EA by using a simple MA theory (and simply changing the various combination), honestly it isn't really hard for anyone else to create profitable EA, it's a just a matter of hard work.  Till the day Popov decided to automated this function, only those who are hard working enough will know how to create profitable EA.  Finding profitable EA is not impossible.  It's only out of reach for those who are simply too lazy to try.  But for those who are determine to find.  I guarantee you, if you use the method above, you will surely find a couple of profitable EA when you apply your sound trading theory to it.


Summary

Is it possible to create profitable EA using FSB Pro? of course!!!

The problem with many beginners are, we are too lazy to test out the various combination to find which is the best optimal setting.  Although I feel that such manual hard work could have been easily resolved by automation.  Unfortunately, there is no such automated function and we lazily relied on the default setting.

To test out this point, I just started another experimentation.  I've been waiting to do it but held back by the sheer amount of work but I finally feel that this experimentation will pay off richly.

What's the experimentation about?

1) Use a LTF setting, the one that you think is the deciding factor, eg H4 crossing or higher than. 
2) Use only this 1 rule from above (H4 crossover, higher than, etc) and save every possible combinations into a different file/EA name.
3) Use a coding method to organised your file name, ex 1 =simple, 2 = weighted 3 = exponential 4 = smoothing and then next option, 1 = Open, 2 = close, 3 = high etc

So when you look at the results, you can see 111, refers to Simple, Simple, Open (on hindsight, I didn't use this coding method and I have to keep referring back to the FSB to check what's the setting.  thus for future setting, I should use this coding method).

Once you have done the 1st 112 combination, save this as your "template" so that in future, you can open these files again and use another parameter input for the next experiment.

The Results?

By simply changing the various combinations, I get very vast results.  It can turn a losing EA into a winning EA and likewise, a winning trading formula into a losing one simply because we use the wrong combination.

In this link https://www.fxblue.com/users/2089513345/stats,

All these EA have exactly the same, 1 opening condition (besides Bar Open/close as the 1st rule and Close and Reverse as the closing rule), yet you can see how various combination can have vast implication on the trading results.

From the results above (via the link), you will realised that if you happened to use the wrong combination, even if you have the correct parameter, correct trading rules, you will still fail in your search for profitable EA because of the wrong use of combination. 

I mentioned this important information because I fear that many enthusiastic beginners who think they have a winning formula and set themselves out to input their trading rules and thereabout received poor trading results and begin to discard their trading theory, thinking that the fault lies in their theory when in fact it lies in the wrong use of combination instead.

Is it humanly possible that we would know out of these 112 combinations which one yield the optimal results?
If you have 3 or more opening conditions, imagine the sheer amount of hard work you have to go through, the number these combinations grew exponentially with each additional trading rules.  It becomes humanly impossible to find the optimal combination without going through these thousands of combination manually (which is not viable).



In Conclusion

Popov, I know I mentioned this issue several times and after this experiment, I want to further emphasize the importance of automating this function as an alternative "optimization method". 

As a long time FSB Pro user who has successfully created hundreds of profitable EA, do trust me that, if you automate this search function for the best combination, users can find profitable EA more effectively, especially for users who have a trading strategy to start with.  This is what I have been doing, over the years to fine tune my EA, unfortunately, it took me so many years because of the sheer amount of combinations I have to go through them manually (and also over the years, FSB's version has undergone many changes).

Re: Hannah's Trade/Portfolio Management Tips

hannahis wrote:

The advantages of using Winner (Scaling In) option.

I hadn't really paid attention to this feature before, but now that I'm temporarily returning to FSB Pro (from EA Studio) I thought I would give it a try.  Thanks for writing about it.

11 (edited by hannahis 2017-08-23 07:12:03)

Re: Hannah's Trade/Portfolio Management Tips

Hi Sleytus,

FSB Winner is also a good feature when generating EA search, as it will cos either exponential loss or exponential profit.  Thus it makes it easier to spot good EA from average ones.

Secondly, FSB Winner is not exactly the same as profit Scaling in (enter additional entry based on profit).  FSB Winner uses average profit.

If FSB Winner can add only on next Bar open/close when profit is above 10 pips (or users define value), that would be ideal.  But I wonder such method may be too complicated for FSB backtesting to handle?

It's rather easy for me to request for this and that...but to translate those wishes into practicality, it takes a genius to work it out.

Re: Hannah's Trade/Portfolio Management Tips

hannahis wrote:

FSB Winner is also a good feature when generating EA search, as it will cos either exponential loss or exponential profit.  Thus it makes it easier to spot good EA from average ones.

Interesting -- I will give that some more thought and experiment.

