{"repo":"vivek-v-rao/Conditional-Skew","free":true,"listed":false,"github":"https://github.com/vivek-v-rao/Conditional-Skew","clone":"git clone https://github.com/vivek-v-rao/Conditional-Skew.git","description":"Fit autoregressive models with skewed generalized error distribution (SGED) noise whose parameters vary with level","language":"Python","stars":26,"topics":["ged","implied-volatility","probability-distribution","quantitative-finance","skew","skewness","time-series-analysis","vix","sged","generalized-error-distribution"],"license":"MIT","category":"trading","readme_excerpt":"Conditional-Skew This project fits autoregressive and regime-switching time-series models with non-Gaussian noise, starting from plain AR models and extending to GED, SGED, and level-dependent SGED specifications. The current examples focus on VIX and SPY , with particular attention to whether the conditional distribution of noise changes as VIX changes, and whether latent Markov regimes improve fit relative to a single-regime AR model. In terms of prior literature, there is extensive work on fitting GED innovations to financial returns and a smaller literature on fitting SGED distributions to returns or conditional-volatility models. By contrast, there appears to be comparatively little work on SGED models in which the scale, skewness, and tail-shape parameters are explicit functions of a lagged state variable such as the level or log-level of VIX . That is the main direction explored here. A useful reference for the SGED itself is Panayiotis Theodossiou, Skewed Generalized Error Distribution of Financial Assets and Option Pricing Files From simplest to most complex: - xxar.py Fits ordinary autoregressive models to each time series in a CSV file, prints autocorrelations, and reports residual diagnostics. - ar noise report.py Shared reporting helpers for residual standard deviation, skew, excess kurtosis, and bin-by-bin summaries based on the previous level. - xar ged.py Fits AR models with GED innovations. This allows heavy tails but not skewness. - ar ged model.py Core GED ","default_branch":null,"files":null,"tree":[],"storefront":"/r/vivek-v-rao","claimed":false,"request_supported":{"post":"https://gitbuyer.com/r/vivek-v-rao/Conditional-Skew/request-supported","requests":0},"note":"indexed from public GitHub; nothing is for sale on this page. Clone it from GitHub. Paid listings live at /search."}