Help on Developing "The MOTHER of ALL" Trading System - page 2
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Nah. I think the project needs to be done with my original thinking and set it up in the n-tier format it was planned for. There are a few shortcuts that I can do but mainly I would like to share this project with the good people in this forum and see if we can take it somewhere. I am proficient with VB and I can do the whole project by myself if it comes down to it. The DLL part could easily be bypassed with external file reading which Codeguru has written extensively about, the basic EA with MM and TP and Trailing stop routines are already present in many EA's developed by the people in this forum, the web page can also be written in PHP or any other language and this whole project is really not that complicated. Jos are you willing to help out this project and handle any one part of the development? If you are willing then let me know, we can start this project this weekend.
I can volunteer part of my services (or at least talk about partnering up). I have a basic EA that already, as part of a cycle, opens ups trades in MetaTrader based on signals received from a third-party signal provider (C2). Actually, my whole set-up is like this:
MT4 ----> Collective2 ----> MT4.
The n-layers, with a different intermediary would be an interesting, and important, twist on this.
I have coded the dlls for MetaTrader that do the signal sending and receiving (and parse XML based responses).
Nothing ventured nothing gained. Using disimilar EA trading philosophies I think would be important because every EA has its strength and weakness. For example, if all were breakout EAs or hedging EAs or simple trend system EAs then you would simply compound the weakness of that system 10x over. Therefore, EAs generating trade signals based on a different philosophy AND different TFs would seem preferrable.
Sounds like a task you're up to, Neo. Go for it!!!
COTtraderBy using disimilar EAs, I doubt you are going to get very many signals, if any. Imagaine an EA designed to peform well during trend and one designed to perform well during range. How often do you think you are going to get the same signal? On top of that, there would be 7 EAs that all need to lineup. I believe too many filters does more harm than good.
Nah. I think the project needs to be done with my original thinking and set it up in the n-tier format it was planned for. There are a few shortcuts that I can do but mainly I would like to share this project with the good people in this forum and see if we can take it somewhere. I am proficient with VB and I can do the whole project by myself if it comes down to it. The DLL part could easily be bypassed with external file reading which Codeguru has written extensively about, the basic EA with MM and TP and Trailing stop routines are already present in many EA's developed by the people in this forum, the web page can also be written in PHP or any other language and this whole project is really not that complicated. Jos are you willing to help out this project and handle any one part of the development? If you are willing then let me know, we can start this project this weekend.
As I wrote earlier I have my doubts about the assumption that many signals combined give a better signal. That's why I would start with a more simpler setup. Just to test that. So I am afraid I will only work on it, when you have statistics who show that your assumption is true. Good luck with the project.
If you know of a 60-80% profitable EA, why not just trade it?
Doing statisical inference on bad trading signals will not drag a holy grail out of the noise, you will just have more noise. If you really want to go down the datamining path why not try neural networks?
ea discussion
I agree with Craig, 60-80% profitable is a good manual trade.
Keep adding lots is not that hard... just click the button a few hundred more times. When you have 8-9 in play, you will start to wince or get an upset stomache. I would imagine though, its one of those trades that will only occur once a few times per week and requires an EA to capture. Scalping momentum so to speak. As far as I am concerned, most of the moves I trade in the marketplace occur on volitility breakouts, after a consolidation period, around news times or ema grail trade support--- all of these leave your moving average EA's as well as oscillators playing catchup. If you don't believe it, run your indicator on non farm payroll day. that ought be interesting.
trades
xxxxxxxxxxx
I can volunteer part of my services (or at least talk about partnering up). I have a basic EA that already, as part of a cycle, opens ups trades in MetaTrader based on signals received from a third-party signal provider (C2). Actually, my whole set-up is like this:
MT4 ----> Collective2 ----> MT4.
The n-layers, with a different intermediary would be an interesting, and important, twist on this.
I have coded the dlls for MetaTrader that do the signal sending and receiving (and parse XML based responses).This is great!! Thank you very much, please PM me so that we can discuss the details!!!
Guys, while I used to agree with most of you on the multiple EA situation, here is what I intend to achieve. 20 Pips a day and 100 Pips a week. I am talking about 1 trade a day or 2 trades total with a very high percentage of winning. I know that you guys think there might not be enough signals or conflicting EA's or things like that. What I intend to do is to gather info first, calculating the signals and the sl/tp of each signal, and if the trade turn out to be a winner, then I will add weight to that signal for the future. Eventually I will come out with a score of all the EA's giving signal divided by the total numbers of EA's, then if the score is high enough, we will take the trade signal.
