Co-located near the venues
Frankfurt, Tokyo, and New York presence chosen to sit close to the matching engines that matter. Median round-trip stays in the single-digit-ms range on premium tiers.
The compute under it all.
Hosted execution boxes co-located near major matching engines, sized to how loud your strategies are running. Or bring your own. Whatever you pick, the same control plane runs on top - so a strategy can move tiers without rewiring.
Anatomy of a server
Four properties, every tier.
Whatever tier you pick, these four properties stay constant. The differences sit in capacity, latency, and where the platform runs - not in safety primitives.
Single-tenant isolation
Each server runs your control plane in a dedicated namespace, with no neighbor sharing CPU schedulers or NIC queues. Your strategies don't compete with someone else's.
Heartbeat-driven monitoring
Every server emits a heartbeat. A missed heartbeat triggers automatic failover or - on Fund Grade - a manual page to operations. You get told before you notice.
Hardened by default
Reproducible base images, no SSH for end users, signed updates only, full audit log piped to the platform. The same image powers Sovereign Vault.
Pick a server
Four shapes for four levels of latency.
Pick the smallest box that fits your strategies' rhythm - upgrading later carries strategies forward, downgrading is a config change. Latency only matters as much as your strategy demands.
Edge Trader
Edge Trader
Light footprint, sized for retail rhythm.
A small but well-placed box. Two cores, four gigs, a single NVMe partition. Plenty for a couple of medium-frequency strategies on a single venue, with predictable cost. The same control plane the bigger tiers run.
Best for
Solo strategies · paper-mode · low-frequency entries
$29
/mo
- 2 vCPU · 4 GB RAM · 80 GB NVMe
- Frankfurt or Tokyo region
- ~25 ms median round-trip to most major venues
- Up to 3 concurrent strategies
- Automatic failover within 30 seconds
- 99.9 % monthly availability target
Quant Box
Popular
Quant Box
Twelve cores, thirty-two gigs, where most operators live.
The default box for a serious operator. Twelve cores worth of strategy budget, dedicated NVMe, sub-eight-millisecond paths to the major matching engines. Headroom to spin up a backtest while a live strategy runs.
Best for
Multi-strategy operators · hedge desks · serious retail
$129
/mo
- 12 vCPU · 32 GB RAM · 400 GB NVMe
- Frankfurt + Tokyo + New York regions
- <8 ms median round-trip to most major venues
- Up to 12 concurrent strategies
- Live backtest sandbox alongside production
- 99.95 % monthly availability target
AI Orchestrator Box
AI Orchestrator Box
Dedicated AI shards. Self-hosted LLM optional.
Built for desks where the AI ops layer is the workload. Dedicated orchestrator shards for the 5-stage chain-of-prompts pipeline, private MCP gateway, prompt/response retention with hot indexing, and an optional self-hosted local-LLM slot for teams that can't ship traffic to external providers.
Best for
AI-heavy desks · agent-pattern operators · teams with self-hosted LLM requirements
$399
/mo
- 16 vCPU · 64 GB RAM · 800 GB NVMe (AI workload tuned)
- Dedicated AI orchestrator shards: 5-stage pipeline runs isolated from other tenants
- Private MCP gateway + dedicated tool dispatch lane
- Prompt + response audit retention with hot index (90 days searchable)
- Optional self-hosted local-LLM slot (Local flavor) - no external provider traffic
- Includes everything in Quant Box
Fund Grade
Limited
Fund Grade
Bare-metal, dedicated NIC, co-located cage.
For strategies that lose money above two milliseconds. Bare-metal hardware, dedicated network interface, racked inside a co-location facility within metres of the matching engine. Custom configurations available.
Best for
Funds · prop desks · market makers · arb operators
From $899
/mo
- Bare-metal · 32+ vCPU · 128+ GB RAM · NVMe RAID
- Co-located in venue cages where available
- <2 ms median round-trip to the venue's matching engine
- Dedicated 10G network interface
- Manual paging to operations on heartbeat loss
- 99.99 % monthly availability with custom SLA option
Self-Managed
Self-Managed
Bring your own hardware, run our software.