Are you suggesting that during strategy generation one could exaggerate the Winner feature to help distinguish good EAs from bad ones, but then back-off to a more conservative value when ready to use the EA?  Or would you keep the original settings used during generation?

13 (edited by hannahis 2017-08-23 07:59:05)

Re: Hannah's Trade/Portfolio Management Tips

Yes, I'm suggesting that using Winner in the generation mode, to sieve out average EA so that you get potentially better ones.

A good entry EA is one that enter with greater accuracy and that it identify a strong and long trend.  Thus if the trend is strong and long, adding more position is viable.

Likewise, for a bad EA that enters in weak trend, adding in a ranging or short trend would be a disaster.

Hence, using Winner function as a alternative filter feature is worth considering.

But do bear in mind what time frame you are using, a higher time frame, such as D1 would mean you are a position trader that has large SL and then adding will make more sense.  It add on next bar open, would mean the next day.

I ONLY use 1 min time frame even if I use D1 because it will add on next bar open and close.


Whether you subsequent would use Winner in your live trading or not depending on your risk management.

I would still use it but instead of the 1 std lot size I use in my testing.  I then change it to 0.01 for the 1st position and then 0.1 for the subsequent positions

All my EA has Winner features and they yield very good results both demo and live.

It takes a long wait for each breakout, so having the Winner option, you maximise your returns with each breakout opportunities.  So that you earn exponentially more than you lose when it is in ranging.

14 (edited by hannahis 2017-08-23 08:41:06)

Re: Hannah's Trade/Portfolio Management Tips

Building Successful EA isn't really going to be difficult.


Here is an example and the results.

In my post above, I mentioned that I did an experiment to test out the 112 combinations.

I only use 1 H4 rule in 1 min time frame/chart and use all the 112 combinations and test it out on demo. (I use Winner = 8 and SL = 500 and BE = 150)

Click here https://www.fxblue.com/users/1111728889 … 8#overview to view the results (use filter date: 20 Aug)

and just based on 3 days trading....can you see the potential in it?  From here I will then select the best combination and then start to add layer upon layer the next trading rules if I want to further improve it.

This method is far more effective than to let the programme search for me.  After you have found your 1st successful rules, then you can either...

1. Use the 1st successful rule/combination to generator subsequent conditions, using FSB Generator

By using "directed" search, i.e. using your 1st rule to "direct" FSB Generator to search for complimentary rules by itself will eliminate million random combinations that aren't complimentary with your 1st rule.  This method will help you narrow your search more successfully.

2. Use the same method to add in another 112 combinations of 2nd rules and then the 3rd rules etc

This method is tedious but most systematical. 


3. Add in your 2nd rule and let FSB optimise it for you.



IMPORTANT: Your 1st LTF must be your deciding factor.  It is your main signaling rule that ultimately confirm for you whether a breakout has occurred or any trading rules that you think is THE most important rule that make or break your EA. 

I build my EA from Larger time frame and then work my way down to the lowest time frame (not the other way round). 


Here is some tips to speed and expand your testing...

I use the same "template" of the 1st batch of 112 Ea and I change it to H1, M30, D1 etc to test out different LTF settings (all these are in 1min time chart but different LTF condition)

And save the Same file name in different folders.  So I have a separate folder for each time frame and I make sure I also have a different Expert folder for each time frame (although all of them has the same file number).

When I install it in MT4, I save the profiles folder and use it in also for H1, M30 etc so that all I need to do is to ensure that the Expert folder has the correct exported LTF conditions such as H1, M30 etc.  Using this method, I don't have to individually install each one of the 112 EA.  I copy and paste the profile and I immediately has the same set up.

NOTE: Provided that the conditions are the same for all these LTF.  Eg if I use MA fast cross MA Slow.  It's the same conditions for all these different LTF, the only difference maybe the parameters.

So I'll have one MT4 account for test out H1, another MT4 for M30 and another MT4 for H4

Re: Hannah's Trade/Portfolio Management Tips

hannahis wrote:

Hence, using Winner function as a alternative filter feature is worth considering.

I like how you explained it in terms of strong and weak trends and how it could be used as a type of filter -- thanks.

Re: Hannah's Trade/Portfolio Management Tips

I build my EA from Larger time frame and then work my way down to the lowest time frame (not the other way round).

Why in that order? Lower timeframes "build" higher TFs, so it would be logical to start from the lowest ones? Otherwise it gets out of sync, doesn't it. I mean "the event" happens first on lower TF and after a while (given it was a significant event) it can be seen on higher TFs as well.

Thanks!

17 (edited by hannahis 2017-08-23 11:24:39)

Re: Hannah's Trade/Portfolio Management Tips

Hi Footon, my dear bro...

