Slow Trade Copier Executions

[JesseOffshore]

Hello,

My strategy is a scalping strategy using stop market orders for order entry. These orders get placed one tick above or below the bar that's giving me entry signal. I use the chart trader and market structure to enter trades and manage the trades with the little paddles on the chart. ATM places stop and profit levels. I do not use a DOM. Only price action.

Very rarely did this ever cause me any noticeable or tangible problems. On very rare occasions, I would get rejected because the market shot off too fast and went past by order. But this did not effect my strategy and its simply the cost of my strategy.

Now I have moved onto using a trade copier from replikanto and multiple funded prop firm accounts, and the problems start. Very clearly whats happening is that the trade copier takes a split second longer to populate and send the orders across all 5 accounts rather than just one account and no trade copier at all. What ends up happening is that either I get rejections for all accounts, or worse, 1 or 2 accounts will get filled, the others rejected on entry, but the ATM still places my stoploss and take profit levels. So in the accounts that get rejected, if I enter a long position, I get rejected, but my profit target is still there so price hits the profit target and puts me in a short position in a fast moving bullish market.

Ive played with this alot and I do not have this problem if there is extra time (split seconds) before price moves into my stop market order. The copier works fine.

So the reason for this post is to find out if I can speed up these orders so that they are as fast as using no trade copier. Or to try and find out if its even possible to speed this up. Ideally I would like to do whatever I can to make all my entries lightning fast.

I use 1 workspace with 1 ES chart that has 2 tabs, a daily chart and a 2000 tick chart. and another MES chart with 2 tabs, a daily and a 1450 tick chart. No DOMS at all. Indicators on all these are very simple. Just an EMA and tick counter, but alot of drawing tools and high priority economic news on the chart. I have very low demand on ninjatrader.

I use Rithmic data and I have 1gigabit fiber connection that tests around 950Mbps connected with CAT 8 cabling.

My computer is unfortunately a 2019 Mac running windows 10 with Bootcamp. The hard drive is partitioned giving the windows side 500GB and its solid state. Windows side is bare with only required applications and mail and chrome. Processor benchmarks at only 8,500 but im not sure if its testing properly due to how its setup. My graphics card is lacking and not enough to power my external monitors: intel iris plus graphics 655

Intel(R) Core(TM) i7-8569U CPU @ 2.80GHz 2.80 GHz

16.0 GB (15.8 GB usable)

Will using the superDOM over the chart trader make any improvement? Will a new PC help? Are there settings within ninjatrader that I can utilize? I just need those few milliseconds back I lost from the trade copier. I dont mind buying a new PC if I get confirmation that it will give me a tangible increase in order entry speed.

I am also emailing Replikanto with the same questions, but I want whatever information ninjatrader can give me on this issue as well.

Thanks.

[brucerobinson]

JesseOffshore - this is a topic I am interested in also and ask if you would kindly post your progress into this. I guess Support will answer here, but I'd be interested in the 'whole picture' i.e. anything Replikanto may contribute, setup etc.

Only thing I can think of to throw in FWIW is a DOM has been 'lighter' (very light) on resources than Chart Trader. Support are better placed to comment on this (e.g. may no longer be the case) and hopefully will, but I use DOMs for that reason - i.e. I don't use Depth of Market or anything else whatsoever - purely use as an Entry/trade management tool

Even at times of market fury, the DOM is never affected (at least not to the naked eye). Whether or not this is relevant re: processor bottlenecking, dividing work between cores, how Replikanto plays with NT is way above my paygrade so I'll shut up and wait to see from those qualified to comment :-)

KR

[bltdavid]

Is that a Mac laptop?
Googling 'i7-8569U' shows it's a mobile processor.

Does that 1 gigabit fiber have 1Gbit speeds on upload
and download, or just download?

Is your Rithmic connection using the 'plugin' approach?
I had some issues with that and turned that off, the
normal non-plugin usage should use less resources.

Your entry strategy made me thing of Apex, but then
you mentioned Replikanto by name -- whereas Apex
sells their own trade copier.

My point is:
Since Apex promotes multiple accounts, and they sell
a trade copier, so if you were with Apex, I would say
definitely ask them for advice -- in fact, I think Darrell
Martin is reselling Replikanto under the Apex brand,
so you might still be ok there.

