Here is an idea...

[Astarfleir]

Instead of trying to find the best EAs around, would it be easier to use existing and proven bad EAs and use them with an order reversal? Like > order instead of < order.

This way a very bad EA becomes a very good EA!

Am I correct?

[soundfx]

Astarfleir,

 

The theory sounds good, though in practice it rarely works.

 

A really bad scalping EA which is always a net loser, with for example 20 pip targets and 500 pip stops could potentially be turned a reasonable long term trading EA - however EA's aren't really needed for such longer term trades and there are likely to be much better manual systems around.

 

Some scalping EA's can end also up being losers simply because of poor spreads and slippage so changing the signals around for these won't make much difference, however changing brokers in some cases may help.

 

The performance of EA's isn't totally about overall accuracy. We can't just say that because a particular EA is 30% accurate and a loser, that when we switch the signals it will be become 70% accurate and a winner.

 

Accuracy of trades depends on the risk/reward ratio. This may be fixed or may be variable depending on price action. For a fixed R/R system like the one I mentioned with 20 pip targets and 500 pip stop (i.e. similar to the original Forex Autopilot which was the predecessor to FAPTurbo), then accuracy will need to be very high for the EA to even have a chance of being a winner over the long term.

 

For example, two losses in a row will mean 1,000 pips down the drain and this will take 50 successful trades before the EA becomes into profit again.

 

If a scalping system with these parameters was 95% accurate then we have 19/20 winners, i.e. (19 * 20) = 380 and (1 * -500) = -500 giving a net loss of -120 pips...and that's at 95 % accuracy!

 

Other EAs may use different algorithms such as the "stacking" or martingale type EAs. This sort of EA checks the current trend and then when price moves against the trend it starts to enter lots of trades in the direction of the trend along what the EA expects to be a retracement. These trades go into drawdown until the trend is resumed, where one by one, they start to come into profit, the EA can then close trades at a particular target one by one or all can be closed at the same time when a certain net profit is realised.

 

If a given market is generally trending over a few months then these EA's can perform great, however when the market dramatically reverses the trend, these EAs are caught on the wrong side of the trend with multiple trades which just get to be bigger and bigger losers.

 

Look at an EA like this in one of the rarer losing periods and you may assume that by changing the signals around that you'll get terrific performance - in fact you'll get much worse performance, because you'll end up trading against the trend all the time apart from when it reverses.

 

I'm sure that there are many more examples which other folks can give of where things aren't as simple as just changing the signals around.

 

As a general rule, if an EA has very simple entry and exit criteria (e.g. 1 or 2 indicators) then switching entry signals may help and is worth a try, but for anything else you'll end up with similar results or much worse results.

[Astarfleir]

Thank you for your reply. Would a 50/50 Take profit/Stop loss EA reversed would work in that line of thought? Let say I live test an EA that has a take profit of 10 pips and a Stop loss of 10 pips but that EA has a very bad track record, would that EA be profitable if the coding is changed from < to > for example?

[soundfx]

No probs. - the answer to your question is "possibly", however ideally the EA which you want to reverse the signals on needs to have a very poor track record - I would say 30% accuracy or worse (and how many EA's have you seen which are that bad?).

 

When we have risk/reward of 1:1 then at least we're starting with a level playing field which is the best scenario to think about reversing entry signals as you rightly point out.

 

Markets change however, and we start to move into the realms of longer term market volatility and the testing period of the EA.

 

The average daily High-Low range of any currency pair you observe will expand and contract in different years and this is why I mention 30% accuracy or worse as being a good starting point for thinking about reversing signals. This is because high average volatility in one particular year could show an EA performing at a very poor 40% accuracy with 10 pip stops and targets, however in years of much lower average volatility the same EA could be rolling along at around 60% accuracy (mainly because the stops get blown less often).

 

If you choose the wrong year in which to reverse the signals based on previous performance, you could easily come unstuck. At least if you start with looking at current market conditions and reverse a very poor EA's signals turning your current 30% accuracy into 70% accuracy then you'll be able to take some good profits before the overall market conditions change for the worse.

 

I've often thought of reversing signals on really poorly performing systems, however when I've tried it out and looked at the results I've often been staggered to find that reversing the signals of a 30% accurate system gives a fractional increase in performance and typically ends up in still being a losing system. Here the issue tends to be the price action basis on which the entry points were specified in the first place - in other words, what should in theory be a great entry point for a Long trade and isn't doesn't necessarily mean it that it will automatically be a good entry point for a Short trade.

 

For example, it could be that the EA is trying to target Long 10 pips without any clear sign that the market is going to move Long that much any time soon. If the EA runs 24 hours each trading day then there will be several patches of very low volatility which it will have to negotiate. If the entry point is flawed in the first place and trades are entered around the middle of low volatility ranging periods, then it doesn't really matter what the overall trend is on the longer term timeframes and whether we're trading with the trend or not, the "noise" can take out our trades on either side, so switching the signal can turn a 30% accuracy system miraculously into a 30% accuracy system lol.

