My EA

Update:
1-I had to remove it from nano real testing as i noticed some subtle bugs could affect the main trading logic.

2- The ATS is under heavy development now. I am assuming it will be ready for the second round of demo/real cent testing on March 1, 2013.

The expected number of trades per day is 1000-1400 trades per day per instrument.

The main limitation for this ATS is the trading conditions of most of MT4 brokerages such as:

- Wider spread
-Execution latency


Thanks.
 
The core strategy (signal and subsequent order management) has to have a strong and verifiable edge. The little tweaks can't be what makes the system profitable over time, as that means you're relying too much on recent market behaviour and your edge is blurred from the natural market noise we observe.

I say this because I'm concerned you're putting a lot of time and effort into the small things as if they are stopping you from being profitable. The small tweaks shouldn't be what makes or breaks your ATS. Sure, the little details can improve performance, and that's important too, but they can't be what makes the ATS profitable, else you'll end up in a constant cycle of curve fitting optimization that won't yield results past what you see on paper from recent testing.

With that aside, where does the EA stand now? What's your timeframe for putting the EA back into production? What goals have you set to get the EA back on track?

Just trying to push you a little. :P Promising start but seeing these bugs and such pop up and stop you in your tracks is kinda concerning. Some of this stuff should have been identified and sorted out long during basic back tests on the core "idea" or strategy.. not found moving forward in near production status like this. Makes me think you skipped a step somewhere.
 
I trade extremely high frequency systems currently and the 3 brokers I would suggest for this are

1) Divisa Capital -- Deep currenex liquidity with very fair execution. I have a tight trailing stop that catches spills and spikes higher and I average only 2 tenths of a pip slippage.
2) http://www.investorseurope.com (Fyshe Horton Finney deep liquidity helps HFT mean reversion systems)
3) http://www.profiforex.com/?id=155585
You definitely want to use my link for Profiforex to benefit from our volume discounts. I do not receive any additional compensation.

They have a 6 cent spread on gold with only $3 commission/rt on all PM and fx!

Additionally, if you are using any scalping system on MT4 please note that MT4 works in sessions of 30 seconds. A quick search on Google about MT4 sessions and how to improve latency by 500ms will send you to a great thread on FF with a good expert that modifies a pending order every 30 seconds to keep the session alive to speed up your execution time.

GL and Trade well.
 
Thank you Jack for your post.

However, there is no skipping but using mql4 to code such trading systems is really challenging. Also, back testing such systems that put hundreds of trades per day is also not trivial task. I am not concerned at all about tweaking my ATS right now. The whole concern is to see the ATS is implementing the trading logic perfectly and resulting expected outcome. We, my coders and i , have to anticipate certain errors and work around them to avoid or to minimize their effects. BTW, we got error 148 with pepperstone, something i did not expect to face with pepperstone ECN razor account. It means we exceeded the allowable number of open trades!. So, the platform do not allow to open any new orders.

The main goal of this project is to develop what i can call a prototype that i could migrate to a real programming language with a real platform to trade with real brokerages.
My next step is to use Tradelink, C# and using FIX API

Again, i am not concerned at this stage to optimize, to tweak or to curve fit.

Quote from Jack_Larkin:

The core strategy (signal and subsequent order management) has to have a strong and verifiable edge. The little tweaks can't be what makes the system profitable over time, as that means you're relying too much on recent market behaviour and your edge is blurred from the natural market noise we observe.

I say this because I'm concerned you're putting a lot of time and effort into the small things as if they are stopping you from being profitable. The small tweaks shouldn't be what makes or breaks your ATS. Sure, the little details can improve performance, and that's important too, but they can't be what makes the ATS profitable, else you'll end up in a constant cycle of curve fitting optimization that won't yield results past what you see on paper from recent testing.

With that aside, where does the EA stand now? What's your timeframe for putting the EA back into production? What goals have you set to get the EA back on track?

Just trying to push you a little. :P Promising start but seeing these bugs and such pop up and stop you in your tracks is kinda concerning. Some of this stuff should have been identified and sorted out long during basic back tests on the core "idea" or strategy.. not found moving forward in near production status like this. Makes me think you skipped a step somewhere.
More...
 
Thanks for your info.

I agree rooicol in FF is always providing new nice tools/concepts. I contacted actually Divisa Capital once, the guy was irrogant and i did not like his tone. I am not sure if this was the only reason that i lost my interest to work with them. However, once i am there, i might contact you for an update :). Seems like 3-6 months from now to have my API ready.

Thanks again.
Quote from Bigpipn:

I trade extremely high frequency systems currently and the 3 brokers I would suggest for this are

1) Divisa Capital -- Deep currenex liquidity with very fair execution. I have a tight trailing stop that catches spills and spikes higher and I average only 2 tenths of a pip slippage.
2) http://www.investorseurope.com (Fyshe Horton Finney deep liquidity helps HFT mean reversion systems)
3) http://www.profiforex.com/?id=155585
You definitely want to use my link for Profiforex to benefit from our volume discounts. I do not receive any additional compensation.