Back to your Mac.
IMHO, yes, you should replace it with a desktop
Windows PC -- Apex has a relationship where the
fastest desktop PC (Radical X13) was on sale.
It has an Intel Core i9-13900KF at 5.8GHz -- I'd
recommend you future-proof it by maxing out the
RAM at 128GB and maxing out the SSD at 2TB
and buy that puppy -- a machine like that is quite
the behemoth and should serve you extremely well
for years and years.

-=o=-

My summary:
From the description of your problem, a new super
fast Windows PC should make the biggest difference.

That is, get off the Mac.

[jeronymite]

JesseOffshore There's "scalping" and there's "Scalping!". If milliseconds matter, multiple accounts are in play, and there's significant capital risk, there are several considerations that must be understood. They are generic, but essential for the serious trader:

• Hardware: The faster and more resource-endowed, the better. Configured for maximum performance. Redundancy where obvious and sensible. Consider all hardware in the "chain of service provision": computer (including memory, CPU, disks, network adapters, graphics adapters, screens, cabling, keyboard and mouse (yes, be sure the batteries are fully charged!)), connections (cables vs WiFi, cables being preferable), modem/router, network infrastructure at point of operation (e.g. copper, fibre, overall technology), etc. Failure mitigation, including Uninterruptible Power Supply (UPS) to suit circumstances. Appropriate physical environment to maintain the kit well (e.g. airflow and temperature regulation). Etc!
• Software: Configured with laser-focus on the primary goal. Nothing extraneous: nothing running that is not material to the goal. Up-to-date and solidly tested in context of the goal. Fit for purpose, as opposed to good enough in the circumstances. Known limitations. Underlying resource usage architecture, in as much as it can be reasonably understood and is known to impose particular requirements on the system (e.g. NinjaTrader's threading considerations). Automate, where possible, to remove latency of action and indecision ... but test, test, test to ensure the automations actually do what you want and do not do what you do not want. Know and understand the impact of "hidden" resource consumers (e.g. Windows processes such as indexing, software updates, settings that allow "default" actions, etc) and decide how to manage those.
• Network Service: Internet Service Provider offering and reliability, including verifiable up/down speeds, time-dependent traffic variations that can cause congestion (e.g. the Netflix special release event nemesis), etc. Local conditions such as network configuration and usage in the place of operation: dedicated to what you are doing, not shared, particularly with "heavy Internet usage" activities such as streaming (e.g. the "unmissable" Netflix event), etc.
• Geography: Distances matter! Network performance is distance dependent, as well as technology dependent. If your target market is exchange-based, consider latency to the point of execution, including all intermediary touch points along the way. Distance is both geographical (i.e. physical point-to-point) and logical (i.e. distance travelled). On the Internet, it's sometimes possible to logically "go round the global network" just to reach the computer in the room next door. Optimise distance (physical and logical) where possible.
• Scale and Scope: Recognise what is possible with scaling your activities, and what is reasonably in-scope and what is not. Do not "push the envelope" based on desire, rather than analysed, reproducibly reliable activity. If more scale or scope are desired or required, there's a process of analysis, planning, implementation, testing and validation to be followed, regardless of this being a retail trader operation, not some grandiose hedge fund. It doesn't need to be excessive, over-the-top activity, but it does need to be well thought out and rigourously applied. Do not succumb to "scope creep".
• Reputation, reliability and robustness: Trading with any significant risk requires a suitable investment in the tools-of-trade you use. They are all part of a business investment. It all should be sourced with the reputation of the providers in mind, and with as reasonable a review as possible of the reliability and robustness of the specific offerings being sourced. Even highly-regarded producers occasionally release things that are "not up to their usual standard", so look carefully before committing. Consider the quality of Support as well. And whilst some bargains can be genuinely "good value", the phrase "you get what you pay for" exists for a reason.
• Process: Trade with due diligence and reliable, proven process. Trade with processes that recognise the potential points of failure and "abnormal events", and have well-rehearsed responses for those things. Keep processes "alive" by updating as learning happens. Trade within the limitations of all the above.

All of that is well and good, but what about your specific situation? In addition to the good comments already provided by others, frankly, bltdavid 's recommendation is top of the list as far as readily identifiable matters go. My own approach aligns well with the concept of "future-proofing" he expresses. Nevertheless, the best hardware can only contribute so much. The software must be capable of what you demand. Again, consult with the relevant software producers on the matter. Ultimately, all the "gear" you use has to be fit for purpose. So make sure your intended usage is crystal clear and you've shown the gear can meet the purpose in practice.

