hyper_make

State persistence, prior transfer, and warmup lifecycle. Read when working on checkpoint/, adding new checkpoint fields, debugging cold starts or stale priors, or understanding serde(default) requirements and backward compatibility rules.

Documents auto_derive.rs first-principles parameter derivation from capital and exchange metadata. Use when onboarding new assets, debugging parameter mismatches, understanding why gamma/max_position/target_liquidity have their values, or adding new derived parameters.

WebSocket management, event loop, rate limiting, reconnection, recovery, metrics, and order execution infrastructure. Use when working on orchestrator/, infra/, messages/, core/, fills/, or execution/ modules, debugging connectivity or order placement, adding message handlers, or investigating stale data and latency issues.

Documents the 9 learning feedback loops, SpreadBandit Thompson Sampling, adaptive ensemble, confidence tracking, and baseline tracker. Use when debugging learning behavior, tuning reward attribution, investigating model weight decay, or understanding how fills translate into parameter updates.

Layered risk system with monitors, circuit breakers, kill switch, and position guards. Use when working on risk/, safety/, or monitoring/ modules, debugging position limits, emergency shutdowns, spread widening, or adding new risk monitors. Covers RiskMonitor trait, severity escalation, and defense-first architecture.

Documents the additive spread composition pipeline from GLFT optimal through to final bid/ask prices. Use when debugging wide spreads, investigating spread component contributions, tuning defensive behavior, or understanding why quotes are wider than expected. Critical for incident triage.

Layer 3 optimal sequential decision-making with Bayesian belief tracking, HJB value functions, and changepoint detection. Use when working on control/, stochastic/, or process_models/ modules, debugging quote/wait/pull decisions, modifying the HJB solver, or adding action types. Covers conjugate updates, BOCD, and value of information.

Systematic analysis of model predictions vs realized outcomes. Use when computing Brier Score, Information Ratio, calibration curves, PnL attribution, or conditional calibration by regime/volatility/funding. Identifies which models are adding noise vs value.

Build prediction logging, outcome tracking, and data pipelines for model calibration. Read FIRST before any model, calibration, or signal work. Covers PredictionRecord schema, async outcome matcher, JSONL persistence, and market state snapshots.

Measure predictive information content of signals using mutual information, correlation, and optimal lag analysis. Use before building models, when adding data sources, debugging signal decay, or selecting features. Covers MI estimation, regime-conditional analysis, and decay tracking.

Wire all model components into the quote generation pipeline. Use when building the quote engine, adding new model components, debugging quote generation, or understanding the full data flow from market data through GLFT optimal spread to final bid/ask prices.

Predict informed vs noise trades for dynamic spread/kappa adjustment. Use when building AS prediction, debugging fill losses, adding liquidation cascade detection, or tuning toxic flow response. Covers trade classification, 26-feature engineering, MLP classifier, and real-time integration.

Model fill intensity using Hawkes processes for state-dependent kappa estimation. Use when upgrading from simple fill-rate kappa, building fill probability predictions, incorporating queue position dynamics, or adding Hyperliquid-specific features (funding, OI) to fill rate models.

Exploit Binance-Hyperliquid lead-lag (50-500ms) for microprice adjustment and quote skew. Use when building cross-exchange signals, adding directional microprice adjustment, improving adverse selection detection for arbitrage flow, or monitoring signal R-squared decay.

Bayesian belief tracking over market states (Quiet/Trending/Volatile/Cascade) using Hidden Markov Models. Use when implementing regime-dependent parameters, building aggressive/defensive decisions, debugging regime-specific losses, or replacing hard-coded volatility thresholds with smooth probability blending.

Automated daily health check of all model components for early warning of degradation and drift. Use when setting up daily reports, debugging performance issues, or reviewing model health after deployment. Covers PnL attribution, calibration metrics, signal decay, regime analysis, and alerting.

Structured triage for live market maker incidents. Use when MM stops quoting, position spikes unexpectedly, kill switch triggers, spreads blow up, rate limits hit, or data goes stale. Decision-tree diagnosis with exact file:line references from past incidents.

End-to-end workflow for onboarding a new asset to the market maker. Use when adding BTC, ETH, SOL, or any Hyperliquid perpetual. Covers viability analysis, config generation via auto_derive, spread profile selection, paper validation, and live deployment.

End-to-end workflow for adding a new predictive signal to the market maker

Step-by-step workflow for diagnosing why the market maker is losing money

Pre-flight checks and deployment procedure for live trading. Use when deploying to mainnet, going live with a new asset, or transitioning from paper to live. Covers code parity audit, config validation, first-30-min monitoring, and rollback.

Set up, debug, and validate the paper trading system