Building EA upward from Lower Time Frame 1st or Downward is a matter of perspective.

I'll try to explain my rationale.

I'm a visual person, so let me explain using wave lengths.

In shorter time frames, you get multiple, short wave lengths, especially if you choose a small parameters, you get multiple peaks and troughs.

Likewise in Higher Time Frames you get lesser but longer wave lengths. 

Imagine you super impose the Higher Time Frame Wave length with the shorter time frame wave lengths, you will noticed,

within 1 H4's Wave Length (1 peak and 1 trough), you will have multiple M1's peaks and trough, multiple wave lengths.

What does that mean?

1min's trough, isn't a significant signal to trigger a Breakout

but H4's trough is a significant signal to trigger a Breakout, hence instead of having so many "false signals" if I use M1, using H4's trough will produce better quality signal.

Once I've identify and "mark" every H4's cross over/higher than (by placing the 1st opening condition), I'm "asking" the software to narrow it's search around the 1st's conditions to find me what other lower time frames setting can help me pin point this entry with greater accuracy and yield higher returns. 

Because there are multiple times a H4's cross over many cross back (if indicator paint over) or a higher than may not signify a breakout.  This is especially true in a "pre breakout" conditions whereby most indicators consolidate and two curves are so closely knitted together.  As a results, if we simply use just 1 trading rule such as H4 cross over or higher than, this EA may not be profitable cos it will experience multiple "false" breakouts. 

Thus adding lower time frames ensure or reduce the incidents of "false breakout" when you have "stricter" rules such that if all these rules occurs, the probability of a breakout is higher.


Thus using the Higher Time Frame as your 1st condition is to use that "marking" as your point of reference/search  Furthermore, using a higher time frames as your 1st conditions, you would have narrowed and eliminated many random combinations and cut down the number of such incidents as compared to using a lower time frames (Eg, in 1 week, there may be only 1 or 2  H4 peaks but in 1 week, there are 10 to 50 peaks in 1min etc just an illustration. 


I hope my rationale make much sense to you.

Re: Hannah's Trade/Portfolio Management Tips

Yes, I can understand smile

But what about the following perspective - the trigger is on lower TFs, like a cross or direction change, and higher TFs support this outlook, meaning we anticipate the trigger on higher TFs and therefore enter early (on the right time wink ). It would like: first a cross on the lowest TF and higher ones show suitable conditions like being below the crossing line and rising towards it. It does makes sense, doesn't it?

19 (edited by hannahis 2017-08-23 12:25:54)

Re: Hannah's Trade/Portfolio Management Tips

1. If you are using lower time frame cross and then higher time frame below the crossing line as your trading rules...it becomes "predictive" instead of reactive mode

One reason why automated trade is better than manual trading (for me) is that in automated trading, I don't need to "predict" such as looking at the H4, nearly crossing and assume the crossing will ever happen and enter in this predictive manner.

In automated trading, we don't need to err on the "predictive" side, what happened if this prediction didn't come true?

The advantage of automated trade, we can "react" when the situation such as a cross over come true.  It actually happened and our EA "react" to it.


2. Although it seems like it's the lower time frame that builds up the higher time frame but in actualities, it's the higher time frames that "decide" and influence the lower time frames.

Most institutions don't care too much with the low time frames to make their decisions.  If you observe big breakouts and check all the time frames, it is mostly when there is a "major" crossing in D1.

When such breakout happened, if you look at the M1 or lower time frames, you will see elongated curves in either ends (buy or sell) and even the occasional divergence from the top or bottom wouldn't change the trend in these short time frames, they are more like a "breather" before it continues to return back to the extreme top/bottoms.

So it is not true to say that it's the lower time frame that "builds" up the higher time frames.

20 (edited by hannahis 2017-08-25 20:36:09)

Re: Hannah's Trade/Portfolio Management Tips

UNDERSTANDING Why you FAIL is your ROAD to SUCCESS


As I've mentioned there are 2 schools of thoughts in strategy selection for a portfolio

(Click here if you haven't read it in another post https://forexsb.com/forum/post/45875/#p45875)


I was very excited when I 1st complied my 1st portfolio.  It contain all the BEST EA I've developed over the months.  And I thought to myself, it will definitely bring me good profit since it contains the best.

The 1st week's trading results was awesome.  And then the 2nd week's results lose all if not most of what I've earn in the 1st week.  And 3rd week, I begin to lose further.  I was disillusioned and I thought aren't these good EA, how come they fail, don't tell me they are not robust enough, maybe they can only work for a while and it will start to fail and it's time to discard it and build better ones (true sometimes).

