Random Walk Theory applying in Expert Advisor | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Probabilities in a Random Walk (Question about one CP's post) 43 replies

Random Walk (The unbeatable!) 354 replies

Is random really random? 231 replies

Expert Advisor with random entries 3 replies

Random walk and Pascals triangle 64 replies

Attachments: Random Walk Theory applying in Expert Advisor
Exit Attachments

Random Walk Theory applying in Expert Advisor

  • #21
  • Sep 21, 2018 7:59am Sep 21, 2018 7:59am
Quoting Ramana
{image}
looks it stared trading!
  • #22
  • Sep 21, 2018 8:06am Sep 21, 2018 8:06am
Updated trading chart for today:

Attached Image (click to enlarge)
Click to Enlarge Name: EU-M5-Profit.png Size: 71 KB
  • #23
  • Sep 21, 2018 8:06am Sep 21, 2018 8:06am
Updated Account Balance:

Attached Image (click to enlarge)
Click to Enlarge Name: EU-M5-Balance.png Size: 40 KB
  • #24
  • Sep 21, 2018 8:08am Sep 21, 2018 8:08am
Profit /Return : 5.23% on 2nd day
  • #25
  • Sep 21, 2018 9:24am Sep 21, 2018 9:24am
Trade Rules # 2

* Per instrument per position either buy or sell - EA is going to open up max 10 positions in regular interval (say 1 hour - these hours can also be modified, eg, you make EA trade only those hours in which you are comfortable and exclude those hours which you do not.

* EA is going to increase lot size upto 10% from 0.01 standard lot.

* EA aim for closing all positions if Total Profit is >$2

* In case Total Profit is in loss - then EA closes single positions with profit >$3, and opens fresh positions

* Each position has SL, TP 100 pips

* Trailing stops also starts on each position
  • #26
  • Sep 22, 2018 2:02am Sep 22, 2018 2:02am
Quoting rekon67
Profit /Return : 5.23% on 2nd day
Cool, I see the same progress with AUDUSD
But I got margin call backtesting this strategy with 2018 data - so EA doesn't close all positions with total loss >$3
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 217 KB
Just an opinion. Based on research
  • #27
  • Sep 22, 2018 9:14am Sep 22, 2018 9:14am
Quoting since80808
{quote} Cool, I see the same progress with AUDUSD But I got margin call backtesting this strategy with 2018 data - so EA doesn't close all positions with total loss >$3 {image}
Back testing is not a proven approach. Otherwise, you would have bunch of systems working absolutely fine giving you desired profit always. But, we both know this is not the case.

I would suggest run the EA on demo account to see how it is working instead using strategy tester for back testing. I can optimize the result in back testing and give you absolutely perfect system. But in reality it would prone to fail.

I give you one pure Data Science example. Training and Testing of data. If you know about predictive modelling, you would know - model accuracy improvement is not correlated to success of the model on Test data. So you train, train, train --- improve accuracy of the prediction on training data but as soon as a new feature is introduced your model is going to struggle.
  • #28
  • Sep 22, 2018 9:33am Sep 22, 2018 9:33am
What I meant is, in Machine Learning term it is called 'Feature Extraction'. The more your model able to do quality extraction of feature the more success in predicting the future it will have, thus, we, human will get benefit out of it.

Below is a screenshot of Macro Economy data, if you know how to read it and execute it in your trading you will be in Top 10% of Fx traders. Same goes for the EA / Model - if it knows how to interpret and use the data - the ratio of unsuccessful traders would decrease till a complete new feature is getting introduced in the market.

Attached Image (click to enlarge)
Click to Enlarge Name: 1.png Size: 53 KB
  • #29
  • Sep 22, 2018 1:01pm Sep 22, 2018 1:01pm
Quoting rekon67
This thread is just a continuation of my previous work. Can we be profitable if we just enter in the market randomly without any extensive training on forex or fundamental / technical knowledge or knowing nothing about Macro/Micro economy? Disclaimer: I used to get feedback that my contents are commercial or rather close to commercial. I used to defend that saying unless you are close to me I am not going to share actual files for free. This time I would like to start the thread with a file. If you are using it - facing trouble - you are welcome...

Random walk - random trend, how you can be profitable if you follow complete uncertainty? Y(t+1)=Y(t)+error which is normally distributed with 0 expected value. Just sum all future errors - e1+e2+e3, according the low of large numbers their sum is 0. It means that your expected profit is 0*all lots sizes = 0.