Edited April 3, 2010 by soundfx

[jdonut]

I have tried that. It doesn't often work. In other words, you reverse the Op_Buy and Op_Sells with each other. Lots of times it just doesn't trade and then when it does trade and you expect it to do the opposite, it might do the opposite but it usually ends up doing that wrong also. I must have tested 200 ea's or more. They all fail. All of them. If they don't fail, the profits are very small. I am trying to make an ea but it is not easy to understand the language for me. I need to know meanings for things and a ) or * doesn't mean much to me. But honestly there are much smarter people than me and - no one has made an ea that consistently wins 6 out of 10 with a moderate profit for a years back test. Something changes in the market. I can get my ea to make $1600 on a 500 depo in a month - January 09 (Back test) by the time it hits May it has less than when I started. Now, a moving average or a fractal...is. The moving averages or fractals or whatever need to be - dynamic - changing with conditions. I am waiting for some sort of dynamic moving average or something. I haven't been able to figure it out yet. I don't think this idea works because the ea was written in one way and even with a signal reverse button, it will most likely fail anyway. I wish you luck finding that which you seek.

[Astarfleir]

Thank you for your input. It was just an idea. I have not give up on EAs yet. I look at charts and all I see are bars. I am sure there has to be an EA with a very conservative Martingale system that would work in every market conditions. No trends, no indicators, no formulas, just positions taken based on tick movements up and down (like the EURCHF) except today! looks like the perfect pair to trade a Martingale EA on it. Not an EA that takes a new positions every certain number of pips but more like positions adapted to the speed of the price movement. Anyways the search continues!

[FXmini]

Melihat hasil pembicaraan para pakar-pakar forex diatas, maka saya tergelitik untuk memberikan sebuah komentar tentang Expert Advisor itu.

Namun sebelumnya saya mohon maaf atas penggunaan bahasa Indonesia ini..

Nah selama bertahun-tahun saya selalu melakukan riset demi riset baik itu indicator, system perdagangan forex/index bahkan kini saya juga sudah melangkah untuk melakukan project pembuatan EA, maka sampai dengan hari ini saya berkesimpulan yang pada dasarnya sama dengan ulasan Mr.SoundFX diatas..

EA itu sebenarnya adalah hanya sebuah mesin yang telah kita programkan berdasarkan historical data bukan mengacu pada data yang akan datang.. nah seperti ulasan diatas, maka kadang sebuah EA yang bagus pada tahun 2009 akan menjadi sebuah EA Malapetaka pada awal-awal bulan 2010 ini mengapa..?

Karena EA yang mungkin bekerja baik pada tahun 2009, tidak dapat menangkal atau tidak mempunyai Artificial Intelegence untuk mengantisipasi kejadian pergerakan harga pada tahun 2010

Saat ini pun saya sedang mengerjakan sebuah project EA yang seupayanya dapat berjalan stabil, namun hal tersebut sangatlah sulit untuk dapat bertahan jika tiba-tiba pasar melakukan gelombang besar.

Apa yang terjadi.. EA tersebut menjadi tidak stabil dan akhirnya dapat membahayakan modal kerja kita.

menurut saya EA yang baik adalah diri anda sendiri.. nah anda harus menyediakan beberapa jenis EA yang harus anda aktifkan jika anda melihat kecendrungan pasar akan bergerak kemana.

Oke pertanyaannya adalah Bagaimana kita melihat kecendrungan pasar akan bergerak kemana..?

jawabannya mungkin kita dapat melihat dengan menggunakan pola pattern, candlestick pattern, fibonachi retreachment.. dan lain-lainnya..

Edited April 3, 2010 by FXmini

[FXmini]

Agar kita bisa sukses menggunakan EA maka saran saya sebaiknya jangan menggunakan hanya 1 buah EA saja tetapi 3 buah EA dengan karakteristik yang beda2 yaiu.. EA yang mengacu pada Trend market (Up / Down EA) lalu EA yang kedua harus mengacu pada Sideways EA dan terakhir kita dapat menggunakan EA yang bersifat Oscilator mingguan..

 

Saya sendiri dalam menggunakan EA kadang masih dibantu dengan manual trading khususnya saya gunakan crossing 2 buah Moving Averages hanya untuk melihat apakah dalam pola dayli saya harus menggunakan EA yang mana..

Misalkan Trend sementara turun, maka saya gunakan EA yang mengacu pada Trend Setter.." Sell " terus dalam beberapa hari.. lalu sesuai sifat market bilamana harga masih dalam rentang trend turun maka EA tersebut saya tetap aktifkan..

Namun sesuai sifat market ... setelah penurunan yang cukup dalam selama beberapa hari. maka sudah menjadi sifat market pada umumnya pasti akan terjadi konsolidasi panjang.. oke.. pada saat itu saya non fungsikan EA Trend itu lalu saya aktifkan EA yang bersifat Sideways..

Kemudian jika ternyata market masih mengalami masa tak menentu maka saya gunakan EA yang bersifat Oscilator untuk memantau.. tetapi bukan untuk melakukan open posisi.. guna EA terakhir ini aga saya dapat mengetahui relasi antara MA manual trading dengan posisi EA oscilator itu sudah sejalan atau belum.

 

Nah sebagai contoh EA oscilator itu dapat menggunakan EA yang berbasis RSI atau CCI namun EA tersebut jangan di set parameternya untuk bilangan TF menit2an.. karena akan menimbulkan cita rasa fals.. tetapi set lah oscilator itu dengan TF harian atau mingguan.. disitu kita bisa melihat apakah open pertama dari EA Oscilator itu sdh sesuai dengan manual trading kita atau malah mengambil posisi oposisi..

 

Oke sampai disini dulu ya..