Projects / Quantitative systems & full-stack simulation workbench

Argus Terminal

Local-first simulation of a Bloomberg-style multi-pane trading, L2/L3 orderbook matching, Monte Carlo VaR/CVaR risk, Black-Scholes-Merton SVI volatility surface, FIX 4.4 / binary SBE wire protocol and Nelson-Siegel-Svensson yield curve workstation built on Node.js, built-in node:sqlite and HTML5 Canvas.

Period

October 2026

Release

v1.0.0

Work type

Independent project

Development

AI-assisted, with verification

01 / Overview

What I built

An educational, single-process Node.js simulation of a multi-asset trading and quantitative analysis terminal (using built-in node:sqlite and a single external dependency, ws, with zero frontend bundler):

  • 01 — Main Trading Terminal & Dual Matching Engines (/): Multi-pane workspace with a <GO> command bar, 6 simulated cross-asset instruments (BTC-USD, ETH-USD, SOL-USD, NVDA-EQ, XAU-USD, IDR-USD), price-time FIFO Orderbook with synchronous SQLite transaction persistence, and a preallocated TypedArray arena SlabOrderbook (capacity = 262,144, intrusive doubly-linked list, O(1) cancel) benchmarked across 250,000 deterministic operations (~2.83M–3.26M ops/sec in-memory vs ~25.9k ops/sec object-array baseline).
  • 02 — Quantitative Risk & Execution Algos: Correlated Cholesky Monte Carlo VaR/CVaR (95% / 99%), Order Book Imbalance (OBI), VPIN-style order-flow toxicity proxy, Stoikov micro-price, OHLCV strategy backtester, and simulated TWAP, VWAP, ICEBERG, and Avellaneda-Stoikov execution schedules.
  • 03 — Derivatives, FIX 4.4 / SBE Wire Codec & Smart Order Router (/options.html): Analytical Black-Scholes-Merton 1st/2nd-order Greeks with verified Put-Call parity (< 1e-8), hybrid Newton-Raphson/bisection IV solver, 5-tenor Gatheral SVI 3D volatility surface, FIX 4.4 parser/builder with Modulo-256 Tag 10 checksum validation, 40-byte binary SBE frame codec, and 4-venue Smart Order Router.
  • 04 — Macro Yield Curve, Bond Desk, LOB Replay & Dark Pool (/macro.html): Nelson-Siegel-Svensson (NSS) yield curve calibration (1M–30Y), fixed-income bond pricing (YTM, DV01, Macaulay/Modified duration, convexity, Z-spread), event-sourced LOB replay time-machine with SHA-256 state hash verification, and a simulated midpoint-peg dark pool (ARGUS-DARK-ATS) with minimum acceptable quantity and VPIN toxicity gating.
Node.jsSQLiteTypedArray / WASMQuantitative FinanceHTML5 Canvas
02 / Evidence

What was checked

Argus Terminal multi-pane trading workspace and L2 order book
Main workspace of vanovri/argus-terminal showing simulated L2 order book depth, OHLCV candlesticks, trades tape, and blotter panes.

Release audit: 25 October 2026. All 6 plain-Node verification suites (64 / 64 assertions across E2E matching, defensive security controls, DevTools layout/a11y invariants, quant risk, BSM/FIX/SBE codecs, and NSS/LOB replay) pass via npm test, alongside a 5-mode 250,000-order ablation benchmark (benchmarks/results/latest.json).

03 / Scope

What this demonstrates

Data-structure layout trade-offs in JavaScript (V8 heap object arrays vs. preallocated TypedArray arenas with intrusive linked lists), deterministic financial and numerical modeling (FIFO matching, BSM Greeks, SVI volatility parameterization, Nelson-Siegel-Svensson yield curves, binary wire framing), and defensive API hardening (parameterized SQLite queries, table allowlisting, SSRF private-IP filtering, and RBAC guards). All market data, venues, and accounts are simulated locally.

Navigate with arrow keys. Enter opens a result. Escape closes search.