There are "ea"s and there are "EA"s.
Well said!
Excellent comment Mr CB :)
All that knowledge and you can fly helis - good grief!
only want profits, don't need all that other junk!
Precisely! Who on Earth needs a reliable, predictable EA that doesn't go off on one if it reinitializes with the wrong values? Hey its only money and reputation.
CB
I've posted this in response to the latest post which suggests that the going price for an EA is $20.
I'm frankly staggered at the number of new members on this forum who either can't describe the precise specification of their strategy, or who can't cut code, together with those who can do neither, but yet will quite happily say this EA is simple, this EA will take an hour, half an hour, whatever.
The truth of the matter is that in order to size the effort required you must be in possession of (and know how to develop) ALL the requirements. And by that I mean:
- The functional requirements - WHAT the EA must do
- And the non-functional requirements - HOW the EA will do it
To those who say - "yea it should take half an hour", I'd ask:
- What about the effort involved in agreeing user requirements?
- What about the need for a persistence layer so your EA can recover following an outage and continue trading?
- What about the flexibility to have a single code-base for multiple pairs?
- What about parameterisation of all the items that the user may have to vary
- What about a TEST mode which makes the EA "strategy-tester friendly"
- What about the flexibility to perform money-management on multiple charts and multiple base currencies with one codebase?
- What about journalling, and email reporting to users?
- What about a services approach to the architecture - designing in modularity from the start?
- What about commenting, user instructions, release notes, data dictionary and version control?
- What about unit-testing?
Only when you define ALL the requirements can you properly estimate the effort required.
So all of you "programmers" who turn out a few functions from your Half Bakery and call it a complete EA - please recognise the difference, and be a bit more up-front. Between you guys charging very little (and delivering very little) and the idiots who suddenly appear with a half-thought-through strategy and declare (from a position of complete ignorance) that the EA will be simple and will only cost xyz, both of you are devaluing the work of those experienced programmers who take the time to develop a professional solution to a well-understood set of requirements.
CB
So, an average MT4 user who has no programming skills has come up with an idea (in your word: an idiot) wants someone with programming skills to write a simple prototype EA to test his idea for $20, and he is not interested in those super fancy fault tolerance (For safety) or journalling, email report to users (For comfort).. And your answer obviously is: All the bells and whistles are come with standard package and it is gonna cost you $1000 for me write something like MA crossover EA...
No wonder GM has gone into bankruptcy because as their CEO acknowledged: They didn't build a car that a buyer wants... Sounds familiar ??
MigMof
Very Very well said....
And there are also some "ea's" that with some Backtesting looks like an "EA" and these "programers" want to sell it for 10000eur :)
migmof wrote >>This reminds me an ad on TV about a person wants to buy simple push bike with a reasonable price, but the manufacturer insists to build a bike with four wheels with ABS brakes and EBD (For safety), air-conditioning with climate control (For comfort). You can guess the ending of the ad :)
So, an average MT4 user who has no programming skills has come up with an idea (in your word: an idiot) wants someone with programming skills to write a simple prototype EA to test his idea for $20, and he is not interested in those super fancy fault tolerance (For safety) or journalling, email report to users (For comfort).. And your answer obviously is: All the bells and whistles are come with standard package and it is gonna cost you $1000 for me write something like MA crossover EA...
No wonder GM has gone into bankruptcy because as their CEO acknowledged: They didn't build a car that a buyer wants... Sounds familiar ??
MigMof
Migmof, you'll see from the subject line that I acknowledge there are "ea's" and "EA"s.
Some of my ea's are fairly quick to develop - perhaps on occasion taking 50-100 lines of code for a quick mock-up of a potential strategy.
However, another of my EA's (which runs live for a major hedge fund) currently sits at 895 lines of code, only 44 of which are for signal generation and a further 41 of which are for money-management. None of the rest of the code is superfluous. It is all there in order to fulfill the requirement that the EA reliably runs unattended. And it does. Therefore it needs to be able to handle all errors, recover its position after an outage and keep the users (both technical and trading) appropriately informed.
The folks I referred to as idiots were those who arrive with a partially conceived strategy, but proceed to estimate the coding effort without knowing or considering all the facts. I'll extend the idiots umbrella to include programmers who fail to see the fact that some users may need some "robustness" features outside of the strategy, suggesting as they do that all EAs are cheap and quick, because that is dis-information.
If you want to write an EA for $20, and then have it run live on behalf of investors, go right ahead mate. Just don't forget to let me know, so I can get the popcorn.
CB
Migmof, you'll see from the subject line that I acknowledge there are "ea's" and "EA"s.
Some of my ea's are fairly quick to develop - perhaps on occasion taking 50-100 lines of code for a quick mock-up of a potential strategy.
However, another of my EA's (which runs live for a major hedge fund) currently sits at 895 lines of code, only 44 of which are for signal generation and a further 41 of which are for money-management. None of the rest of the code is superfluous. It is all there in order to fulfill the requirement that the EA reliably runs unattended. And it does. Therefore it needs to be able to handle all errors, recover its position after an outage and keep the users (both technical and trading) appropriately informed.
The folks I referred to as idiots were those who arrive with a partially conceived strategy, but proceed to estimate the coding effort without knowing or considering all the facts. I'll extend the idiots umbrella to include programmers who fail to see the fact that some users may need some "robustness" features outside of the strategy, suggesting as they do that all EAs are cheap and quick, because that is dis-information.
If you want to write an EA for $20, and then have it run live on behalf of investors, go right ahead mate. Just don't forget to let me know, so I can get the popcorn.
