PARTIAL · vol model
Hypothesis
The Heterogeneous Autoregressive model of realized volatility (daily/weekly/monthly RV components) forecasts volatility well enough to drive sizing and stop placement.
Math — HAR-RV
$$ RV_{t+1} = \beta_0 + \beta_d RV_t^{(d)} + \beta_w RV_t^{(w)} + \beta_m RV_t^{(m)} + \varepsilon_{t+1} $$
Method
Fit HAR-RV on intraday realized variance per symbol, forecast next-day RV, drive vol-targeted sizing and ATR-stop scaling; benchmark vs GARCH (N-023).
Results
| Vol forecast $R^2$ | 0.45 |
| Beats GARCH(1,1) | modestly |
| Directional content | none |
Adopted on the risk side: HAR-RV forecasts realized vol slightly better than GARCH and is cheap to fit, so it drives sizing and stop distances. No directional edge — like every honest result here, the forecastable quantity is variance, not return.
The most reliable thing you can predict in markets is how much they will move, not which way. Build the risk engine on that, and stop asking price to be predictable.