Statistically testing every indicator with my Genetic Algoritm | Page 5 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

codification of the market / Genetic Algorithms 36 replies

S/R Area Algoritm 17 replies

What Constitutes a Statistically Significant Sample for Back Testing? 4 replies

Has anyone had luck with high frequency using either neural nets or genetic Pr? 3 replies

Genetic Algorithm for Metatrader 1 reply

Attachments: Statistically testing every indicator with my Genetic Algoritm
Exit Attachments

Statistically testing every indicator with my Genetic Algoritm

  • #81
  • Jun 16, 2014 1:42am Jun 16, 2014 1:42am
Quoting Macdon
{quote} Hi Proximus. What if you do what you did with the 30M TF on the 4HR TF? This is more a medium TF and seems to be used more and more by traders. I'd be really keen to see those results and I'm sure others here would too. Also, can't you increase the lot size as your equity grows with that 1 yearly trade of 15,000 PIPS? You could buy another lot every 1,000 PIPS or so. Thanks.
Of course its used by many traders, in fact i bet most profitable traders dont even touch lower timeframes than H4. But if you are scalping like me then you need profitable timeframe below H4, on H4 it seems that you cant make more than 15%/month which is lame.On M30 however it is possible to make 30-40% which is better, but trading is much harder on M30 because most of the markets are incredibly choppy at that level.

I could add that into the formula, yes, maybe this week i will test H4 aswell, seems like indicators can adapt better there
"There's a sucker born every minute" - P.T. Barnum
  • #82
  • Jun 16, 2014 2:36am Jun 16, 2014 2:36am
Quoting Proximus
{quote} Of course its used by many traders, in fact i bet most profitable traders dont even touch lower timeframes than H4. But if you are scalping like me then you need profitable timeframe below H4, on H4 it seems that you cant make more than 15%/month which is lame.On M30 however it is possible to make 30-40% which is better, but trading is much harder on M30 because most of the markets are incredibly choppy at that level. I could add that into the formula, yes, maybe this week i will test H4 aswell, seems like indicators can adapt better there...
I understand what you're saying. You can do more trades on a lower TF and thus have a greater profit on turnover, but 15% per month is a good POT too!

Thanks.
  • #83
  • Jun 16, 2014 4:39am Jun 16, 2014 4:39am
Quoting Macdon
{quote} I understand what you're saying. You can do more trades on a lower TF and thus have a greater profit on turnover, but 15% per month is a good POT too! Thanks.
15%/ month is a great achievement, do doubt, but not for me, i have a higher expectation

I have a pretty good understanding of the markets and experience of it to make more, however if i`d have a hedgefund or a managed account ,then it would be enought because external investors can invest into it, thus the profit would increase also if the fund would be popular.

If i could make 15%/ month consistently i will open a hedgefund, however for personal trading i can make better

For 6% risk, once i made 200%/day for a week.
"There's a sucker born every minute" - P.T. Barnum
  • #84
  • Edited 6:44am Jun 16, 2014 5:01am | Edited 6:44am
Baboon popped the balloon. All the baboons were crying. The great baboon was sad afterwards, when others left without reasoning.
The great baboon and his adventures..

I saw a guy repeating the same fooking bullshit once. Later he found out that he was the only one who believed his own bullshit.

Saying stuff above in this matter makes you ridiculus. That at least.

Ill go now.
  • #85
  • Jun 16, 2014 9:07am Jun 16, 2014 9:07am
Anyone who followed my previous thread where i constructed a gambling system:

http://www.forexfactory.com/showthread.php?t=467309

Should know that i mostly design systems, and indicators, i also trade manually but i dont like it, i`m more into mechanical trading since i dont trust in the human nature of emotions and hesitation while trading.

So based on that indicator there, the APROXIMATOR V1.0.mq4

I was working on a better version of it, and a much simplier one, combining elements with the one Holy Grail indicator of mine which was used by eitan1010 in his thread Eitan's Project.

It was a succesful aproach, so its time to update the indicator.

Maybe today, but probably tomorrow, i will upload here the new version of the indicator which i`ve been working on as a side project since the last week.And let's see how well it will work in the market.
"There's a sucker born every minute" - P.T. Barnum
  • #86
  • Jun 16, 2014 11:04am Jun 16, 2014 11:04am
Good read, thanks for your work on this.
  • #87
  • Jun 17, 2014 2:43am Jun 17, 2014 2:43am
Quoting Proximus
{quote} Sorry for the late response, i was busy this week. Lets see, first i thought it shows that buggy because i dont have volume data on my historical data, but then i tested it on HKD/JPY with fresh downloaded data with volume info and it was still buggy.I have build 646 , no idea why it is bugged like that... {quote} Yes it can be helpful, but the technical analysis should not be then only analysis.Use fundamentals, and sentiment analysis too, it may improve your edge.
Hmm,