They have a 6 cent spread on gold with only $3 commission/rt on all PM and fx!

Additionally, if you are using any scalping system on MT4 please note that MT4 works in sessions of 30 seconds. A quick search on Google about MT4 sessions and how to improve latency by 500ms will send you to a great thread on FF with a good expert that modifies a pending order every 30 seconds to keep the session alive to speed up your execution time.

GL and Trade well.
More...
 
Quote from mcgene4xpro:

Thank you Jack for your post.

However, there is no skipping but using mql4 to code such trading systems is really challenging. Also, back testing such systems that put hundreds of trades per day is also not trivial task. I am not concerned at all about tweaking my ATS right now. The whole concern is to see the ATS is implementing the trading logic perfectly and resulting expected outcome. We, my coders and i , have to anticipate certain errors and work around them to avoid or to minimize their effects. BTW, we got error 148 with pepperstone, something i did not expect to face with pepperstone ECN razor account. It means we exceeded the allowable number of open trades!. So, the platform do not allow to open any new orders.

The main goal of this project is to develop what i can call a prototype that i could migrate to a real programming language with a real platform to trade with real brokerages.
My next step is to use Tradelink, C# and using FIX API

Again, i am not concerned at this stage to optimize, to tweak or to curve fit.
More...

Hundreds per day, wow.. and I thought my EA that pushed 10-15 a day was ballin :P

I can see how that might be hard to keep track of, and test properly outside of forward testing.

Max orders isn't something that's related to being an ECN or not, it's about server load and messaging. Remember, MT4 is a ticket based platform, each fill is a separate trade. Often, the people who hold 100+ open positions in their account at once are trading baskets with many multiples of entries, all separate tickets, individually tracked by the MT4 server, and then when these baskets hit a profit threshold, the EA will try to close them all at once. MT4 server has to confirm the closing of one ticket per request first before it can close the next (even if you get craft and run different trade blocks managed by different terminal instances for the same account.) This process is fast, but on 100+ tickets it still can take a while. So most brokers put in a limit to start of 50 or 100 open tickets per account.

So really, this is more about MT4 / MT4 Server. It's pretty silly really, the execution they patch through to at the banks can process more transactions a second than we can count... and even MT4 Server can process huge volumes so long as they are from separate accounts... but 8+ years ago when they wrote MT4 Server, MetaQuotes had the bright idea to queue transactions on a per account basis and require the ticket to be confirmed filled before moving to the next in queue. No idea what they were thinking at the time.

Anyway, if I recall, you can get this limit raised at Pepperstone for live accounts, but you need ask. They might inquire why your EA might open that many tickets at once, not about the strategy or anything just how to best work with you on getting batches of orders put through efficiently. Or they might not and just raise it... donno... never had to do it myself yet.
 
Thank you Jack for sharing those details and suggestion.
The only issue that bothers me regarding Pepperstone is their API trading which is only for large accounts.

Quote from Jack_Larkin:

Hundreds per day, wow.. and I thought my EA that pushed 10-15 a day was ballin :P

I can see how that might be hard to keep track of, and test properly outside of forward testing.

Max orders isn't something that's related to being an ECN or not, it's about server load and messaging. Remember, MT4 is a ticket based platform, each fill is a separate trade. Often, the people who hold 100+ open positions in their account at once are trading baskets with many multiples of entries, all separate tickets, individually tracked by the MT4 server, and then when these baskets hit a profit threshold, the EA will try to close them all at once. MT4 server has to confirm the closing of one ticket per request first before it can close the next (even if you get craft and run different trade blocks managed by different terminal instances for the same account.) This process is fast, but on 100+ tickets it still can take a while. So most brokers put in a limit to start of 50 or 100 open tickets per account.

So really, this is more about MT4 / MT4 Server. It's pretty silly really, the execution they patch through to at the banks can process more transactions a second than we can count... and even MT4 Server can process huge volumes so long as they are from separate accounts... but 8+ years ago when they wrote MT4 Server, MetaQuotes had the bright idea to queue transactions on a per account basis and require the ticket to be confirmed filled before moving to the next in queue. No idea what they were thinking at the time.

Anyway, if I recall, you can get this limit raised at Pepperstone for live accounts, but you need ask. They might inquire why your EA might open that many tickets at once, not about the strategy or anything just how to best work with you on getting batches of orders put through efficiently. Or they might not and just raise it... donno... never had to do it myself yet.
More...
 
Quote from mcgene4xpro:

Thank you Jack for sharing those details and suggestion.
The only issue that bothers me regarding Pepperstone is their API trading which is only for large accounts.
More...