CB
No one has suggested every EA should be $20. People paying $20 shouldn't/wouldn't expect the end product will have bells and whistles like you have described above. I hope people paying $20 for an EA probably just want something quick and simple to test out their half baked strategy and nothing more. With all those bullet-proof features is only gonna slow down the back testing...
On the other hand, I just hate to see few people in this forum always try to demonize others who are requesting EA work with little money. In those three flamingos world, people who can't code are so inferior and they should use their arms and legs as collateral for any coding work. Realistically, in a demand and supply market, if all the developers find the requirement are unacceptable to code for $20, they will say No (idiots excepted). The requester will then have to raise the price until someone willing to do it. So what's all the fuss about for ea for $20 !!!
Quote: "If you want to write an EA for $20, and then have it run live on behalf of investors, go right ahead mate. Just don't forget to let me know, so I can get the popcorn." <== You are missing the point, no one is preparing to run the $20 developed EA in a live account...It is just a prototype to test out someone's thoughts/strategy. Oh, please enjoy your popcorn, just make sure you don't choke...
Quote: "I'll extend the idiots umbrella to include programmers who fail to see the fact that some users may need some "robustness" features outside of the strategy" <== There was a story about a blond wants to learn how to drive. She found a very experienced instructor. After 40 hours lessons, she has basically mastered the driving skills and plan to take a driving test. The instructor insists she is not ready because she needs another 40 hours lessons to learn how to check/change engine oil, check/change brake oil, check tyre pressures and change tyres, you can guess the ending...
PS: how do we know that the someone promised "EA" is not a "eA" (expansive Ash)...
Backtesting need not be slowed down by "Belgian Whistles". One of my huge EAs has a TEST mode which disables all the collateral functionality and it tests very quickly.
I'm the last to demonize newbies. As long as someone arrives here with a willingness to learn, I'm one of the first to help. Check my posts.
Apart from that, I don't disagree with any of your response.
There will be folk who will believe that all EAs are quick and easy. If anything - I hope this thread will highlight to those folk the key differences in cost, effort and functionality between a quick and dirty mockup and a solution that can be used in the wild with large amounts of other people's money.
CB
I've posted this in response to the latest post which suggests that the going price for an EA is $20.
I'm frankly staggered at the number of new members on this forum who either can't describe the precise specification of their strategy, or who can't cut code, together with those who can do neither, but yet will quite happily say this EA is simple, this EA will take an hour, half an hour, whatever.
The truth of the matter is that in order to size the effort required you must be in possession of (and know how to develop) ALL the requirements. And by that I mean:
- The functional requirements - WHAT the EA must do
- And the non-functional requirements - HOW the EA will do it
To those who say - "yea it should take half an hour", I'd ask:
- What about the effort involved in agreeing user requirements?
- What about the need for a persistence layer so your EA can recover following an outage and continue trading?
- What about the flexibility to have a single code-base for multiple pairs?
- What about parameterisation of all the items that the user may have to vary
- What about a TEST mode which makes the EA "strategy-tester friendly"
- What about the flexibility to perform money-management on multiple charts and multiple base currencies with one codebase?
- What about journalling, and email reporting to users?
- What about a services approach to the architecture - designing in modularity from the start?
- What about commenting, user instructions, release notes, data dictionary and version control?
- What about unit-testing?
Only when you define ALL the requirements can you properly estimate the effort required.
So all of you "programmers" who turn out a few functions from your Half Bakery and call it a complete EA - please recognise the difference, and be a bit more up-front. Between you guys charging very little (and delivering very little) and the idiots who suddenly appear with a half-thought-through strategy and declare (from a position of complete ignorance) that the EA will be simple and will only cost xyz, both of you are devaluing the work of those experienced programmers who take the time to develop a professional solution to a well-understood set of requirements.
CB
Do it well or dont do it at all - Integrity - well said Cloudbreaker.
So how does this idiot get a very professional programmer to look at a complex EA?
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
I've posted this in response to the latest post which suggests that the going price for an EA is $20.
I'm frankly staggered at the number of new members on this forum who either can't describe the precise specification of their strategy, or who can't cut code, together with those who can do neither, but yet will quite happily say this EA is simple, this EA will take an hour, half an hour, whatever.
The truth of the matter is that in order to size the effort required you must be in possession of (and know how to develop) ALL the requirements. And by that I mean:
- The functional requirements - WHAT the EA must do
- And the non-functional requirements - HOW the EA will do it
To those who say - "yea it should take half an hour", I'd ask:
- What about the effort involved in agreeing user requirements?
- What about the need for a persistence layer so your EA can recover following an outage and continue trading?
- What about the flexibility to have a single code-base for multiple pairs?
- What about parameterisation of all the items that the user may have to vary
- What about a TEST mode which makes the EA "strategy-tester friendly"
- What about the flexibility to perform money-management on multiple charts and multiple base currencies with one codebase?
- What about journalling, and email reporting to users?
- What about a services approach to the architecture - designing in modularity from the start?
- What about commenting, user instructions, release notes, data dictionary and version control?
- What about unit-testing?
Only when you define ALL the requirements can you properly estimate the effort required.
So all of you "programmers" who turn out a few functions from your Half Bakery and call it a complete EA - please recognise the difference, and be a bit more up-front. Between you guys charging very little (and delivering very little) and the idiots who suddenly appear with a half-thought-through strategy and declare (from a position of complete ignorance) that the EA will be simple and will only cost xyz, both of you are devaluing the work of those experienced programmers who take the time to develop a professional solution to a well-understood set of requirements.
CB