Try this one, other source. I'm looking forward to 4h tests this is my favourite perspective
Attached File(s)
File Type: mq4 Trend Magic.mq4   2 KB | 394 downloads
  • #88
  • Jun 17, 2014 9:26am Jun 17, 2014 9:26am
Quoting Proximus
Anyone who followed my previous thread where i constructed a gambling system: http://www.forexfactory.com/showthread.php?t=467309 Should know that i mostly design systems, and indicators, i also trade manually but i dont like it, i`m more into mechanical trading since i dont trust in the human nature of emotions and hesitation while trading. So based on that indicator there, the APROXIMATOR V1.0.mq4 I was working on a better version of it, and a...
Here is the thread opened for this indicator as promised:

http://www.forexfactory.com/showthre...60#post7542160

In the meantime i will try to make time to test those indicators in H4, but at the moment i`m busy so please forgive me
"There's a sucker born every minute" - P.T. Barnum
  • #89
  • Jun 27, 2014 11:08am Jun 27, 2014 11:08am
Quoting Szpons
{quote} Hmm, Try this one, other source. I'm looking forward to 4h tests this is my favourite perspective {file}
Still it is bugged for me, i dont know why i have the latest MT4.


Anyway i got some free time this weekend so i will test the H4 versions of each indicator.I could also the H1 but from M30 to H1 there is not much difference, but to H4 it is.Besides i`ve looked across many threads on the forum and looks like every trader trades on H4 nowadays, you can probably spot the best trends on this timeframe.So lets do it!
"There's a sucker born every minute" - P.T. Barnum
  • #90
  • Jun 28, 2014 1:25am Jun 28, 2014 1:25am
Quoting Proximus
{quote} 15%/ month is a great achievement, do doubt, but not for me, i have a higher expectation I have a pretty good understanding of the markets and experience of it to make more, however if i`d have a hedgefund or a managed account ,then it would be enought because external investors can invest into it, thus the profit would increase also if the fund would be popular. If i could make 15%/ month consistently i will open a hedgefund, however for personal trading i can make better For 6% risk, once i made 200%/day for a week.

I just happen to stumble across this thread and read all post up to this point. Maybe the successful people just gather information and then leave silently.
While the "newbies that do post on this thread" have zero clue on what to learn from an experience trader Proximus.

What I "gather" from this thread:

#1 Proximus have an EA with "Artificial Intelligent" to change indicator's parameter for best curve fit win result.

#2 The indicators were tested in a 12 year period with a range of "24-50%" accuracy signal result AND it profit..
Which mean that mean there you can win with such terrible accuracy.

#3 Some indicator have "49% accuracy" and 12 year result a LOSS.
While some indicator have "30% accuracy" and 12 year result a WIN.

#4 The top indicator on 30 min time frame only make 3604 trade in "12 years" .
That mean 12year / 3604 trade => 300 trade per year and there is 260 trading day in 1 years. So you pretty much "have 1 trade at least each day on average."

#5 Even with all the skill Proximus has, he self-claim that he can not make "15% per month consistently".

#6 The Final Verdict for the "newbie" that see this thread. After reading everything in this thread you should have come to conclusion that "Indicator are just tools. It's how you use the tool to make you profit. Proximus use HIS EA to use the "indicator at the right time with the right parameter to adapt to the market so it win..."
  • #91
  • Jun 28, 2014 1:55am Jun 28, 2014 1:55am
Quoting victor_ray
{quote}After reading everything in this thread you should have come to conclusion that "Indicator are just tools. It's how you use the tool to make you profit.
This is the worthwhile conclusion of the thread. I would expect that people have come here not to hear this type of answer, however.
"Serenity, you sound like a frustrated Canadian living your life in the US"
  • #92
  • Jun 28, 2014 2:20am Jun 28, 2014 2:20am
Is it possible to test the indicators that performed the worst in the opposite direction? ie sell when the indicator gives a buy, and buy when the indicator gives a sell signal?
  • #93
  • Jun 28, 2014 5:16am Jun 28, 2014 5:16am
Quoting victor_ray
{quote} I just happen to stumble across this thread and read all post up to this point. Maybe the successful people just gather information and then leave silently. While the "newbies that do post on this thread" have zero clue on what to learn from an experience trader Proximus. What I "gather" from this thread: #1 Proximus have an EA with "Artificial Intelligent" to change indicator's parameter for best curve fit win result. #2 The indicators were tested in a 12 year period with a range of "24-50%" accuracy signal result AND it profit.. Which...
I think you should re-read the thread.
  • #94
  • Jun 28, 2014 5:47am Jun 28, 2014 5:47am
Quoting victor_ray
{quote} I just happen to stumble across this thread and read all post up to this point. Maybe the successful people just gather information and then leave silently. While the "newbies that do post on this thread" have zero clue on what to learn from an experience trader Proximus. What I "gather" from this thread: #1 Proximus have an EA with "Artificial Intelligent" to change indicator's parameter for best curve fit win result. #2 The indicators were tested in a 12 year period with a range of "24-50%" accuracy signal result AND it profit.. Which...
1) Similar but not exactly, curve fitting as it is wont really help you in any way because the market changes too much.It is actually curve fitting + natural selection type algoritm to select the best curve.

