How TradeForge Works

TradeForge operates through a structured pipeline: detect opportunities, evaluate risk, execute with discipline, and monitor continuously.

The Four-Step Process

1

Signal Detection

The system monitors funding rates across supported markets. When rates reach statistical extremes (based on 30-day rolling percentiles), a potential trading signal is generated. This is a structural observation, not a price prediction.

2

Risk Evaluation

Before any signal can become a trade, the Risk Engine evaluates current conditions. It checks circuit breaker status, available margin, position limits, and market volatility. If any safety constraint is violated, the signal is rejected.

3

Execution

Approved signals are sized according to the risk budget and current volatility. Every order is placed with a mandatory stop-loss. Position size is capped, never increased during stress conditions. The system executes with discipline, not urgency.

4

Monitoring

Open positions are continuously monitored. Exit conditions include: funding rate normalization, time limits, or risk triggers. Every state change is logged with reasoning. The dashboard provides real-time visibility into system behavior.

The Risk Hierarchy

TradeForge uses a hierarchical architecture where risk control has ultimate authority.

Core principle: The Risk Engine can reject or reduce any trade. Strategy logic cannot override risk logic. If there's a conflict, safety wins.

How decisions flow:

Edges propose — Strategy modules detect signals and suggest trades
Risk Engine evaluates — Checks all constraints and calculates safe position size
Execution follows — Orders are placed only if approved, always with stops

Circuit Breakers

The system includes multiple safety mechanisms that automatically restrict trading when conditions warrant caution.

Caution Threshold: 2%

When daily drawdown reaches 2%, the system increases scrutiny on new positions. Larger trades require additional confirmation. Position sizing becomes more conservative.

Daily Halt: 3%

At 3% daily drawdown, new position opening is suspended. Existing positions continue to be monitored and managed. Trading resumes the next day if conditions allow.

Peak Halt: 8%

If drawdown from peak reaches 8%, all trading activity stops. This is a full stop requiring manual review and acknowledgment before resuming.

Non-Negotiable Rules

These rules are hard-coded into the system. They cannot be disabled or bypassed, even by the user.

Every position must have a stop-loss — No exceptions. Positions without defined exits are not allowed.
Risk beats return — The system is designed for survival first. Optimization for returns comes second.
Uncertainty reduces action — When data is stale, conflicting, or missing, the system tightens rules. It never loosens them.
Crashes enter safe mode — After any system interruption, trading resumes in a reduced-risk state requiring manual acknowledgment.
Edges never auto-enable — Users must explicitly opt-in to any strategy. Nothing activates automatically.

Current Scope and Limitations

Transparency includes being clear about what the system does not do. These are intentional constraints in the current version.

V1 supports:

BTC/USDT and ETH/USDT perpetual futures
Binance exchange only
Funding rate percentile strategy
Web dashboard access

V1 does not include:

Other trading pairs or spot markets
Multiple exchange support
Additional strategy types
Public API access
Mobile trading controls

We start narrow to prove stability across different market conditions before expanding scope. V2 features will only be added after V1 demonstrates reliable behavior.

Logging and Audit Trail

Every decision the system makes is recorded with reasoning. This applies to both trades that were executed and trades that were rejected.

What gets logged:

Signal generation events with market conditions at the time
Risk evaluation outcomes with specific constraints checked
Order placement details including size calculations
Position state changes with trigger conditions
Rejection reasons when trades are blocked

This audit trail serves multiple purposes: debugging unexpected behavior, understanding system decisions, and maintaining accountability. Logs are retained according to tiered storage: 30 days hot, 1 year warm, then archived.