I failed to understand some factors and on hindsight, realised how inexperience I was.

These are my failures

1.  I assumed that my EA will make money in whatever market.  In another words, I overlooked the important factor to know what was my EA built for, ranging markets or Breakouts?

2.  I failed to understand what's the current market situation and whether it "fits" what my EA was built for

3.  I failed to understand and accept losing is part of the game



1. Understand What your EA is built for.

Many FSB users relied on backtesting, walk forward, etc to select their EA and don't like the idea of using demo/nano accounts to test out their EA to understand it further beyond just those statistical numbers.


You may know that those EA selected meet certain acceptance criteria but do you know what your EA is built for?
Since you aren't the person who build the EA and select those trading rules, hence you may not even know

a) whether the EA is for ranging market or for Breakouts.

And if you don't test it out in Demo/nano accounts, you won't know how does your EA behave.

b) Do you know how much will it lose in an average ranging week.  How much it would win in a breakout (approximately).



2. Understand the current market situation (is it favourable for your EA)

And if you have identify what your EA is built for, next is to know when you "launch" your portfolio, do you know whether the current market situation is favourable for your EA?

a) If my EA is a ranging EA, I will time the "launching" of my EA right after a breakout and likewise, if my EA is a breakout, I'll time the launching of my EA after a "certain" periods of ranging, where I think the breakout is going to happen (here is where your trading knowledge will come in handy).

b) If you don't have the trading knowledge and completely clueless of reading the market situations.  Then at least, when you understand what your EA is built for (ranging or breakout), you have to mentally prepare yourself for "failures/losses"



3. Accepting Loses is part of the game in Forex

a) Since we all know there isn't a "holy grail" EA, accepting loses is part of the game.  The conquest is not to avoid loses, the quest is how to play the sum of probability game whereby you win more than you lose.

b) once you expected your EA to fail in those times when the market is not favourable to how your EA is being built/develop, you are not discouraged/disillusioned by the loses.  In fact, you will expect it and accept is as a matter of fact.  And not burden yourself down mentally or psychological and see it as "failures".

c) like all business, there is operating cost.  To keep your EA "operating" all the time, there are "operating" cost when market situations aren't favourable to your EA.

d) The challenge is then how to keep your operating cost down.  That's part of money and risk management skills.  Balancing your portfolio with the right mix of ranging and breakout EA.



Implications:

I often heard of FSB users who said that they were disappointed that after selecting the best EA that pass all their acceptance criteria and when they put their EA to demo/live/nano trading, the EA was unprofitable.

Hey, take heart, even my EA, personally built by me also suffered the same fate.

The problem lies in those pointers I gave above.

Since many don't know what their EA are built for, they may not know why they fail and hence they thought ALL the fault lies in the EA. 

If I put my Breakout EA in a prolong ranging market, let me assure you, it will fail miserably.

These users failed to understand the current market situations wasn't favourable for the EA and they "set themselves up" for failures.

Let me give you a very concrete examples.

If you have a breakout EA and you observe the MT4 charts, this year there are 3 "unusual prolong ranging" markets that will "kill a lot of Breakout EA"

Looking at MT4 D1 chart, you will be able to identify these 3 prolonged ranging

1. Mar 30 April till April 21

2. May 23 till June 26

3. Aug 4 till Aug 25 (a breakout just happened)

A usual ranging lasted approximately 2 weeks but these prolong ranging lasted nearly a month.

So if you happened to put your Breakout EA for live/demo trading, so eager to test out how good was your walk forward selection, you are going for a big disappointment.  And if you failed to understand 1) what your EA is built for, 2) what's the current market situation you are putting your EA to test, is it favourable or unfavourable for your EA, you will then be very discouraged and think FSB backtesting isn't that good and your EA is lousy.

So during those prolong ranging markets, I expected my portfolio to fair poorer than it should be and I don't take these loses as "failures" or take it to heart.  In fact, because I know it's a ranging market, it expected and accepted those loses as part and parcel of the forex game.  And I actually was trying to rush and prepare my next portfolio to be ready to be launched at this coming breakout (which happens today at the very moment of my writing and unfortunately, I missed this wave....never mind, like any surfer, it's about preparing and timing one for the next big wave to come).


In Summary

So understanding those 3 pointers above will prevent you from suffer any unnecessary heartaches and learn not to take failures too hard.

