Thread: Statistically testing every indicator with my Genetic Algoritm

Yup like the name suggests, I'will test each major indior in the MT4 platform, also you can submit custom made ones (preferably those that give BUY/SELL signs and can be mechanical, not things like pivot points or similar to that).

Mission

I've build in the previous 4 months a intricate Artificial igence that can exchange upon indiors profitably, plus it can assess and compare the every indior's validity.Now I'm not going to place my AI in public, nor some information how it created it,but will use it to assist you folks out who are using indiors for trading.Certain myths regarding indiors must vanish, a few of you say indior are BS and cant make constant profit with it, a few of you say they are the Holy Grail, and can make huge profit.

I say, you can make money by investing them mechanically, however, you can't utilize them statically, which means that you cant use the very same parameters like in the MACD 12,26,9 that were usen in 1970, the market changed much since then.

Strategy

My genetic algorithm, segments price into clusters randomly, and calculates the best parameters to the bunch, then after new clusters/ information comes, it learns out of those too, and tries to adapt to the best possible parameters, to the maximum fitting enviroment.

It's pretty inteligent at doing so.

So I will compare these indiors, by their formula not by their parameters, so dont ask me parameters/ settings of the indiors that are the best, because there is not any such thing, in different states, different parameters are great,but I will inform you atleast the minimal parameters which must be utilised to be atleast breakeven, however you still must optimize the indior after each transaction or bunch of trades.

Its not like it is possible to utilize MACD 12,26,9 in 2002 trending market with same succes like in 2010 ranging market.Thats nonsense.

But what I'm going to reveal is that though parameters can vary, specific indiors will still be overall better than many others.

Due to their formula, guesses the price flow more accurately, together with no parameter, better than other indiors.

So I might make a top 10 list after we're done, but lets only test them.

Also not that Drawdown and things like these are not quantified, so it may be that particular indiors are more dangerous than others, but its generally not higher than 25 percent, but if it's then indior has a rather hart time calling the price and will probably not be profitable anyway.

Criteria

Simple, 2 numbers are crucial here, I will measure each indior, just how much pips/pipettes can it acumulate in the previous 12 decades of time,by opening a transaction at every signal and shutting it at opposite. SL and no more TP will be used, so Drawdown isn't measured!
Second, I will assess the accuracy rate, how accurately predicts price flow.

I will use the EUR/USD just, and M30 timeframe.I have 12 decades of rather accurate historic statistics on M30.

Also I picked M30 because I discovered that you cant make money in the long term on TF's under 30 with those indiors,or atleast you would be quite limited.None of the MT4 indiors can filter out sound under M30 effectively.

Today it may be that particular indiors will perform much better on USD/JPY or even AUD/USD, especially the trend spotters.

But I believe the EUR/USD is nonetheless a fantastic benchmark, and since many men and women exchange it, I will test it just here.Also if I were to test it in each pair that will take forever, and it takes much computing power to get this done, so EUR /USD just

1 other number will function as nr of transactions, the sample dimensions.

Why?

Why i do it ? Simple, its free, and I want to help you out men that are still beginners and have doubts regarding indiors.Also I dont spoil anything, I have far better indiors compared to classical ones, however those who just have classical MT4 indiors, this is a superb performance analysis of them

TOP 10 INDICATORS OF ALL TIME:
1) Parabolic SAR (PSAR)
two ) Single Simple MA
3) Bullpower
4) Bearpower
5) Relative Vigor Index (RVI)
6) Simple MA Crosses
7) Ichimoku (Kijun/Tenkan cross)
8) Relative Stength Index (RSI)
9) Moving Average Convergence Divergence (MACD)
10) Money Flow Index (MFI)

Outcomes on H4:
https://www.forexforum.co.za/general...d-thought.html

Ok lets start

1) MACD =Moving average convergence divergence (mains egy histogram above/below 0 to provide signs )

Pair: EURUSD
Timeframe: M30
Years tested: 12
Internet pipettes acumulated (win-lose): 26024
Total transactions: 461
Truth rate: 32.537%
Minimum parameters/price cluster: Fast and Slow period has to be atleast 100 or more to become atleast breakeven based on market structure, anything beneath that won't be profitable in extended duration, and has to change parameters to fit ranging and trending markets, Signal period doesnt matter with this egy
(12,26,9 is a fantasy )
Profitable: Yes

two ) MACD =Moving average convergence divergence (secondary egy with sign lineup above/below 0)

Sorry I cant test that, 3 parameters are overly much, I dont have enough CP to complete it at realistic time.It would take weeks to compute that, too many permutations

The trade sample could be contested, but thats all we've got, and we span trought 12 decades therefore, atleast we tested it in each condition.