Not sure what to tell you, most brokers are like that, at least for FIX support. I not going to rush in to defend Pepper on this since I too would like FIX access without the commitment of capital they want, but I understand why they have the requirements in place and I dont' blame them for it either.

Remember, FIX might be an open standard, but it's just a messaging framework. Brokers usually have to hire an integration provider to create a FIX gateway for their unique back end.. this isn't a simple matter when you think pre-trade, processing, post trade, clearing, accounting, and settlement... It's more than just "your" connection to the broker after all.. unless you just want a FIX bridge to MT4 Server, which would be easier to setup but then limit you again to MT4 Server's limitations, so pretty pointless outide of learning how FIX works.

They also gotta buy quite a bit more hardware to service FIX clients, and since users of FIX are often institutions with specific needs or very order / ticket / messaging intensive reasons to want to go FIX over another platform there's usually a lot more money spent servicing and supporting these users (...you try training up support staff to be FIX and programming experts just to employ them as customer service reps, not gonna happen cheaply.) So often, if a broker offers FIX, it's with the requirement that you're gonna commit capital or do a decent bit of business to make serving your FIX account cost effective for them. This is a normal thing for most brokers. No way it's worth it for just a few grand of an account balance.

That being said, you can try Dukas for their JForex API, or IB's TWS Gateway API, or any number of other brokers offering their own 'proprietary' API that really is just a programming interface to their desktop client (just like MT4's MQL4 with the use of DLLs and other extensions), not a direct messaging service to a broker's execution servers like with FIX. The same brokers who offer their desktop APIs, usually also offer FIX API, but again, with higher requirements.

Not much more to say about it, I mean, if you look around and see that no broker is offering FIX without such commitments, there's probably a good reason for it.
 
..and just to be clear. I'd absolutely love to get my hands on a decent FIX connection to Pepper myself. They've offered me a demo before to play with but I'm not crazy about account balance commitment it takes to get a live connection going. So I research and hunt around, doing so uncovers more about FIX, and why brokers have the requirements they do to support it..

In the end, and realistically speaking, my needs are met by MT4 and MQL4 for now. Seems it's extendable enough through DLLs and some tricks that I won't be out growing it anytime soon. I'm sure if I got into HFT stuff I'd want to go pure FIX, but that's not in the cards as I see things turning out so I'm not worried.
 
Well-explained.. Thanks Jack.

When i got to the next stage of development in which i will have the code in C# with FIX, i will do my best to shop around and even try to negotiate the terms. Taken into consideration that my ATS is volume intensive trading system that might compensate easily the deposit requirement.

I also actually consider the other examples you mentioned specially jforex java API. However, i would still like to enjoy the higher leverage than 1:100 of Dukas.

I am even considering white labeling that i could connect my EA within or directly. However, i am still at least 6 months away from this stage so i am not rushing but just gathering relevant info.

The good news about my demo which started on Monday that i am up 7.4% now in profit with about 400 trades per day only EU. I am glad that i didnot notice till now any major bug.
Quote from Jack_Larkin:

Not sure what to tell you, most brokers are like that, at least for FIX support. I not going to rush in to defend Pepper on this since I too would like FIX access without the commitment of capital they want, but I understand why they have the requirements in place and I dont' blame them for it either.

Remember, FIX might be an open standard, but it's just a messaging framework. Brokers usually have to hire an integration provider to create a FIX gateway for their unique back end.. this isn't a simple matter when you think pre-trade, processing, post trade, clearing, accounting, and settlement... It's more than just "your" connection to the broker after all.. unless you just want a FIX bridge to MT4 Server, which would be easier to setup but then limit you again to MT4 Server's limitations, so pretty pointless outide of learning how FIX works.

They also gotta buy quite a bit more hardware to service FIX clients, and since users of FIX are often institutions with specific needs or very order / ticket / messaging intensive reasons to want to go FIX over another platform there's usually a lot more money spent servicing and supporting these users (...you try training up support staff to be FIX and programming experts just to employ them as customer service reps, not gonna happen cheaply.) So often, if a broker offers FIX, it's with the requirement that you're gonna commit capital or do a decent bit of business to make serving your FIX account cost effective for them. This is a normal thing for most brokers. No way it's worth it for just a few grand of an account balance.

That being said, you can try Dukas for their JForex API, or IB's TWS Gateway API, or any number of other brokers offering their own 'proprietary' API that really is just a programming interface to their desktop client (just like MT4's MQL4 with the use of DLLs and other extensions), not a direct messaging service to a broker's execution servers like with FIX. The same brokers who offer their desktop APIs, usually also offer FIX API, but again, with higher requirements.

Not much more to say about it, I mean, if you look around and see that no broker is offering FIX without such commitments, there's probably a good reason for it.
More...
 

Attachments

You must log in or register to reply here.
Back
Top