None of this has touched on your "trading infrastructure" (i.e. broker, accounts, margins, etc), as one assumes you have taken all that into consideration too. Neither does it really consider your trading methodology/plan, as, again, one assumes you have accounted for that in your deliberations. The only comment would be that if all else is "as good as it can be", you may need to revisit these things and adjust appropriately for your desired outcomes.

Hope that's all useful generally, and also for your current situation. May your trading be safe and profitable.

Thanks.

[JesseOffshore]

Yes, its a Mac laptop. A generic speedtest just now provided 946 down and 304 up testing to local servers.

Another thing I should note is that my physical location is southern Thailand, south east asia.

I am trading Apex funded accounts but my strategy has nothing to do with them.

I have tried the 'plugin' approach but apex have updated their 'connection guide' to not do that. I created a helpdesk ticket to clarify which connection method is best and they said to follow their updated connection guide. I also asked if I should change the gateway server, considering my physical location is Thailand, Singapore is very close, but they said to use Chicago.

Yes the replikanto copier and the apex copier are the same, so maybe ill reach out to Apex and forward all this info.

I am not ready to spend $10k on a supercomputer just yet. Im only just now starting to scale up in contracts so im not very far into the money.

Ideally I really just want to know how to make this all faster from a technical side so I can make an educated and measured decision on what to do next.

[bltdavid]

I am not ready to spend $10k on a supercomputer just yet.

Gotcha.

You don't need to spend $10k on a new computer.
Spending $3k is probably enough, or even $2k.

The point is:
How do you answer the question:
Is the Mac your main bottleneck?

You will have to investigate this yourself
With a real PC.
A real fast PC.

Remember, you will not know the tangible/measureable
benefits of moving to a brand new fast PC until you
actually do the move to a brand new fast PC and
test it.

If $3k is a lot of money, consider setting up a lease
arrangement and pay monthly.

I mean, I still think getting off the Mac should be an
extremely serious focus of your investigation.

Just my 2˘.

[jeronymite]

JesseOffshore I agree with bltdavid -- you are most likely to improve your situation by switching to a PC with some decent specs. See what you can afford and what you can get for the price.

Thanks.

[NinjaTrader_Jason]

Hello JesseOffShore,

"So in the accounts that get rejected, if I enter a long position, I get rejected, but my profit target is still there"

If your entry order gets rejected, no ATM orders are submitted. ATM orders are only submitted after the entry order fills. Have you only experienced this behavior with the trade copier or also without the trade copier? I have not seen such behavior before. Are you sure ATM orders were submitted after the entry order got rejected or could it be you checked 2 different accounts?

You mention the extra time is not there without the trade copier which indicates the trade copier is related. I'm not sure how the trade copier functions or how it needs to be configured. Please check with the vendor of the trade copier if this can be as expected.

Typically the SuperDOM is faster as it does not contain all the items you have in your charts. I suggest to test the trade copier with the SuperDOM and then compare it with charts.

Besides all the helpful posts from the other users in this thread, you can find more performance tips at the link below.

https://ninjatrader.com/support/helpGuides/nt8/NT%20HelpGuide%20English.html?performance_tips2.htm

[JesseOffshore]

Gotcha.

You don't need to spend $10k on a new computer.
Spending $3k is probably enough, or even $2k.

The point is:
How do you answer the question:
Is the Mac your main bottleneck?

You will have to investigate this yourself
With a real PC.
A real fast PC.

Remember, you will not know the tangible/measureable
benefits of moving to a brand new fast PC until you
actually do the move to a brand new fast PC and
test it.

If $3k is a lot of money, consider setting up a lease
arrangement and pay monthly.

I mean, I still think getting off the Mac should be an
extremely serious focus of your investigation.

Just my 2˘.

JesseOffshore I agree with bltdavid -- you are most likely to improve your situation by switching to a PC with some decent specs. See what you can afford and what you can get for the price.

Thanks.

I appreciate you guys help and you are both right. The supercomputer comment was mostly aimed at the behemoth mentioned earlier lol. Unfortunately I am no longer in the states and dont have any plans to go back right now, so EZtradingcomputers is out of the question. Most countries outside the USA dont have trade agreements like the USA does, and they slam you with astronomical import taxes on electronics. I have budgeted around $3k that ive put aside for a laptop and another $1500 for traveling monitors for when im away from my desk at home. I do want a behemoth, but its going to have to wait for a little while longer. Probably time to step up and upgrade the "business" for the short term.