Originally Posted by ;

Its not like it is possible to use MACD 12,26,9 in 2002 trending market with same succes like in 2010 ranging market.Thats nonsense.

Every kind of market can be found during every period in time. Flip timeframes to acquire the market you would like. Just remember that as you go down available pips per swing have a tendency to go down.

Originally Posted by ;

quote Every type of market can be found during each period in time. Flip timeframes to acquire the market you desire. Just keep in mind that as you go down available pips each swing tend to go down.

I cant go lower than m30 anything lower than that is only overly noisy.And if I move higher, the sample size becomes too small, and the transaction duration, months/years. Thats not a very fast Method of making money, better put your money in bank then

Count me like your strategy. .

Originally Posted by ;

Count me like your approach. .

Sure, after I finish with the fundamental MT4 indiors you can submit custom made ones too.But preferably in .mq4 format, because I cant connect the .ex4 documents for my algo.

And preferably those that have 1 or two parameters, anything more than that could take days/weeks/years to calculate, and likely be overly complied to even fit the market.

Ongoing...

3) RSI= Relative strength index (Above 30 BUY / Below 70 SELL)

Pair: EURUSD
Timeframe: M30
Years Analyzed: 12
Net pipettes acumulated (win-lose): 35332
Complete Transactions: Two (yep only 2 year long Transactions, 1 won 1 Missing )
Truth rate: 50%
Minimum parameters/price cluster: Not really good in this TF (14 setting is a Fantasy ), Greater timeframes advised, unless You Would like to Maintain a Transaction for 4 years Available...
Profitable: Yes
Comments: Although it gave Greater profits than the MACD, it only had 2 Transactions, I tested it higher timeframe where it gave Fairly good results, also in Different pairs it was Okay, but since we test it upon EUR/USD M30, I will discard this from the TOP 10 Due to This lack of data on M30 timeframe

4) Bullpower (buy above 0/ Market below 0)

Pair: EURUSD
Timeframe: M30
Years tested: 12
Web pipettes acumulated (win-lose): 21337
Total trades: 3056
Accuracy rate: 23.52
Minimum parameters/price Bunch: Top period is Required, but it can make profit on M30
Profitable: Yes
Remarks: High periods needed Such as on the RSI, but it has Far more trades than the RSI, although its accuracy rate is Reduced, I Believe it more reliable than the RSI on EURUSD M30

5) Bearpower (buy above 0/ Market under 0)

Pair: EURUSD
Timeframe: M30
Years tested: 12
Web pipettes acumulated (win-lose): 29986
Total trades: 2796
Accuracy rate: 22.46
Minimum parameters/price Bunch: Top period is Required, but it can make profit on M30
Profitable: Yes
Remarks: Less accurate then Bullpower, but it Created more profit, Using bit longer but fewer Poor Transactions

Thought I was interested of triumph rates of those in constructed indiors if it is possible, can you examine them on Daily TF as well as When analyzing Moving Averages evaluation with periods of 5, 10, 20, 50, 100 and 200. . .exponential would be ok. For Moving Average Crossovers evaluation groups of (5, 10); (10, 20) and (50, 200) I'm a silent lover of Exponential Moving Averages but employed them a little

Originally Posted by ;

Excellent idea I was always interested of triumph rates of these in built indiors Also if it is possible, can you examine them on Daily TF as well as When analyzing Moving Averages evaluation with spans of 5, 10, 20, 50, 100 and 200. . .exponential would be ok. For Moving Average Crossovers evaluation groups of (5, 10); (10, 20) and (50, 200) I am a silent fan of Exponential Moving Averages but used them a tiny

I shall leave the moving averages to the last, because there'll be a number of combinations there.First I examine the Oscillators group.

I pick the M30 because it is the smallest greatest TF.No doubt the Daily and higher timeframes will make each indior profitable, but the challenge is the profit will be slow.

Can you hold a commerce open for 4 weeks or like that? I dont believe so, month, because its better to have 100 small transactions in 4 then 1 huge commerce, because you're able to compound the profit and create more.

If each trade in average will give 10-15 pips and'd 1000 transactions in a year would create atleast 1000x more profit, compared to 15000 pips in 1 commerce which lasts the entire year.

Why ? Because you increase the lot size in the meantime because the equity grows.While if you had only 1 commerce, the swaps will probably eat all profit, and you also had to worry the whole year for that trade. . .Bad idea.

On Daily TF its easy to trade, but we need quick profit right