Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
Trading Rig Upgrade for Backtesting/Optimization in Multicharts
I'm looking to upgrade my trading infrastructure. Quite limited right now.
I'd appreciate it if I can have the forum's two cents as to the following two questions:
As of now I've been trading regular MC on a 2020 LG Gram 17 laptop, i5-1035G7 6M Cache,
1.20 GHz base clock, turbo up to 3.70 GHz, 16 RAM (Costco edition).
This is obviously not ideal, especially for backtesting and optimization purposes, which takes a lot of time. I'm working with just one tick chart at a time but lots of input parameters. I've undervolted it and am using a cooling pad so it never runs hot. I've added one external monitor.
1) Do you think increasing my laptop RAM from 16 to 40 GB will be of any help with optimization speed?
I'm also thinking of buying an old workstation so as to have a more appropriate rig to run automated trading 24/5. I have a very limited budget.
2) Do you think the following model would do better than my LG Gram for optimization purposes?:
Dell Precision T5600 with two Xeon E5-2690 2.9GHz, 20M cache, total 16Core, 64 GB with SSD.
I could get one for about $700 USD shipping included and upgrade the graphic card for something decent like GTX 970 (+$100).
Thank you
Can you help answer these questions from other members on NexusFi?
My personal experience with running Optimisations and WFO's on MultiCharts is to maximise the Processor spec, and aim for the highest number of Cores (including Hyperthreading). MC actually does not use a lot of RAM on the two high-spec PC's that I perform the strategy testing on ... maybe 16GB at most on a 64Gb PC and a 32Gb. But MC definitely makes the most of the 32 cores, and you can manually drop the Cores down if you want to leave a few available for other processes while MC Optimises/WFO in the background.
I hope this helps.
Hi,
I'm a Multicharts user, who does a significant amount of optimizations/backtests.
it's clear your concerned about causing a early demise of your laptop and I assume you feel the optimizations are taking longer than you'd like?
In my opinion increasing your ram will have a marginal effect. It's really more about base clock speed since your not overclocking and have in fact undervolted. Your concerned about heat in the laptop, adding ram will marginally increase the internal heat production.
Computationally tick charts are more work, you didn't say how much tick data your working with or the resolution? Large amounts of tick data will be helped by additional ram.
If you wanted to use the computer while optimizing then you could increase the ram but I wouldn't go above 32Gigs. and reduce the cores used from 8 to 7 or six depending on what your going to be doing besides using Multicharts.
I do think that the workstation would do a great job of optimizing but I don't think you should put any extra money on a graphics card since the graphics display on MC is not complex and unless your running a very large monitor or multi-monitors that you don't think the oem graphics card can handle I wouldn't upgrade the card. I bot a used Dell T7200 for the very same reason and have build three computers from components a couple of years back. All 3 I-7 3700K with 32gigs of ram each able to drive 4 monitors each.
As someone who has admittedly "overkilled" it in the areas of computers. Is your current computer really holding you back or is this a want rather than a need? If it's just a "nice to have" I'd save your money and wait because computer technology is progressing so rapidly you'll be able to get a much more capable machine in the near future.
I'm fairly new to programming and automation, so I'm still in the progress of discovering my needs. What I do know is that loading my three years historical data (2000 tick charts) takes about 9 min every time I turn on MC.
My initial thought was that increasing RAM would help at least with the backfilling, if not with the optimization. But my 16 GB memory is not even used at 50% when backfilling, so it actually doesn’t sound like adding RAM would have any impact here.
And the laptop uses about 35% memory while optimizing
Well, reading you makes it clearer that there's no way around my 4 cores and their poor clock speed. I need to accept the fact that more RAM will not improve anything at all, even though that's the only option for upgrading my current LG Gram.
Thermal throttling kicks in very quickly in LG Grams so that's why I've thought overclocking was out of the question (I could be wrong). I thought I'd get better performance by undervolting (just slightly anyway) since it helps to avoid throttling. But I think I'll look further into the possibility of overclocking instead.
It took me a month to learn enough coding to code my first strategy and do simple historical backtests. WFO is still to be done. And I want to add 4-5 other strategies over the next few months. When I fire up an optimization, I usually have between 10 min and 2 hours to wait depending on the settings I choose. It’s not the end of the world compared with what many of you are doing, but I feel it is somewhat of a drag as I'm testing and retesting a lot of components.
I would feel things moving much faster if I could cut down my optimization time by half. I see myself working on this for many months to come, hence the thought about buying a used desktop that would speed up things. Of course, I’m doing other stuff while waiting for optimization reports so it's not wasted time, but it cuts the workflow and makes it more difficult to stay "in the zone", which is precious. A used workhouse would also be better suited for running 24h, although I don’t care too much if my LG Gram dies in 3 years because of that.
I was about to pull the trigger on a $120 32 RAM stick to speed things up a bit with my current laptop. But it’s now clear it’s not going to be of any help. I need to accept my reality: the bottleneck is my CPU and no amount of RAM will change that.
From my experience, the number of cores along with the amount of RAM really determines how long running tests will take from a hardware perspective. On the other hand, how much data you have loaded in a chart is a huge factor if RAM is limited.
As for my machine, I have a 8-core 9900K-I9 processor with 128GB ram. I have some very large charts that I run tests with and that RAM is crucial! I have to say that sometimes I wish I had an AMD system with the crazy amount of cores, which would be optimal for backtesting.