[mattcrown]

I was trading APEX copier on a cheap laptop. the fills we all over the place and half the time the copier didn't even work at all. i got a high speed computer and all was vastly better.

[JesseOffshore]

Hello JesseOffShore,

"So in the accounts that get rejected, if I enter a long position, I get rejected, but my profit target is still there"

If your entry order gets rejected, no ATM orders are submitted. ATM orders are only submitted after the entry order fills. Have you only experienced this behavior with the trade copier or also without the trade copier? I have not seen such behavior before. Are you sure ATM orders were submitted after the entry order got rejected or could it be you checked 2 different accounts?

You mention the extra time is not there without the trade copier which indicates the trade copier is related. I'm not sure how the trade copier functions or how it needs to be configured. Please check with the vendor of the trade copier if this can be as expected.

Typically the SuperDOM is faster as it does not contain all the items you have in your charts. I suggest to test the trade copier with the SuperDOM and then compare it with charts.

Besides all the helpful posts from the other users in this thread, you can find more performance tips at the link below.
https://ninjatrader.com/support/help...ance_tips2.htm

Thanks for the reply.

Let me clarify what I mean by "rejected" so there's no confusion. I want to go long and place a buy stop order 1 tick above the bar giving me an entry signal. Trade copier is activated and will copy orders from leader account to 4 follower accounts.

A common result in volatile markets (an hour after a news event, once things settle a bit) is that I will get 2 or 3 pop up windows that say something like "Can not place a buy order below current market price" So I will be in the market in 2 or 3 accounts but the others, im assuming the last ones in the string of orders, get rejected with this pop up. No issues when not using the copier, but when using the copier the market moves past my order level. Hope this makes sense.

I can promise you that whatever happens in that split second, occasionally itll throw me in a short position exactly where my take profit was supposed to be from the ATM.

Rithmic data is down so I cant recreate it at the moment.

My main question that I really want to know, is if there is any way to speed this process up. Just that split second I lost to the trade copier? Will a new PC actually speed this up?

The question is important not because I dont want to buy a new PC, its because if this isnt something that a new PC will help, like fundamentally its not possible because the operation uses very little resources, I will need to ditch the copier and get larger accounts or change my strategy.

[JesseOffshore]

I was trading APEX copier on a cheap laptop. the fills we all over the place and half the time the copier didn't even work at all. i got a high speed computer and all was vastly better.

This is good to know. Can you give me any details on the setup? Specifically your processor and whatnot?

[NinjaTrader_Jason]

JesseOffshore, this could also be an inflight execution. Both ATM orders are filled which causes a reverse position. However the entry order is not rejected in this scenario. ATM orders are only submitted when an entry order fills, not when a entry order is rejected.

If you like me to investigate this, please send me your log and trace files from within the NinjaTrader platform so that I may investigate.

You can do this by going to the Control Center-> Help-> Email Support

Ensuring 'Log and Trace Files' is checked will include these files. This is checked by default.

Reference the following ticket number in the body of the email: Att Jason H, 03640376

Please include at what time the entry order was submitted, at what price, for what instrument and for what account.

A new computer will likely speed it up, but it is hard to say by how much. The only way would be to test it.

[brucerobinson]

Wow, this is turning into “War & Peace”.

Lots of suggestions - so how do you get more of a steer on if the stray orders are ‘caused’ by the copier’s load, or the copier’s speed of execution before/other than a $2-3k test?

Running the NT Utilisation Monitor will show you where the copier ranks. If it is low, it suggests it isn’t choking things up

Putting together your description of the issue@Ninjatrader_Jason’s reply, @mattcrown’s experience reminded me of I problem I experienced some long time back whilst my NT setup would struggle in high volatility.

I use DOM’s with ATMs to enter multiple orders with SL & TP OCOs

I would manage the trade manually using market orders on the DOM until the final manual contract exit would cancel the ATM.

Very occasionally, instead of the final ‘click on the DOM’ placing a single contract order to flatten the remaining position, and cancelling the ATM, I would find I was in a new ATM entry in the opposite direction but with one contract less than the ATM quantity. I can’t recall the details of this and am not saying it is precisely what you’re experiencing, but after much combing of executions etc. Support concluded that it was what they called an ‘in-flight execution’ where relative speed of a single market order execution vs an ATM execution in high volatility with the platform not performing in a sprightly manner.

I won’t go further but what triggered the thought was the discussion with Jason that what you’re describing shouldn’t be possible. Technically true. Yet here we are….. And it is/was. So the Q is, how can what IS happening be happening….

