Built for long optimization runs
Designed for long history datasets and 100k to 300k pass optimizations in MetaTrader 5.
AMD EPYC MT5 Backtesting
MT5 Backtest Farm on AMD EPYC
Private cloud for your Strategy Tester
If one optimization already takes 30-50 hours, move heavy MetaTrader 5 backtests to a private cluster of AMD EPYC 9454P servers and reduce waiting time for 100k+ pass runs, depending on EA logic, symbols, data and tester settings.
Private agent cluster
No sharing and no per-pass fees, only your isolated farm with direct control over the compute.
Powerful node class
AMD EPYC 9454P 48-core Genoa nodes with fast NVMe storage and stable Windows Server deployment.
Your terminal stays the master
Your existing MT5 terminal stays the control center while the servers do the heavy lifting as remote agents.
Private Cloud for Your Strategy Tester
If one optimization already takes 30-50 hours, offload MetaTrader 5 backtests to a dedicated cluster of AMD EPYC 9454P servers and shorten 100k+ pass runs, depending on EA logic, symbols, data and tester settings.
Built for heavy EAs
Designed for long history datasets and 100k to 300k pass optimizations in MetaTrader 5.
Private MT5 agent farm
No sharing, no per-pass fees - just your own isolated farm with full control and better data privacy.
EPYC 9454P hardware
Powered by AMD EPYC™ 9454P 48-Core "Genoa" (Zen 4) with fast NVMe storage.
Plug-and-play with your MT5
Your existing MT5 terminal stays the master - our servers join as remote agents and do the heavy lifting.
Who is this for?
Traders and teams whose MT5 optimizations no longer fit on one machine.
If even a 64-core or 128-thread CPU still leaves you waiting for hours, the problem is not your patience. You have reached the point where a bigger desktop stops scaling and a dedicated backtest farm becomes the practical next step.
Individual algo traders
- Run heavy MT5 robots with long history and 100k+ pass optimizations.
- Want to reduce 30-50 hour runs and iterate faster after a real workload test.
Teams, prop firms and services
- Several people share hardware and queue for optimizations all day long.
- Need stable and reproducible MT5 backtests across symbols and strategies.
How it works
Two ways to use our servers with MT5 Strategy Tester
You can move MT5 fully onto a high-performance server, or keep the master terminal on your local machine and attach our CPUs as remote tester agents.
Full VPS or dedicated server for MT5
You rent a Windows VPS or a dedicated machine with Intel i9, AMD Ryzen, or EPYC class CPUs and work through standard Windows RDP.
- Install MT5, load history, and run backtests directly on the server.
- All CPU cores, RAM, and NVMe disks are reserved for your own workload.
- Best when you want one powerful machine for live trading and research.
Connect our CPU cores as remote MT5 agents
If you prefer to keep MT5 on your home PC or laptop, you can add our servers as remote agents in Strategy Tester.
- Your local MT5 remains the master, but passes are sent to remote agents over the network.
- MT5 can use your own CPU and the farm cores together for higher total speed.
- No workflow change is required, only more agents in the Strategy Tester.
The problem
Why your home PC stopped being enough for MT5 optimizations.
Modern EAs are heavier, use more symbols, and run across 7-10 years of history. What used to be a quick optimization becomes a long continuous workload, even on a strong workstation.
1
Heavy EAs and long history
Every pass has to process symbols, filters, and risk rules over years of ticks, so 100k to 300k runs can take days on one machine.
2
Scaling hits a wall
MT5 does not scale linearly forever. Memory bandwidth, disk I/O, and non-parallel parts of the tester become the bottleneck.
3
Your workstation becomes a bottleneck
While a multi-day optimization runs, the machine stays busy and noisy, and your research work slows down with it.
The solution
Your MT5 stays the master. We give it a farm of dozens of cores.
Keep the familiar master terminal and add a private cluster of AMD EPYC servers as MT5 tester agents. Your workflow stays under your control, while the heavy lifting moves to dedicated hardware.
1
Master terminal stays in place
No account migration and no workflow reset. The local MT5 instance remains the brain.
2
EPYC nodes join as agents
Dedicated AMD EPYC servers attach to the Strategy Tester as a private farm and take optimization passes.
3
Scale from one machine to many
Combine local power with the farm and scale up whenever you need extra throughput.
How it works technically
How it looks inside MT5.
From your side there is no new software. You enable optimization in the Strategy Tester, tick the farm agents, and MT5 starts splitting passes across local and remote cores.
1. Choose EA, symbol, dates, and optimization mode
Set Fast genetic or Slow complete algorithm in the Strategy Tester depending on whether you want speed or full coverage.
2. Tick local and farm agents
Enable your own CPU cores plus the Local Network Farm agents hosted on Winservers.NET.
3. Start the optimization
MT5 distributes passes automatically across all agents. You watch the results in the master terminal as usual.
MetaTrader 5 Strategy Tester with local cores plus remote EPYC agents
Farm configurations
Standard MT5 farm configurations.
Pick a starting point that matches how often you run optimizations and how quickly you want 100k to 300k pass runs to finish. The cluster can scale up or down as your research pipeline grows.
Farm 1
Starter Farm
One EPYC 9454P node for traders who want to add serious compute to an existing workstation without moving everything at once.
How it works
Add one remote MT5 tester node to your current setup, keep the master terminal where it already is and send only the heavy passes to the farm.
Who it is for
- Traders who want a first step beyond a single local workstation.
- Users who run 100k+ passes but do not need a full cluster yet.
- Teams that want to test the workflow before scaling nodes.
