SigmaX
System

From signal to seal, in five stages.

SigmaX is built on XEOS — an execution intelligence engine that treats every decision as a causally-linked event. Here is what happens between a signal firing and a fill being booked against your account.

01

Signal

A strategy identifies an opportunity — with a confidence score, a proposed size, and a regime fingerprint.

  • Signals are versioned and tracked by trace_id from generation through disposition.
  • Each signal carries its model version, input features, and decision rationale.
  • Signals that fail regime gates are logged with the gate that rejected them and why.
02

Gate

Before reaching the market, every signal passes through a sequence of risk and regime filters.

  • Volatility regime check — is current vol within the strategy's operating envelope?
  • Liquidity check — can the requested size be filled within the slippage budget?
  • Drawdown limit check — is the account within its risk parameters?
  • Protection mode override — if triggered, only exits are allowed.
03

Execute

Orders route to venue. Every microsecond of the execution path is captured.

  • Requested price, fill price, and timestamps recorded at nanosecond resolution.
  • Slippage (execution cost) and fees recorded per fill.
  • Missed fills recorded as "opportunities not executed" — with the reason.
04

Account

Realized PnL is decomposed against the theoretical ideal — the shadow portfolio.

  • Shadow PnL: what you would have made with frictionless, instantaneous execution.
  • Executable PnL: what you actually made.
  • Leakage = Shadow − Executable, itemized by cause.
05

Audit

Every decision, gate, fill, and reconciliation is sealed into an immutable audit record.

  • Each record carries a parent chain — you can trace a fill back to the signal that created it.
  • Records are cryptographically signed. Tampering is detectable.
  • The client portal exposes the audit drawer for any trade on demand.

Language

Internally, engineering uses precise technical terms. On the client side, we translate those into the outcomes they describe — without losing any information.

InternalClient-facing
slippageExecution cost
missed fillsOpportunities not executed
regime killProtection mode
leakageEfficiency loss
shadow_pnlTheoretical PnL
executable_pnlRealized PnL
adverse selectionMarket-timing cost