Quant Research for Prop Trading - FICC
Factor Storage
DolphinDB provides purpose-built storage for FICC data structures — supporting both curve key point data (tenor-based yields) and curve-fitting parameters (Nelson-Siegel and similar models), with full historical tracking of curve evolution for rapid reconstruction of any past curve shape for backtesting and scenario analysis.
Spread data, macro indicators, and credit ratings are co-managed with high-frequency market data in a unified store, with automatic cross-frequency alignment and interpolation. Researchers can efficiently run cross-instrument, cross-tenor, and cross-market historical spread analysis to inform trading strategy development.
Factor Computation
DolphinDB delivers professional fixed income analytics for FICC research — with real-time parallel curve fitting and spread computation across multiple yield curves. Supported construction methods include Nelson-Siegel, cubic spline, and piecewise linear, covering government bonds, credit bonds, SHIBOR, and more. Cross-instrument, cross-tenor, and cross-market spread analysis and carry calculation are built in.
A full computation framework handles FICC-specific term structure factors — level, slope, and curvature — alongside credit spread factors and macro factors. Researchers can implement factor logic for rates strategies, credit spread arbitrage, and basis trading, with native integration of macroeconomic data for scenario analysis and stress testing.
Historical Pricing & Valuation
DolphinDB provides comprehensive historical pricing capabilities for FICC research, addressing the price gaps common in illiquid bond markets. Multiple valuation methods support both backtest execution pricing and mark-to-market position valuation.
Built-in curve construction methods — Bootstrap, Nelson-Siegel, and Nelson-Siegel-Svensson — efficiently build spot rate curves from observed yield-to-maturity data, with cubic spline and linear interpolation available for deriving rates at arbitrary tenors. For non-traded bonds, discounted cash flow pricing off the spot curve supports configurable adjustments for credit spread and liquidity premium. Where comparable bonds exist, a rules-based engine automatically identifies reference bonds by issuer, tenor, and credit rating for rapid relative value pricing.
Featured Tutorials & Best Practices:
Unified Stream-Batch Factor Development
DolphinDB brings unified stream-batch curve analytics to FICC research. Curve fitting algorithms and spread computation logic developed in the research environment deploy directly to live data streams — updating curve shapes and trading signals intraday in real time. Backtest and production environments share identical curve construction methods, eliminating strategy degradation caused by implementation discrepancies.
Historical batch curve reconstruction and real-time incremental updates run on the same framework — researchers rapidly evaluate the impact of different fitting methods on strategy performance and deploy the optimal approach to production. A single codebase spans the full pipeline from macro strategy research to trading signal generation, ensuring seamless research-to-production transfer.
Model Training & Inference
DolphinDB provides time-series-native modeling and inference capabilities for FICC research across rates, credit, and FX. Built-in ARIMA, GARCH, and factor modeling tools cover yield curve forecasting, spread regression, liquidity factor construction, and term structure modeling. XGBoost and other ML frameworks integrate via plugins, and large-scale Monte Carlo simulation runs natively within the system for options pricing and risk measurement.
DolphinDB's LibTorch plugin loads PyTorch-trained LSTM and Transformer models for high-performance inference. Combined with AI DataLoader's efficient organization of large-scale historical time-series, the system supports deep learning-driven research including macro time-series forecasting, cross-asset co-movement analysis, and credit indicator prediction — seamlessly integrated with DolphinDB's factor computation and real-time curve update pipelines.
Factor Lab
A fully integrated platform for the entire factor lifecycle — from data management and research to evaluation and strategy backtesting. Every factor's computation logic, data dependencies, historical performance, and applied strategies are automatically tracked, with full lineage tracing and impact analysis.
Teams share a unified view of IC, IC_IR, long-short returns, and other key metrics across the full factor library — enabling performance comparison, portfolio optimization, and collaborative research without duplicated effort.
Request a free trial → https://dolphindb.com/product#starfish
Featured Tutorials & Best Practices:
Backtesting Framework
An integrated event-driven backtesting framework combining data replay, simulated matching, and strategy execution in one engine. Historical data streams simultaneously to the matching engine and strategy callbacks — strategies generate orders based on spread movements, curve shape signals, and other indicators, passing through DV01, duration, and concentration risk checks before entering the matching engine. Bid-ask spreads, liquidity constraints, and execution latency are accurately modeled, with positions, valuations, and risk exposures updated in real time.
Built on a C++ core and covering rates, government bond futures, interest rate swaps, FX, and credit bonds, DolphinDB delivers up to tens of times the performance of Python-based frameworks like Backtrader and MetaTrader4 at full-market, mid-to-high frequency scale.
Order Matching Simulation
DolphinDB supports custom matching simulation for FICC strategies, purpose-built for OTC bond market dynamics. Execution is simulated using historical trade data and two-way dealer quotes across varying liquidity regimes.
Liquidity Tiering: Configurable liquidity parameters by bond type, remaining maturity, and credit rating — on-the-run Treasuries execute at tight spreads with immediate fill; semi-liquid instruments carry partial fill ratios or execution delays; off-the-run and illiquid credit bonds are assigned longer queue times with mark-to-market exposure modeled during the waiting period. Bid-ask spreads adjust dynamically — widening under elevated volatility or applying incremental market impact costs for large block orders.
Financing Cost Integration: Financing costs are derived from position data and historical repo rates, with configurable repo availability constraints and leverage limits. For interest rate derivatives, matching simulation combines theoretical pricing models — Black-Scholes, Hull-White, and similar — with historical bid-offer spreads.
AI-Powered Investment Research
DolphinDB's AI research system integrates large language models into the FICC quantitative workflow, enabling analysts to complete complex analytical tasks through natural language. The FICC Analytics Agent handles pricing and risk metric computation across rates, credit, and options — automatically interpreting requests, selecting the appropriate model, fetching data, and returning results. The Portfolio Management Agent queries portfolio duration, convexity, and credit exposure, and generates structured analytical reports.
The system synthesizes macro data, monetary policy signals, and live market data across multiple sources to power intelligent scenario analysis and stress testing — bringing institutional-grade FICC analytics within reach of non-technical researchers, while giving senior quant teams a faster, smarter research workflow.