Monthly price
€270/month
per node · Windows license not included
Win 11 Pro recommended — ~10× cheaper than Server
Key advantages
- 48 Zen 4 cores, 256 GB DDR5 ECC RAM and NVMe storage.
- Easy to compare against your current machine before scaling.
Farm 2
Pro Farm
Three EPYC 9454P nodes for traders and small teams that need more throughput and shorter queues for larger optimization batches.
How it works
Your master MT5 launches the job and the farm splits optimization passes across three dedicated agents with reserved CPU and RAM.
Who it is for
- Algo traders with several EAs, symbols and long history sets.
- Teams that need predictable turnaround time for frequent reruns.
- Users who want a stronger answer than a single VPS.
Monthly price
€750/month
for 3 nodes · Windows license not included
Win 11 Pro recommended — ~10× cheaper than Server
Key advantages
- 144 total cores across dedicated EPYC servers.
- Good fit for 100k to 300k pass workloads across multiple symbols.
Farm 3
Custom Farm
Five or more EPYC nodes sized around your EA logic, symbols, pass count and the time window you want to hit.
How it works
We size the cluster around your workload and can expand node count when the optimization queue grows.
Who it is for
- Teams that run repeated research jobs and need a tailored layout.
- Users who want to scale beyond the standard node counts.
- Customers who want larger farms or longer commitments.
Pricing
Custom pricing
with volume discounts up to 20% · Windows license not included
Win 11 Pro recommended — ~10× cheaper than Server
Key advantages
- Cluster size matched to symbols, history depth and target runtime.
- Flexible path when one or three nodes are no longer enough.
Each farm node is delivered with Windows Server 2022 Evaluation preinstalled for initial testing. All prices exclude Windows licensing.
A note on Windows licensing: Windows Server is licensed per core. On a 48-core EPYC node, a proper Windows Server license adds substantial cost — often more than the hardware rental itself. Windows 11 Pro handles MT5 remote agents equally well, costs roughly 10 times less than a comparable Windows Server license, and is the option we recommend for most backtest farm nodes. We can help you choose the right license when you contact us.
Final step
Ready to reduce long MT5 optimization queues?
Tell us which EA you are running and what kind of optimizations you launch. We will propose a farm configuration that fits your workload and budget.
Case study
Example: 7.5 years of XAUUSD history, 296,000 passes.
Example scenario: same EA, same 7.5-year XAUUSD dataset, same 296,000-pass optimization. Only the compute setup changes, and the waiting time changes with it.
16-core desktop
The run is possible, but it can occupy the machine for an entire weekend or longer.
128-thread workstation
In this example scenario, a strong CPU can compress the job to around 50 hours, but it still consumes almost two trading days.
Threadripper + 3x EPYC farm
The workload is split across roughly 200+ effective threads, so the same optimization may finish in several hours instead of several days in this example scenario.
Actual timings always depend on EA logic, modeling mode, and data quality. Each farm is sized around your setup and your target time for a typical optimization.
Onboarding
How we bring your MT5 farm online in 1-2 days.
The process is intentionally simple: you describe your current workflow, we size the farm, deploy the servers, and help you connect them as remote agents.
1
Discuss your workload
We clarify which EAs you run, history depth, typical pass count, and what acceptable optimization time means for you.
2
Propose farm options
You get 1-2 concrete configurations: node count, RAM, NVMe, and whether to rent Windows licenses or bring your own.
3
Deploy and configure
We provision the servers, install MT5 MetaTester agents, configure ports and security, and prepare the farm for jobs.
4
Connect from MT5
We send IPs, ports, and agent passwords and guide you through adding the farm in MT5 as a Local Network.
5
Run and scale
You start running real optimizations. If needed, we help you scale nodes up or down as the backlog changes.
6
Keep iterating
Once the farm is online, you can refine EAs, rerun passes, and keep your local workstation free for analysis.
Typical setup time for a standard farm is 1-2 business days after payment and OS licensing.
FAQ
Typical questions about MT5 farms.
Short answers to the questions traders usually ask when they move from a single MT5 box to a dedicated backtest cluster.
How is an MT5 farm different from a normal VPS or server?
A normal VPS or single server usually runs one or a few MT5 terminals for live trading and light testing. An MT5 farm is a cluster of dedicated servers configured as MetaTester agents for heavy backtests and optimizations, while your terminal remains the master.
Do I need to move my MT5 terminal to your servers?
No. Your MT5 terminal can stay on your workstation or on an existing VPS. You simply add our servers as remote agents in the Strategy Tester. If needed, we can also host the master terminal, but that is optional.
Can I use the farm for optimizations and live trading at the same time?
Technically yes, but it is usually better to reserve one machine for live trading and use the farm mainly for optimizations so heavy tests never affect production.
Where are the servers located and what latency should I expect?
The farm nodes are hosted in modern European datacenters with stable backbone connectivity. Latency matters much less for backtests than for live execution, but the network is kept fast and reliable.
Is there a minimum term? Can I try one month first?
Yes. The minimum term is one month, so you can start with a small farm, run your real optimizations, and then decide whether to continue, scale up, or scale down.
Can I pay for the farm with crypto?
Yes. Winservers.NET supports crypto payments for trading infrastructure, including Windows VPS, dedicated MetaTrader servers, and MT5 backtesting setups.
K.S.M. Trade Sp. z o.o. | NIP/VAT PL5273128546 | REGON 52957272900000 | Address: ALEJA JANA PAWLA II 27 WARSAW, POLAND 00-867