quant-statistics

Testing financial time series for stationarity involves running ADF and cointegration tests to ensure series are suitable for regression. This skill applies statsmodels-backed ADF tests to asset returns, providing test statistics and decision thresholds to confirm data stationarity before deploying capital.

Yes, you can use GARCH models to forecast conditional volatility for crypto or equity returns. This skill fits GARCH(1,1) plus variant summaries to forecast conditional volatility, persistence, and leverage effects, generating interpretable risk signals computed via the arch package.

To fix heteroskedasticity and autocorrelation in regression residuals during factor model backtesting, apply Newey-West or WLS remedies. This skill evaluates residual diagnostics and deploys these corrections to ensure statistical significance of factor premiums.

Calculating bootstrap confidence intervals for Sharpe ratios is best achieved through numpy-backed resampling of asset returns. This skill provides bootstrap analysis to generate defensible metrics for strategy insights and risk factors rather than relying on intuition alone.

Yes, this statistical testing suite supports cointegration and Granger causality tests for pair trading. It runs these tests to validate pair-trading strategies and interprets z-score signals for spread trades, ensuring quantitative research is grounded in rigorous metrics.

High persistence in volatility modeling indicates that shocks to volatility decay slowly over time. The leverage effect in GARCH summaries shows that negative returns increase future volatility more than positive returns, with both metrics computed by the arch package to provide interpretable risk signals.