I think there are about 50 if not 100 EA's out there based on MACD, SMA crossing, breakout, divergence, and/or many similar methods. Why do we keep on searching/programming for that "Perfect" EA by reinventing the wheel? My proposal is to use what we have, and based on statistics, we can come out with a combination of EA's that will give us an edge over this trading business. I have done a little calculation and if we just start to add one extra lot every 100PIPs of gain, capping at 50 lots, a simple investment of $10,000 with 100:1 leverage will net you over 2.0 Million in a year. I think my project will be profitable, and I am only shooting for 1 trade a day, 20 PIPs of Profit, and we can increase lots once we make more capital.
For instance, an EA that is based on Breakout, generates a signal at 2:00AM, then another EA based on SMA crossing generated a signal on 2:30AM, then another EA based on MACD Divergence generated a signal at 2:45AM, so now we have 3 EA all predicting a move, and if we have another EA based on Murray Math or Fibonacci also predicting the same move, doesn't it just confirm the move more? I mean if all we are shooting for is a simple 20 PIPS, I think we might actually get it, don't you? Also when you are manually trading, don't you take into accounts FIBS, High/Low, Murray Math, CandleStick formation, and/or 3-bar reversal and things like that? What I am doing with my EA is exactly the same thing, but just more complex in nature.
Some people might say too many indicators is not good. And I agree with you because you will overinterprete the market. But most of the time when the market is in a state of undecisiveness, my EA will probably save you from a losing trade. So by assigning a value to each EA based on their success rate, we can pretty much eliminate false signals (hopefully). I am very excited with this idea and I guess some of you might not be on the same page with me. How many of you actually use an EA to do your trades with your real accounts and trade FULL LOTS? I don't, and that is because I don't trust an EA to make that decision for me, I test them on demo accounts, but when it comes to trading real money, I just don't trust them. I guess you can say the same thing about this project, but I rather trade less and book my 20 PIPS than to have an EA that spews out 10 signals a day. Once I get it to generate 1 good signal a day, that is pretty good for me, specially I am trading 1 FULL LOT, which is $500 a Week, then next week will be $1000 Week, then the week after that will $1500, $2000, $2500, and so on. I don't want quantity, I just want quality, and I don't want to get 100 pips per trade, I just want my 20.
So what happens when the EA is not generating a signal. Man, I don't know, how about we trade the news? That is what I do most of the time, trade the news. I am always available to trade the news, and at least I will trade 20 times a month for the big news, including all interest rate decisions, CPI, PPI, GDP, and all of those indicator releases. But having an EA that will trade for you when you are not trading the news is very very convenient, and I would rather trust this EA, than any other EA out there. Just my 2 pennies...
Ok, enough ranting. I am still looking for people to help out with this project. Please cast your doubts away. No pain No gain. You are participating in an event that might change the whole conception and future outlook of automated trading. Why not be a part of it than to be left behind when we actually get somewhere with this project?
Thank you.
If you know of a 60-80% profitable EA, why not just trade it? Doing statisical inference on bad trading signals will not drag a holy grail out of the noise, you will just have more noise. If you really want to go down the datamining path why not try neural networks?
Craig:
I re-read your post again and I have to say that you got a point, but what I am doing here is not datamining, nor am I trying to drag a holy grail out of the "noises" as you put it. I am simply determining the relationship of different school of thoughts which the EA's are based on, then trading when the signals agree with each other. If anything, this EA will probably stop you from trading during the "noisy" hours.
I think neuralnetworks is good, but again I would be reinventing the wheels, why not use the EA's which are already successful and go from there? I have seen so many people who complaint over and over again on what certain EAs will work on certain market conditions, but there are really no way to determine what market condition will which EA work the best. Some people use time filters, some people use News release filters, and some people just trade a certain day of the week. What some people are trying to do is to do a backtest and cut out the losers by adjusting stop loss, time trades, and any other factors, which means that people aren't seeing the whole picture. Is like this story that I have heard of a bunch of blind men trying to feel an elephant, and each one will tell you a different impression on what an elephant is, but none of them is right. My project will put all the parts together and more. It will not only be limited to EA's, that is just the first phase of the project, eventually we can tap into long term fundamental trading, signal strength, market sentiment, and so forth. By having a system to deliver such signals, we have to infrastructure to do the impossible. I think when we finally do get there, that is and will be the "HOLY GRAIL" of trading.