Small quant teams
Teams of 2–10 moving beyond scripts, fragmented credentials and exchange-specific operations into private environments, unified risk and paper trading.
AI + exchange-neutral infrastructure
Let AI collaborate.
Keep trading controlled.
Project Anchor brings multi-model AI collaboration, strategy research, unified risk controls, auditable execution and operations into one private infrastructure. Providers and exchanges can change; control stays with the system owner.
Private deployment · Customer-owned accounts · No custody · No return promise
Built for professional teams that need controlled infrastructure.
The first customers are not beginners looking for signals or promised returns. They are teams with research, strategies or client projects that need to turn scattered scripts into a safe, auditable system.
Research desks and funds
Research desks, family offices and digital-asset funds that need roles, data boundaries, multi-strategy governance, evidence and optional local AI.
Fintech and software firms
Software and technical service firms that want reusable exchange adapters, risk controls and a private or white-label operating layer.
One control plane across the trading lifecycle.
The platform is not tied to one exchange and never collapses strategy logic into an order interface. Every step has a boundary, explicit state and reviewable evidence.
Strategy research
Normalized market models, candidate evaluation, stressed costs, out-of-sample observation and paper-trading preparation.
Risk control
Position, exposure, daily loss, drawdown, frequency, permissions and circuit breakers are enforced before the exchange boundary.
Auditable operations
Health, alerts, read-only reconciliation, approvals and operational evidence are visible in one maintained surface.
Work with more than ChatGPT—without giving any AI control of trading.
The planned AI Provider Gateway can connect OpenAI, Anthropic Claude, Google Gemini, Ollama/local models and compatible services. Customers can bring their own keys, deploy locally or route by task.
What AI may do
OpenAIClaudeGeminiOllama / Local
01Explain risk decisions and anomalies without changing rules
02Turn natural language into reviewable research experiments
03Compare strategies, summarize incidents and draft reports
04Route providers by data class, cost and latency
What AI must never do
×Read exchange secrets, signing material or full credentials
×Bypass deterministic risk, human approval or circuit breakers
×Create, sign or submit a production order
×Present model output as promised returns or investment advice
Current status: no model runtime is embedded today. The provider gateway is an implementation roadmap; architectural compatibility is not the same as completed provider certification.
A separated path from signal to execution.
Strategies express intent, deterministic risk decides whether to continue, adapters handle venue differences, and audit records the outcome.
01
Normalize data
Map markets, assets, accounts and instruments.
02
Strategy intent
Research logic never holds trading credentials.
03
Unified risk gate
Block before any external request.
04
Adapt and audit
Track venue calls, receipts and evidence.
Exchanges are adapter targets, not product boundaries.
Initial adapter work targets Kraken and Binance. The same safety contract can extend to additional exchanges, brokers and data sources.
Kraken
Adapter target- Read-only account and public-market access
- xStocks research data and simulation preparation
- Permission, geography and capability checks
Binance
Adapter target- Sandbox and execution-path validation
- Normalized order, receipt and error mapping
- Production execution requires separate authorization
Project Anchor is independent software infrastructure. Exchange names identify technical adapter targets only and do not imply endorsement, authorization or partnership.
Safety is architecture, not a feature toggle.
Credential isolation, least privilege, IP restrictions, approval, circuit breakers and read-only reconciliation form the control boundary. The public site reads no account data and exposes no order controls.
Private deployment
Each customer owns an isolated runtime, account connection and data boundary.
Least privilege
Research, read-only and execution credentials remain separate, with scenario-specific IP and permission controls.
Evidence driven
Runtime state, approvals, risk decisions and external receipts form reviewable evidence.
Demonstrate risk discipline before subscribing.
The simulated evaluation uses deterministic rules, a fixed UTC day boundary and immutable rule snapshots. Reaching the objective still requires human review; passing certifies risk readiness only and does not provide capital, a profit split or live access.
$10,000 simulated budget
8% objective · 2% maximum daily loss · 5% maximum total drawdown · at least 5 active days · best day capped at 40% of total profit.
$50,000 simulated budget
8% objective · 2% maximum daily loss · 5% maximum total drawdown · at least 7 active days · best day capped at 35% of total profit.
$100,000 simulated budget
7% objective · 1.5% maximum daily loss · 4% maximum total drawdown · at least 10 active days · best day capped at 30% of total profit.
The evaluation engine is implemented and covered by automated tests, but public paid enrollment remains closed until identity, payment, refund, appeal and target-jurisdiction legal reviews are complete.
Charge for verifiable value.
International plans are priced in USD. Evaluation, software subscription and private deployment remain separate; a passed evaluation can be credited toward an annual software contract within 30 days. We do not charge by AUM or share trading returns.
Free Readiness
$0Product tour and sample check
- Public rulebook
- Delayed-data sample assessment
- No certificate or persistent account
- A low-friction first look
Risk Evaluation
$49–$349per simulated evaluation
- Simulated risk budget and fixed rules
- Maximum daily loss and total drawdown
- Audit evidence and human review
- No capital or profit split
Professional subscription
$99/monthor $990/year
- 1 user and 1 research workspace
- 1 read-only exchange connection
- Backtesting, paper trading and risk controls
- 30-day audit history
- AI Gateway BYOK
Team and private deployment
from $299/monthEnterprise deployment by quote
- Team permissions and approvals
- Multiple strategies and adapters
- Private VPC, on-premises and SSO
- Custom SLA and security review
- Optional managed operations
Prices exclude tax, cloud resources, GPUs, third-party AI and licensed market or historical data. AI defaults to customer-owned provider keys. Evaluations run only in simulation and provide no company capital, live-trading access or return promise.
Commercial pilot
Start with a risk-readiness assessment.
Review the public rules and complete a simulated risk evaluation before choosing the $99/month software subscription or a private team deployment. Connections use customer-owned accounts and credentials; we do not custody funds or promise returns.
Stage oneFree readiness check
Stage twoSimulation and risk certification
Stage threeSubscription or private deployment
Business contactFree forwarding for hello@anchor-infra.com is being activated. Do not send sensitive information until activation is confirmed.