Research on financial models, market regimes, institutions, and investor decisions.
- Executed investigations
- 06
- Empirical studies
- 04
- Numerical validations
- 02
- Interactive models
- 05
Research program
Contents
- 01Portfolio construction & market riskFindings and their decision boundaries.
- 02Derivatives & computational financeValidation studies and model alternatives.
- 03Factor exposure & empirical attributionRegression output, holdout stability, and model risk.
- 04Market regimes & institutional transmissionFramework, synthesis, and proposed research.
Published research
Seven public results from six executed investigations
All seven results →Asset pricing · Numerical verification
Cross-method option-pricing validation
Numerical validation
Black–Scholes, CRR, PDE, and Monte Carlo prices agreed within declared numerical tolerances.
≈10⁻³ tree/PDE error
Inspect the validation record →Portfolio construction · Estimation risk
Portfolio optimization under estimation error
Empirical finding
The fitted max-Sharpe portfolio weakened materially on later data.
Sharpe 1.12 → 0.44
Inspect the investigation →Market risk · Forecast calibration
Historical VaR under coverage and independence tests
Empirical finding
Breaches arrived too often and in clusters; the forecast failed its calibration tests.
135 breaches · 107 expected
Inspect the investigation →Derivatives · Stochastic assumptions
Constant-volatility dynamics under empirical tests
Empirical finding
Observed volatility and option prices contradicted the tested constant-volatility dynamics.
≈3% → 93% realized volatility
Inspect the investigation →Cross-asset regimes · Diversification
Stock–bond diversification across changing correlation environments
Empirical finding
The measured stock–bond relationship changed sign.
−0.35 → +0.37 correlation
Inspect the investigation →Factor models · Product inference
ETF factor exposure and alpha after multiple testing
Empirical finding
Advertised exposures appeared; no tested alpha survived multiple-testing correction.
0 alphas at 5% FDR
Inspect the investigation →Numerical methods · Simulation error
Monte Carlo convergence, bias, and variance reduction
Numerical validation
Sampling error followed the N⁻¹ᐟ² law; enough paths is defined by the decision's tolerance.
Error ∝ N⁻¹ᐟ²
Inspect the investigation →Working instruments
Interactive models
Five implementations run live in the browser. Change an assumption; watch the answer move.
Markets and institutions
Regimes rewire how shocks move through markets
Interpretation standard
A quantitative conclusion is inseparable from its model, data vintage, evaluation design, and error estimate. Confidence is warranted only within those boundaries.
Research standard