A zero-dependency quantitative finance platform for the JVM: research, validation, execution, and connectivity — from walk-forward backtesting to a measured sub-microsecond tick-to-order path, in one Java library.
21 technical indicators (batch + O(1) streaming), strategy DSL, screener, Monte Carlo, ML (GARCH, gradient boosting, 2-state regime HMM).
Nine standard factors, rank IC/IR evaluation, walk-forward and permutation validation, cost-aware factor backtests (spread, impact), sector/beta-neutral construction, decay and attribution reports.
Pair conventions and dual-calendar dates, swap-points curves, FX swaps, NDFs with local fixing calendars, delta-quoted vol surfaces, vanna-volga, digitals/touches/barriers — Monte Carlo cross-checked.
Walk-forward analysis with stitched out-of-sample equity, deflated Sharpe, VaR backtesting (Kupiec + Christoffersen), point-in-time universes with delisting returns for survivorship-honest backtests.
Dynamic benchmark algos (VWAP/TWAP/Arrival/IS/Close/Open/POV that re-decide on live spread, depth, vol, volume curve, impact and alpha), portfolio-level basket scheduling (leg balance + capacity by marginal basket risk from a streaming covariance matrix), smart order routing (readable, zero-alloc, and full-checklist adaptive with lit + dark venues and learned adverse selection), opportunistic liquidity seeking with a completion floor, iceberg, dark pool, and a TCA-style backtest that grades each benchmark algo over a session's bars.
Unified streaming SignalEngine (imbalance, microprice, volatility, liquidity, momentum, composite — equity + FX) plus the models feeding execution: intraday volume/vol/spread seasonality (day-type aware), L2 queue position, iceberg detection, Lee-Ready trade classification, passive-fill probability, prequential-IC-gated online alpha learning, cross-asset lead-lag, Kyle-lambda learned impact, jump-robust (bipower) volatility, Avellaneda-Stoikov optimal quoting, closing auction reserve — with learned state persisted across sessions and the whole loop proven by a five-day overnight simulator.
ITCH-style L3 feed decoding into a full-depth book with exact own-order queue position, consolidated NBBO, order-flow imbalance signals, LULD + market-wide circuit breakers, venue message-rate throttling.
Garbage-free FIX 35=W/X quote decoding, per-LP size-tier ladders with sweep and full-amount pricing, last-look analytics (reject rate, hold, post-reject markout) driving expected-all-in LP routing, symmetric maker-side last look, WMR fixing schedules, direct-vs-synthetic crosses.
Black-Scholes/Garman-Kohlhagen, Black-76, Heston stochastic vol (semi-analytic + Monte Carlo cross-check), American trees, SABR, vol surfaces, higher-order Greeks (vanna, volga, cross-gamma), autocallables (memory coupons, knock-in) with RFQ dealer-panel analytics, options-book Greeks and scenario grids, Monte Carlo hedging error.
One netted view across desks and products: equities cash + options and FX spot/swaps/NDFs/options decompose into a common factor space (currency-level legs, so pairs net their shared USD); inventory-skewed quoting, internalize-or-route economics with client price improvement, cost-aware minimum-variance hedging of the excess, and routing that starts with the firm's own inventory before dark pools and lit — with the diversification benefit priced in the report.
The 14-step workflow: four portfolio VaR flavors with expected shortfall (delta-normal, delta-gamma Cornish-Fisher, Monte Carlo, historical), EVT tail fitting (peaks-over-threshold GPD), rank correlations, PCA and Gaussian/Student-t copulas, stress testing with closed-form reverse stress, FRTB liquidity-horizon ES, Basel traffic light and P&L attribution — mapped end-to-end in MARKET_RISK.md.
Curve bootstrap, day counts and calendars, bond analytics; risk parity, Black-Litterman, constrained optimization, position sizing.
Live WebSocket feeds, FIX 4.4 with resend recovery and persistent sessions, tick capture & deterministic replay, paper venue, live dashboard.
git clone https://github.com/AshJha0/Quant-Finance-Library && cd Quant-Finance-Library
mvn package
java -jar target/quant-finance-library-*.jar backtest \
--csv bars.csv --symbol EURUSD --strategy sma --fast 10 --slow 30 --out report.html