Replay · time
FerroReplay
Deterministic time, replay, and journal primitives that make time an explicit dependency in reproducible systems.
Tier 02
PLATFORM
Anvil
The deterministic runtime.
Anvil composes MorphIQ Labs infrastructure into journaled runtime workflows for strategies, replay, and operations. Signals, valuations, intents, risk decisions, and fills are recorded as reproducible state transitions; if a result cannot be replayed, the platform treats it as a defect.
Runtime discipline
Determinism is the design.
Anvil makes reproducibility part of the runtime contract. Inputs are sequenced, time is explicit, state changes are journaled, and replay has to arrive at the same result. Where primitive laws can be specified, proof gates sit beside replay, idempotency, and regression tests.
Journal
Every external input is sequenced and written to an append-only journal. The journal is the system of record; everything else is derived.
Replay
Given a journal segment, Anvil replays the state transitions that produced a result. The path is derived from recorded events, not guessed from downstream state.
Strategy SDK
A narrow, typed surface for writing strategies. Strategy code emits intents through the SDK boundary; risk, execution, journaling, and replay stay under platform control.
Composition
Reusable components. Runtime discipline.
Anvil is the composition layer for MorphIQ Labs components. Matching, pricing, signal analysis, feature extraction, market data, and replay stay reusable underneath; the platform assembles them into journaled workflows that behave the same way under test, under load, and across environments.
Matching
FerroMatch
Replay-exact limit-order-book matching with price-time priority, configurable tick semantics, and explicit tiebreakers.
Pricing / risk
FerroRisk
Pricing, implied-volatility solving, Greeks, volatility surfaces, and portfolio risk measures with formal proof gates around stable primitive laws.
Signal · ML features
FerroWave
Wavelet feature extraction and signal analysis for non-stationary time series, with streaming features and regime primitives.
Market data
FerroFeed
Market-data normalization with provider-agnostic event streams, quality gates, and replay semantics.
Who Anvil is for
For teams that own the runtime.
Anvil is built for the people responsible for systematic workflows after research leaves the notebook: platform engineers, strategy developers, operators, and risk reviewers. It gives them a shared runtime contract for backtests, paper runs, production runs, replay, and audit.
Systematic operators
Need deterministic behavior across staging, production, and recovery. Anvil's journal and replay model is the substrate.
Strategy developers
Need research code and production code to run against the same runtime contract. The SDK boundary keeps strategy intent separate from risk, execution, and journaling.
Risk and compliance
Need execution records that can be replayed and reviewed on demand. The journal carries the event path; nothing is guessed from downstream state.
Technical surface
The runtime has its
own technical home.
This overview stays at the portfolio layer. Anvil's technical surface covers runtime architecture, the strategy SDK, journal and replay semantics, operational runbooks, and evaluation materials for teams going deeper.