It struck me that your ‘orphaned ATM entries’ made by the copier under high volatility/load bear a striking resemblance to me finding that instead of inputting a single market order to close out of a position I found I was in a new ATM Entry in the opposite direction but already one contract exited, the difference between executing a single order versus processing the ATM (equivalence to with/without trade copier & impact it’s code efficiency & load has), how nimble your workspace is with/without, whether hardware will make any difference.

Anyhoo, might give clues as to what is actually taking place re: orphaned entries which ‘can’t happen’ from the cobweb-infested depths of my memory.

[bltdavid]

Let me clarify what I mean by "rejected" so there's no confusion. I want to go long and place a buy stop order 1 tick above the bar giving me an entry signal. Trade copier is activated and will copy orders from leader account to 4 follower accounts.

Yeah, I can see what's happening.

But wait, 1 tick?

That's some very tight parameters. It's very easy to see how your Buy Stop entry orders
can get rejected because price has moved quickly -- so quickly that this happens before each
follower account entry order is fully submitted -- and sooo, when price moves so fast above
your designated entry price, one or more of the submitted entry orders on the follower
accounts get rejected.

When this happens on your follower accounts, you immediately get a popup for that
account stating the entry order was rejected due to 'Can not place a buy order below
current market price'. These popups should come to the front, if they don't, that's
another issue.

With more control, a trade copier could be told to convert a rejected Buy order into a
Market Buy order -- meaning you would at least get in, though some slippage might
happen. Such a feature would guarantee the follower account still has an entry position
so that when the leader's target/stop orders are replicated to the followers, those two
orders are not orphaned.

A common result in volatile markets (an hour after a news event, once things settle a bit) is that I will get 2 or 3 pop up windows that say something like "Can not place a buy order below current market price" So I will be in the market in 2 or 3 accounts but the others, im assuming the last ones in the string of orders, get rejected with this pop up. No issues when not using the copier, but when using the copier the market moves past my order level. Hope this makes sense.

Yes, it does. I can see what is happening. Makes perfect sense.

I hope you're catching these rejected order popups immediately when they occur.
I'd suggest you close the leader position very soon after dismissing those popups,
better with small profit, of course.

I mean, don't hesitate, close the leader position immediately.

The point is, if you don't manually close the leader position quickly -- before your stop
or target is hit -- it's possible for one of the orphaned stop/target orders in your follower
accounts to get filled inadvertently.

That is, for a long position in the leader account, if a follower account's entry order was
rejected, but the copier still replicated the stop/target orders to that follower, then if you
let your leader position hit target, the follower orphaned targets are 'Sell Limit' and those
follower accounts all of a sudden have a short position.

Same thing with your stop order -- if you allow the long leader position to hit the stop order,
then any orphaned sell stop orders also create an inadvertent short position in the follower
accounts where the initial entry order was rejected.

I can promise you that whatever happens in that split second, occasionally itll throw me in a short position exactly where my take profit was supposed to be from the ATM.

This is probably your fault.
Now wait, hear me out -- let's discuss what behaviors may be causing
these short positions to actually occur.

To wit:
You should be seeing these popups immediately when they occur.
You should close the leader position immediately when you see the first popup.
You should be dismissing these popups immediately when they occur.
Go to CC > Positions, see anything? Immediately right click and Flatten Everything.

The point is:
Closing the leader position immediately ensures that no orphaned stop/target
orders have a chance to get filled.

If you see short positions in your follower accounts where your target was
supposed to be -- me thinks you need to be more pro-active with the steps
above. What I mean is, I'm hoping that part of the solution for this portion
of your problem is just you being quicker with the tasks of closing, dismissing,
and flattening everything.

Don't be angry, but do you see my point?
Are you doing enough, after the rejection, to control the events leading to
these inadvertent short positions?

My main question that I really want to know, is if there is any way to speed this process up. Just that split second I lost to the trade copier? Will a new PC actually speed this up?

Yes, I think a new PC would speed up things.
[Would it be enough, I don't know -- but trading from a Mac is surely incurring an unnecessary penalty.]

Here's an idea:
Since you are trading manually, and dependent on the speed of the charts ...
In NT7, a setting existed called 'display update interval', which could be made smaller (ie, make charts update faster).
In NT8, no such setting exists, but plenty of complaining about this does exists, see here and here -- see addon here.

Try using that addon.
Maybe using a faster refresh rate will help.

Just my 2˘.