As much as you prefer to use walk forward for your EA search and selection.  Do understand there are these statistics only give you an overview of what's the probably end results of your EA, It doesn't tell you, the process.  It can't tell you, how your EA will behave in real market situation and observing how your EA behave in a demo/live accounts will give you an added perspective and greater confidence in what to expect, so that when your EA started to lose money, you won't panic cos you know it's part of the game.  If you don't understand your EA, you will lose confidence and stop your live trading and discard your EA and then you are back to square one all over again and soon you will realised this failure cycle will keep repeating itself and if you don't understand why, you will soon quit the forex game.

21 (edited by sleytus 2017-08-25 23:24:38)

Re: Hannah's Trade/Portfolio Management Tips

Thanks for writing about this -- very interesting.  I have certainly been frustrated at times when an EA that has great stats then performs poorly.  Explaining this in terms of breakout and ranging markets gives me a way to better think about why an EA may perform in a way I didn't expect.

I had a few of questions -- partly because I am relatively new to this and, also, because I am too lazy to do the research myself.

1. When you create an EA, do you intentionally design it for breakouts or ranging, or is it only later that you determine which type of market it trades best in?

2. If your design is intentional -- then I assume you know ahead of time which indicators or combination of indicators work best for different markets?

3. Suppose I have an EA (e.g. EURUSD H1) and I would like to test it to see if it performs differently for breakout or ranging markets.  Can you point me to a specific DataHorizon that I could use to see the difference?  And, if so, doesn't it make sense to keep in my bag of tricks two DataHorizons -- one for breakouts and one for ranging -- that I can routinely refer back to in order to check whether my latest and greatest EA has a problem with one or both of those types of markets.

4. Related to (3) -- do you think it would be helpful to create a library of DataHorizons that encapsulate particular market conditions and which can be repeatedly used to test newly generated or hand-created EAs?

5. How does one take this into account when trading -- do you turn EAs on and off based on market trend or do you just let things run and take your lumps as they come (hoping that in the end the overall performance will be in your favor)?

Thanks, Hannah...

Re: Hannah's Trade/Portfolio Management Tips

Hannah -- Your post gave me an idea.  I've started a new thread under "Forex Strategies" entitled "Using Synthetic DataHorizons For Generating Strategies".

Re: Hannah's Trade/Portfolio Management Tips

Give me your support if you like my post (no obligation)

If you have been reading my post and for some reasons you would like to support me when you decided to buy any FSB product, do use my referral link here https://forexsb.com/160.html

Do drop me a private message (PM) before your purchase so that I can personally thank you and offer any tips of what suits you the best.

Re: Hannah's Trade/Portfolio Management Tips

What you think over this Idea...

Strategie search with big Spread (30 in EurUsd) and run then with Spread 0-5....

I think the Strategie can good or what you think?

25 (edited by hannahis 2017-08-30 13:35:35)

Re: Hannah's Trade/Portfolio Management Tips

First of all, if you have been reading my post, you probably would know by now that I approach FSB from a trader's perspective.

In another words, I use FSB as my trading tool to improve my trading skills and thereby make better and better EA.

if you are hoping to have a tools that can provide highly profitable EA without much hard work, then you will be in for a nasty surprise.  But given adequate dedication and thinking, you will able to develop better EA, which otherwise would be tough to do so on your own.

HOW TO SET "OPTIMAL" STOP LOSS

In my earlier post, I did an experiment, I use the "deciding" factor rule such as H1 or H4 and use Close and Reverse as my closing and then I use all the 112 combinations available (such as the Fast Method, Slow Method and Base Price) and I test run these 112 EA on a demo account. 

The original Stop Loss (SL) I use was 1000 (was a mistake, I thought I set it to 500).

Using Fx Blue service to collect my trade statistics, I can then able to understand whether I've set too high SL or too low SL.

Here is an example if you want to understand further.

1.  Click on the link https://www.fxblue.com/users/2089513347/stats
2.  go to "Strategy" tab
3.  Double click on "net profit" to see the best EA

Now scroll to the right and look for

1. Ave Win
2. Ave loss
3. Worst Trade


So if here you can see that

1. Ave Win is about 450 (only those profitable EA, I don't count those unprofitable EA)
2. Ave Loss is about 350 (for those profitable EA)
3. Worst Trade without SL is 480
    Worst Trade when stopped by SL is 1000


So from the statistics above, I realised for my profitable EA, SL 1000 is too high.  if I were to set my SL between 250 to 300, I may reduce my ave loss further.

Hence, when I've finished testing this group of EA, I'll then shortlist the profitable ones and create 2 groups, one with SL 250 and another with SL300 to see which has better results (then the original and between both of these groups)

Now if you happened to see that your SL is mostly around your SL setting, that means most of your trade are stop out by SL and if your Pf is low then it may means that you have set too tight SL and you may want to consider increasing your SL to give your EA more "trading space".