Choose Your Trading Desk Simulation
Institutional HFT StatArb Engine
Cross-venue statistical arbitrage on crypto pair assets (BTC/USDT & ETH/USDT). Candidates engineer an adaptive 2D Kalman filter for dynamic hedge ratio estimation, execute Half-Kelly sizing, and defend a strict 3% trailing drawdown limit.
β_t = β_{t-1} + w_t • Z_t = (Spread_t - μ_t) / σ_t
btcusdt_ethusdt_ticks_5000.csv (5,000 L2 Ticks)Avellaneda–Stoikov Market Maker
Ultra-high frequency limit order book quoting with inventory penalty risk aversion. Dynamic reservation price calculations, L2 queue priority modeling, and adverse selection mitigation against informed liquidity takers.
r(s,q,t) = s - qγσ²(T-t) • δ&supa;+δ&supb; = (2/γ)ln(1+γ/κ)
btc_l2_microstructure_ticks.csv (Order Book Snaps)Dynamic Jump-Diffusion Hedger
Options surface market making with Merton jump-diffusion shocks. Candidates engineer real-time Delta-Gamma-Vega risk neutral hedging under stochastic volatility and extreme Poisson jump arrival regimes.
∂V/∂t + ½σ²S²∂²V/∂S² + λE[ΔV] = rV (Merton Jump)
btc_derivatives_jump_surface_ticks.csv (8K Ticks)Hierarchical Risk Parity (HRP)
Machine learning unsupervised graph tree clustering for robust covariance allocation. Candidates bypass Markowitz matrix inversion instability by engineering single-linkage dendrogram clustering and recursive bisection.
d_{i,j} = √(½(1 - ρ_{i,j})) • Recursive Bisection Weights
multi_asset_covariance_matrix.csv (Cross-Asset Series)Proprietary Trading Desks Side-by-Side Deep-Dive
Compare mathematical complexity, data ingestion frequency, risk constraints, and audited performance benchmarks across all 4 flagship desk environments.
| Desk & Mandate | Asset Class Traded | Data Resolution | Core Mathematical Engine | Risk Guard Limits | Target Sharpe | Action |
|---|---|---|---|---|---|---|
|
SIM 01
Millennium / Jane Street
Statistical Arbitrage
|
BTC/USDT • ETH/USDT | 100ms L2 Ticks | Engle-Granger • 2D Kalman Filter | ≤ 3.0% Trailing DD | 3.92 (+18.4%) | Solve Desk 01 → |
|
SIM 02
Citadel Securities
High-Frequency MM
|
BTC/USDT L2 Depth | Sub-Millisecond Snaps | Avellaneda-Stoikov HJB PDE | ±10 Units Cap | 4.18 (+24.3%) | Solve Desk 02 → |
|
SIM 03
Optiver / IMC Desk
Options Volatility
|
Deribit Options Surface | Real-Time Vol Ticks | Merton Jump • Greeks Hedging | ≤ 0.05 Net Vega | 3.78 (+21.6%) | Solve Desk 03 → |
|
SIM 04
Two Sigma / AQR
Quantitative Allocation
|
20-Asset Global Matrix | Intraday Covariance | Hierarchical Graph Clustering | κ < 85 Condition No. | 3.15 (+16.8%) | Solve Desk 04 → |
The 5-Module Trading Desk Pipeline You Will Engineer
Every desk simulation requires building a full production-grade trading system, from raw tick ingestion to autonomous risk verification.
L2 Market Ingestion
Streaming WSS ticks & Snappy Parquet tick reader with zero buffer loss.
C++ WASM • ArrowState Estimation
Adaptive 2D Kalman filter, HJB reservation pricing, or Merton jump shock state.
NumPy • SciPyAlpha & Signal Engine
Spread Z-score thresholding, order book queue priority, & volatility skew pricing.
Vectorized • NumbaRisk Guard & Sizing
Half-Kelly sizing, inventory penalty limits, Vega delta neutrality, and trailing DD guards.
Hard Risk EnforcerGround Truth Matcher
Autonomous backtest evaluation matching desk ground-truth within ±2.50% variance.
Verified Talent DossierGeneric Coding Sites vs. TrainQuants IRL Simulations
Why quantitative hedge funds and proprietary market makers recruit from TrainQuants over traditional coding platforms.
LeetCode / HackerRank / Codeforces
- Isolated 5-line toy puzzles (Invert binary trees, palindrome string matching).
- Zero market microstructure (No order books, no tick queues, no adverse selection).
- Abstract time limits (O(N) algorithms without hardware or network latency bounds).
- No architectural engineering (Single function submission, no modular pipelines).
- Ignored by Quant Recruiters (Doesn't evaluate real-world alpha generation capability).
Proprietary Floor-Ready System Builder
- End-to-End Trading Desk Systems (Data ingestion → Architecture → Modular code → Verification).
- Institutional Market Data APIs (REST & WSS feeds, Snappy Parquet, L2 ticks, Candidate API keys).
- Sub-Millisecond Latency Budgets (Audited microsecond CPU processing limits per tick).
- Real Strategy Signal Briefings & Charts (Visual trigger points, Z-scores, entry/exit markers).
- Verified Talent Passport Dossier (Audited Sharpe, Drawdown, and PnL stamped for recruiters).
The 4-Step Engineering Protocol Inside Each Sandbox
When you click "Solve in Sandbox", you enter an isolated institutional quantitative lab structured across 4 progressive steps:
Problem Briefing & Data API Hub
Access your authorized sandbox API key, download raw L2 tick datasets (CSV/Parquet), review the detailed strategy mandate, inspect entry/exit signal briefs, and study the interactive sample chart with target benchmark stats.
System Architecture Blueprint
Inspect the multi-module trading architecture graph. Analyze data-flow transitions from raw market feeds through state estimation, signal generation, risk filters, and order execution.
Interactive Python Code Studio
Implement the strategy in our browser-based IDE across 5 modular components. Run live test cycles, inspect execution logs, and tune parameters against latency budgets.
Desk Ground Truth Matcher
Backtest your engineered code against desk benchmarks within ±2.50% tolerance. Watch real-time tick-by-tick market stress-tests and earn your verified candidate dossier.