rough-vol-forecastlisted

# rough-vol-forecast You hand over a ticker and a set of forecast horizons (default 1, 5, 20, 60, 120 trading days). The skill fits daily-return realized vol on a 2-year window, then applies three vol-scaling models across each horizon: - **Traditional Brownian**: sigma(h) = sigma_daily × sqrt(h). Standard sqrt-time scaling. - **EWMA (RiskMetrics)**: same sqrt-time scaling but on a decay-weighted vol estimate that responds faster to recent regime. - **Rough vol (Bayer-Friz-Gatheral 2016)**: sigma(h) = sigma_daily × h^H with H = 0.14 (Livieri et al. 2018 empirical default). Damps long-horizon growth substantially. Answers "how much does horizon really matter for vol?" — which turns out to be the big 2024-25 vol modeling debate. ## When to invoke - "What's my 60-day forward vol on SPY?" - Comparing vol assumptions in options pricing / position sizing - Auditing whether sqrt-time scaling is over-estimating your scenario vol - The user says "rough vol", "Bayer Friz Gatheral", "vol scaling", "horizon vol" Not for: options pricing (this is not a calibrated rBergomi engine). Not for regime detection (use change-point-detector or market-regime). ## What you need - A ticker (`--ticker`) - `MASSIVE_API_KEY` exported - Stocks Basic minimum Optional: - `--horizons` (default `1,5,20,60,120`) - `--lookback-days` (default 504) - `--hurst` (default 0.14, Livieri et al. 2018 estimate on daily equity data) - `--ewma-lambda` (default 0.94, RiskMetrics) ## What you get b