Systematic alpha
generation.
Walk-forward validated. Out-of-sample tested. Every metric reported below comes from genuine predictive performance, never fitted to historical data.
Infrastructure at a glance.
Implied Volatility Surface.
The IV surface σ(K, T) maps implied volatility across strike prices and expirations, revealing skew, smile curvature, and term structure dynamics in real time.
Skew
OTM puts trade at higher IV than equidistant calls. The volatility smile is structurally asymmetric.
Term Structure
Short-dated vol captures event risk; long-dated vol reflects structural expectations. Relative value lives in the spread.
Butterfly & Curvature
Convexity of the smile at each tenor encodes tail-risk pricing and dealer positioning.
Monte Carlo path engine.
200 geometric Brownian motion paths simulate plausible trajectories for the underlying asset. Percentile fan bands (5th to 95th) quantify the probability distribution of outcomes.
Regime-conditioned volatility inputs from econometric models
Full re-pricing of multi-leg structures across each path
P&L distribution analysis for position sizing and risk
Tail-risk quantification at 1% and 5% confidence levels
Volatility regime detection.
Econometric conditional volatility feeds a 3-state regime classifier. The model dynamically reconfigures signal weights and strategy parameters based on whether the market is calm, mixed, or stressed.
Structural Volatility Features
Conditional volatility, persistence (α + β), news–asymmetry (γ), and volatility risk premium (IV/RV ratio), all derived from rolling 252-day GARCH estimation windows.
Event-Aware Adaptation
Earnings announcements, central bank decisions, policy meetings, and budget events are proximity-encoded. The model anticipates regime shifts around scheduled catalysts.
Feature concepts.
332 features in the weekly production set, organized into conceptual domains. Each captures a distinct dimension of market microstructure and risk.
Greeks & Sensitivities
Delta, gamma, vega, theta, vanna, charm. Option price sensitivities to spot, volatility, and time, computed at every grid point.
IV Surface Dynamics
Skew slope, butterfly spread, risk reversal, ATM level, IV rank, and term structure curvature. The shape of market expectations.
Open Interest Dynamics
Put-call ratio, OI clustering at strikes, MWPL utilization, and OI change velocity. Positioning reveals directional conviction.
Gamma Exposure & Dealer Flow
GEX measures the gamma exposure of market makers, predicting where hedging flows will amplify or suppress moves.
Microstructure Signals
Bipower variation, jump component detection, tick-rule order flow imbalance. High-frequency signals extracted from 5-minute data.
Macro & Term Structure
Futures basis (contango/backwardation), carry, yield curve regime, VIX correlation, and cross-asset beta. The macro context for stock-level signals.
Strategy combinations.
A GPU-accelerated combinatorial engine evaluates millions of multi-leg options structures, selecting the optimal payoff for each prediction signal.
Futures-settled option pricing. The standard for Indian NSE derivatives. Uses the forward price directly, eliminating dividend and borrow-rate assumptions.
Classic spot-based model with continuous dividend yield adjustment. Used when futures data is unavailable.
Implied volatility extraction via Brent's method, a hybrid bisection/secant solver with superlinear convergence, searching over [10⁻⁶, 5.0] with 10⁻⁷ tolerance.
Full analytical Greeks: Delta, Gamma, Vega, Theta, Vanna, and Charm, computed at every strike in the chain for position-level risk decomposition.
Grid Construction
All valid strike×type multi-leg combinations enumerated from the live chain
Scenario Matrix
P&L re-priced across ±10% spot grid using Black-76 with vol shock
IV Smile Calibration
4 stickiness models: strike, delta, symmetric, hybrid
Ranking & Selection
Filtered by risk constraints, ranked by risk-adjusted return
Every structure is evaluated for profitability across both positive and negative spot moves. The engine selects strategies that profit from the magnitude of the move, not its direction. Defined maximum loss. No naked exposure.
Walk-Forward Validation
Every prediction is made strictly out-of-sample. 78-week rolling training windows advance weekly, ensuring the model never sees future data. 359 sequential validation folds across 10 years of data.
Regime-Adaptive Architecture
The model detects whether the market is in a calm, transitional, or stressed regime and dynamically reconfigures. Different market environments demand different signal weights and risk profiles.
Non-Directional Edge
We trade volatility magnitude, not direction. Options structures are designed with defined maximum loss and non-directional payoff profiles. Alpha comes from predicting which stocks will move, not how.
No Overfitting Guarantee
Magnitude-weighted sample importance with temporal decay. No hyperparameter tuning on test data. No p-hacking. Every reported metric is from genuine out-of-sample walk-forward predictions.
All results presented are based on simulated walk-forward backtesting and do not represent actual trading performance. Past or simulated performance is not indicative of future results. AnchorLane Capital is currently in the research and development stage. No investment advice is being offered. Options trading involves substantial risk and is not suitable for all investors.