Built around a single question:
what causes quant strategies to fail?
Most systematic trading programs do not die from a lack of edge. They die from overfitting, unmanaged correlation, ungoverned position sizing, and operators who override the system at the worst possible moment. IronCurve is engineered as a direct response to each of those failure modes.
A diversified book, not a single strategy.
IronCurve operates a portfolio of 37 instrument-timeframe combinations across four asset classes. Each combination is evaluated independently, but allocated capital within a unified risk budget. The result is a book where no single instrument, signal, or market regime can dominate the equity curve.
A multi-layer risk governor.
Risk is enforced at four layers, not one. Portfolio-level caps, asset-class caps, timeframe caps, and per-trade sizing all operate simultaneously. When any layer reaches its ceiling, new entries are blocked at that level while the rest of the book continues operating normally. There is no scenario in which the system can take unbounded risk — it is mathematically impossible.
The system’s objective function is not maximum return. It is the smoothest achievable equity curve at the chosen target return. Drawdown control is a first-class design constraint, not a post-hoc overlay.
Fully autonomous. Infrastructure-grade.
The system runs 24 hours a day, five days a week, on a dedicated virtual private server with uptime monitoring and alerting. Execution is handled entirely by code. The human operator monitors the output, not the trades.
Operational discipline is codified: pre-flight checklists, circuit breakers, a daily monitoring log, and a strict rule against intraday intervention. Discipline is not an aspiration — it is a process.
What we optimise for.
Capital preservation precedes return.
A 50% drawdown requires a 100% recovery. Every design choice is evaluated first on its contribution to downside control, and only then on its contribution to upside.
Compounding over time, not speed over it.
The goal is a multi-decade operation, not a year of outsized returns. Consistency and durability outrank peak performance in every parameter decision.
Diversification is load-bearing, not cosmetic.
Four asset classes and 37 instrument-timeframe pairs exist so that any single regime, instrument, or correlated cluster cannot take the book down. Diversification is an explicit risk constraint, not a marketing line.
The operator is a failure mode.
Systematic trading fails most often because a human intervenes. The system is designed to require as little human input as possible, and the operating protocol treats manual override as the most dangerous action available.
Transparency with boundaries.
Methodology, philosophy, and results are published openly. Signal logic, risk-cap specifics, and parameter values remain proprietary. A public record is a credibility asset; giving away the system is not.
Every line of code is auditable.
The entire codebase — signal generation, portfolio governor, execution layer — is documented and audited. No black boxes, no uninspected dependencies. If a decision is made, it can be traced.
A deliberate list of exclusions.
IronCurve is not a high-frequency arbitrage operation. It is not a discretionary macro book. It is not a signals service, a copy-trading product, or an educational platform. It does not promise outsized returns, and it does not take unbounded risk to chase them.
IronCurve is a systematic, multi-asset, risk-first trading operation designed to compound capital over decades with controlled drawdowns. That is the entire mandate.