With broker charges your expected profit is negative.

Simple proof you're wrong.
  • #30
  • Sep 22, 2018 9:49pm Sep 22, 2018 9:49pm
Quoting dispersion
{quote} Random walk - random trend, how you can be profitable if you follow complete uncertainty? Y(t+1)=Y(t)+error which is normally distributed with 0 expected value. Just sum all future errors - e1+e2+e3, according the low of large numbers their sum is 0. It means that your expected profit is 0*all lots sizes = 0. With broker charges your expected profit is negative. Simple proof you're wrong.
Yes, randomness with complete zero memory would be a 'chicken-walk' so totally agree on the point. Here the EA do walk with a 'short-term' memory so if it is not comfortable and losing, it changes trade direction with assumption that next trade will be in profit.
  • #31
  • Sep 23, 2018 3:30am Sep 23, 2018 3:30am
Quoting rekon67
{quote} Back testing is not a proven approach. Otherwise, you would have bunch of systems working absolutely fine giving you desired profit always. But, we both know this is not the case. I would suggest run the EA on demo account to see how it is working instead using strategy tester for back testing. I can optimize the result in back testing and give you absolutely perfect system. But in reality it would prone to fail. I give you one pure Data Science example. Training and Testing of data. If you know about predictive modelling, you would know...
Rekon, I'm math and you are right absolutely.
I just ask you about fixed SL=>$3. Why it doesn't work in the tester?
Just an opinion. Based on research
  • #32
  • Sep 23, 2018 3:49am Sep 23, 2018 3:49am
is this EA written just for pure technical content? what else feature does it have.
Profit growth tells you the trader ability.
  • #33
  • Sep 23, 2018 9:58am Sep 23, 2018 9:58am
Quoting since80808
{quote} Rekon, I'm math and you are right absolutely. I just ask you about fixed SL=>$3. Why it doesn't work in the tester?
Why it is not working in test, there could be multiple reason - the EA considers trading hours from the time of activation, before executing trade it refers the trade hours this feature does not work with past data, so with past data reference for opening a trade it may take incorrectly.

Secondly,it compares with active positions, how long those are active. Should it close the old position first - it tries to find out those answers? Which is again does not work well with past data.

If your data has any gap then also it will fail in back testing.

That is why the best way to test it by running it on a demo account - keep it live and check the result. Question why it took this trade, why it closed that trade etc. Post the finding here for betterment.
  • #34
  • Sep 23, 2018 10:04am Sep 23, 2018 10:04am
Quoting voketexpert
is this EA written just for pure technical content? what else feature does it have.
I released it with some basic feature, the original version is too complicated so I have to shut that down.

Here, it has a short term memory to execute trade - that is ofcourse not a perfect approach, but it takes simple steps to solve trading fx. I have already posted few action which it does. I will share more details later.
  • #35
  • Sep 23, 2018 10:45am Sep 23, 2018 10:45am
What I meant is, to improve the tool, we need to supervise it closely. Can it satisfy us while answering the below questions for example. If not, then we should contribute to make it enable answering these questions.

Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 33 KB
  • #36
  • Sep 23, 2018 11:42am Sep 23, 2018 11:42am
Here is the snap of live trading before market open - it worked 48 hours, closed 28 Buys and 25 Sales with 0.57 lot trading and $36.30 profit.

The positions remained open 4 sales with 0.07 lot and 7 buys with 0.07 lots. So, it is doing a lot balancing to save the equity.

Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 66 KB
  • #37
  • Sep 23, 2018 12:08pm Sep 23, 2018 12:08pm
Here is a pattern of opening and closing multiple trades by doing lot balancing along with maintaining trading hours - latest snap from trade history.

Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 70 KB
  • #38
  • Sep 24, 2018 1:27am Sep 24, 2018 1:27am
the same real time - EA doesn't close positions<=-$3
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 209 KB
Just an opinion. Based on research
  • #39
  • Sep 24, 2018 2:07am Sep 24, 2018 2:07am
Quoting since80808
the same real time - EA doesn't close positions<=-$3 {image}
EA only closes single position with >3 usd, and overall >2usd, it closes single position with loss >100 usd
  • #40
  • Sep 24, 2018 3:06am Sep 24, 2018 3:06am
Quoting rekon67
{quote} EA only closes single position with >3 usd, and overall >2usd, it closes single position with loss >100 usd
Wow! ((( mega SL with micro TP????
Just an opinion. Based on research
0 traders viewing now