ShreyasSomnathe
Quantitative Researcher & Trader building ML-driven systematic strategies for global energy and index futures
Sharpe Ratio
Live Strategies
Projects
What I've Built
Production trading systems, research platforms, and quantitative infrastructure for energy futures markets
Crude Oil Forward Curve ML Trading System
Production Trading System
Production ML system for crude calendar spread trading using PCA of forward curve (level/slope/curvature), VAR for lead-lag relationships, and Kalman filtering for regime detection. Integrated EIA, FRED, Cushing inventory, and refinery data across 3 live strategies with FastAPI execution and PostgreSQL.
sharpe
3.4
strategies
3 Live
assets
Energy Futures
Quantitative Trading Analysis Engine
Research Platform
Python platform for high-frequency CL futures analysis (1.44M rows, 842 patterns) with Shannon/Renyi/Tsallis entropy, chaos theory (Lyapunov exponents, Grassberger-Procaccia), Markov chain models, and network topology analysis. Risk frameworks: Kelly Criterion, VaR/CVaR, Sharpe optimization.
rows
1.44M
patterns
842
methods
10+
assets
CL Futures
Energy Trading Infrastructure & Dashboard
Full-Stack Platform
Full-stack trading infrastructure with Python strategy engine (mean-reversion, momentum, volume breakout), FastAPI + WebSocket backend, and React dashboard with live P&L, order blotter, TT API integration, strategy metrics, and system monitoring.
strategies
3 Types
latency
Sub-sec
monitoring
Live
api
TT API
CL Intraday Spread Research
Quantitative Research
18.3M minute-level bars across WTI crude oil forward curve. 37 hypotheses tested with institutional-grade methodology. Discovered bifurcated market structure and butterfly reversion strategy with abundant signal capacity.
bars
18.3M
hypotheses
37
signals
7.4K/yr
methods
12+
Research
CL Intraday Forward Curve Microstructure
How does information propagate across the crude oil forward curve within a single trading day? This framework analyzes minute-level dynamics across WTI outright + calendar spreads, revealing a bifurcated information structure with exploitable divergences.
Bifurcated Information Structure
Hurst < 0.5 vs > 0.5
Beta Magnitude Collapse
Daily 2.5 → Intraday 0.02
Session Energy Cycles
63-90% Compression Rate
EIA Paradox
Amplify OR 1.48x, Suppress 1M 0.74x
Butterfly Reversion
2.3 min half-life (2,400x faster)
Trade Archetype Winner
7,400 signals/yr | Top Archetype
Quantitative Methods
Visualizations
22 plotsBeta Analysis
4 plotsBeta Convergence by Timescale
Nonlinear Beta by Volatility Quintile
Failed Propagation Hit Rates
Session-Conditional Beta
Lead-Lag
2 plotsCross-Tenor Correlation at Minute Lags
VAR Impulse Response Functions
Regimes
3 plotsRolling PCA Explained Variance
Regime Classification Timeline
Regime Transition Probabilities
Volatility
3 plotsSession Energy Buildup Curves
Bar Type Distribution
Hourly Volume Profile
Events
3 plotsSession Volatility Comparison
EIA Volatility Ratio Paradox
Settlement Window Volatility
Fractal
3 plotsHurst Exponents by Tenor & Timescale
Butterfly AR(1) Decay
Wavelet Energy Distribution
Advanced
3 plotsCorrelation Breakdown Forward Returns
Shannon Entropy Evolution
Snap-Back Reversion Rates
Strategies
1 plotsTrade Archetype Comparison
Background
Experience & Education
Quantitative Researcher / Trader
Blueberry Capital | Gurgaon, India
- •Developed ML models for trading crude oil forward curve, achieving portfolio Sharpe ratio of 3.4 through systematic signal generation across multiple contract maturities and term structure regimes
- •Built and deployed 3 live strategies integrating alternative data: EIA weekly petroleum reports, FRED macroeconomic indicators (GDP, industrial production, rates), and proprietary market microstructure features
- •Architected end-to-end execution infrastructure in Python (FastAPI, WebSocket) for automated order generation, risk checks, position management, and real-time P&L monitoring
- •Engineered feature pipelines processing energy fundamentals, forward curve dynamics, inventory surprise signals, and macro regime indicators with ensemble ML (XGBoost, neural networks)
Quantitative Trader
Futures First | Hyderabad, India
- •Developed statistical models for energy futures (crude oil, natural gas, heating oil, RBOB, gasoil) generating alpha across multiple contract maturities
- •Executed algorithmic strategies: event-driven, trend-following, and mid-frequency arbitrage boxes for consistent daily profits
- •Applied regression analysis for hedge ratios across Abu Dhabi crude, Brent, WTI; Monte Carlo stress testing; seasonal analysis for strategy optimization
- •Managed multi-commodity portfolio (intraday to seasonal): mean-reverting, trend-following, arbitrage, inter-product and calendar spread strategies
B.E. in Computer Science
Nagpur University
Nagpur, India | 2018 - 2022
CGPA: 9.2 | Coursework: Machine Learning, Deep Neural Networks, Data Mining, AI
Executive Program in Algorithmic Trading
QuantInsti Quantitative Learning
Mumbai, India | Jan - Jun 2022
Quantitative Trading, ML in Finance, HFT, Risk Management, Derivatives
Certifications
Technical
Skills & Tools
Programming
Quantitative
ML & Methods
Infrastructure
Contact
Let's Connect
Open to quantitative research roles, trading opportunities, and collaboration in systematic strategies