2) & 3) Yes strange but true, the less accurate is the most efficient,while the most accurate is the less eficient.It is probably because of the fact of the Reward>Risk phenomena, where probably all 2:1 ,3:1 systems could be profitable and leaving the "1 pipper" or in other words the 1-5 pip takeprofit and say 500 pip stoploss or worse no stoploss systems in shame.
The 2:1 RR myth is actually true

4) Yea, if you look on a M30 tf, you probably need 6-7 bars to trade, i dont think you can scalp m5 bars profitably, atleast not on retail platforms not.

5) Not ...yet

6) Yes thats right, a good trader can make the worst indicator the profitablest one just by eliminating the most false signals, in the same way the worst trader can make the best indicator useless.It's relative, however this was an objective study of mine, and because it was tested like this, it probably only matters for EA traders or mechanical traders, for discretionary people i dont think they even use indicators.
"There's a sucker born every minute" - P.T. Barnum
  • #95
  • Jun 28, 2014 5:48am Jun 28, 2014 5:48am
Quoting lappies
Is it possible to test the indicators that performed the worst in the opposite direction? ie sell when the indicator gives a buy, and buy when the indicator gives a sell signal?
Sorry to say, but thats just nonsense.There is a whole thread here about this aproach:

http://www.forexfactory.com/showthre...97#post7552497
"There's a sucker born every minute" - P.T. Barnum
  • #96
  • Jun 28, 2014 5:57am Jun 28, 2014 5:57am
Hello

Interesting thread, thank you.
I don't understand the result you get with RSI, the RSI(60) and below cross many times the 30-70 level in M30, on EURUSD
how is it possible to get 2 crosses in 12 years ?

thank you
No more Fight, the victory has come...
  • #97
  • Jun 28, 2014 6:08am Jun 28, 2014 6:08am
Quoting nonor
Hello Interesting thread, thank you. I don't understand the result you get with RSI, the RSI(60) and below cross many times the 30-70 level in M30, on EURUSD how is it possible to get 2 crosses in 12 years ? thank you
Of course it crosses many times, but none of those are profitable, or atleast not with my criteria or atleast underperforming.I buy/sell at a signal and close it at the opposite, no stoploss and no tp is used.
The results will be shuffled by selecting the most profitable.Now this doenst mean that you cant use SL and TP, it is better to use it, but it is only done this way to compare the performances of indicators.

So the fact that your RSI 60 is crossing many times the 30/70 means that that setting is probably not the most efficient one,not necessarly a loser but just not effective since if it would be effective then it would have been selected by my AI.

The RSI has a pretty strange formula and probably wasnt designed for EUR/USD, try using it in JPY pairs or in trending pairs.
"There's a sucker born every minute" - P.T. Barnum
  • #98
  • Jun 29, 2014 3:42am Jun 29, 2014 3:42am
Ok, thank you.
It's a fact that RSI is well used in many manual and automated systems and your results goes against the common belief.
is it possible to test the two following variation of RSI ?

Thank you
Attached File(s)
File Type: mq4 RSI_Figurelli RSI.mq4   3 KB | 787 downloads
File Type: mq4 RSI_RapidRSI.mq4   3 KB | 656 downloads
No more Fight, the victory has come...
  • #99
  • Jun 29, 2014 10:19pm Jun 29, 2014 10:19pm
Quoting nonor
Ok, thank you. It's a fact that RSI is well used in many manual and automated systems and your results goes against the common belief. is it possible to test the two following variation of RSI ? Thank you {file} {file}
I dont know how well its used, and maybe the discretionary element plays a huge role in diferentiating the loser from the winner, however my objective analysis says this.Numbers dont lie.

I will test those 2 also, but my analysis says that the RSI isnt really made for the EUR/USD, could be that those "succesful" guys use it on other pairs or in higher timeframes?
"There's a sucker born every minute" - P.T. Barnum
  • #100
  • Jun 30, 2014 6:31am Jun 30, 2014 6:31am
Any thoughts about finding the best parameters for the best performing indicators for the past 3 months, then repeating it monthly (for the past 3 months) and always trading with the actual "optimized", i.e. best performing settings...?
0 traders viewing now