We ship a reproducible image and the same control plane that runs the managed tiers. You bring the hardware, the network, and the rack. Common when you already have co-location, when sovereignty matters, or when latency targets are weirder than what we offer.
Best for
Operators with their own infra · regulated entities · custom latency targets
License
· annual
- Reproducible OCI image + Helm chart
- Same control plane, same TUI, same observability
- Pairs with Sovereign Vault for full self-hosting
- Quarterly platform updates with backwards compatibility
- Direct engineering contact for migration
- No availability SLA from us - your hardware, your call
Side by side
Every tier, same row.
| Edge Trader | Quant Box | Fund Grade | Self-Managed | ||
|---|---|---|---|---|---|
| vCPU | 2 | 12 | 16 | 32+ | Your spec |
| RAM | 4 GB | 32 GB | 64 GB | 128+ GB | Your spec |
| Storage | 80 GB NVMe | 400 GB NVMe | 800 GB NVMe | NVMe RAID | Your spec |
| Latency profile | ~25 ms | <8 ms | <8 ms | <2 ms | Your call |
| Form factor | Container | Container | Container + AI shard | Bare-metal | Your hardware |
| Regions | FRA · NRT | FRA · NRT · NYC | FRA · NRT · NYC | Co-lo cages | Anywhere |
| Concurrent strategies | 3 | 12 | 12 | Custom | Your call |
| Dedicated AI orchestrator shards | - | - | ✓ | Optional | Your call |
| Self-hosted local-LLM slot | - | - | Optional | Optional | Your call |
| Availability target | 99.9 % | 99.95 % | 99.95 % | 99.99 %+ | Your hardware |
Managed vs self-managed
Same image. Different rack.
Managed servers are TradeMire's job to keep alive. Self-Managed lets you bring your own iron and run the same image. Where the box sits is the variable - the platform stays identical.
TradeMire-managed
Edge / Quant / Fund
Customer-managed
Self-Managed
Hardware
Provisioned by TradeMire on vetted upstream providers, tracked by serial number, replaced on failure within hours.
Yours. Could be a tower in a closet, a rack in a co-lo, or a sovereign cloud account - the platform image runs on all of it.
Patching cadence
Security patches applied during low-traffic windows with rolling restart - your strategies don't see the bounce.
On your schedule. We publish CVE notes; you decide when to take the update.
Observability
Telemetry, traces, and logs ship to TradeMire's observability stack - exposed back to you through the platform UI.
Standard OpenTelemetry exporters. Pipe to your own Grafana, Datadog, or whatever your team already runs.
Perimeter
Inside TradeMire's network boundary, accessed through the platform's authenticated control plane.
Inside your network boundary, behind whatever VPN, mTLS, or WAF policy your team requires.
FAQ
The operator's questions.
Does latency really matter for retail-size strategies?
Honestly - usually not. A daily DCA strategy doesn't care if it fires now or fifty milliseconds later. Edge Trader is the right call for that. Latency only starts mattering when you're competing for the same fill against other automated traders, which happens at smaller and smaller AUM each year. If you're not sure, start lower and move up.
What happens to my running strategies if a server fails?
Edge Trader and Quant Box auto-failover within thirty seconds - state is replicated and the new instance picks up where the old one stopped. Fund Grade pages a human in operations on top of that. Self-Managed depends on your own resilience setup.
Can I install custom binaries on the server?
Not on managed tiers - the image is sealed for predictability. On Self-Managed, you control the host, so anything goes. Strategies are described declaratively, so most extensions live in strategy templates rather than binaries anyway.
How is co-location pricing structured for Fund Grade?
The base price covers the bare-metal box and our control plane. Cage fees and venue cross-connects are passed through at cost - we'll quote them per venue when you tell us which matching engines you care about.
Can I move from a managed tier to Self-Managed without redeploying strategies?
Yes. Strategy definitions, vault settings, and operational history are platform-level state - they live in the control plane, not in the server image. Pull the same definitions onto your own image and they keep running.
Provision a server
Pick a region. Pick a tier. Trade.
Provisioning is a guided wizard - region, tier, billing cycle. Most servers come online inside ten minutes. Fund Grade and bare-metal builds we'll quote and turn up by hand.