Volatility forecasting across horizons: how to read model comparisons
AI-assisted research by PyInvesting. Sources and limitations are provided below. Educational content; not personalized investment advice.
AI-assisted article drafted and checked with AI against the sources below. Educational information, not personal investment advice.
How model rankings vary by forecast horizon
Federal Reserve research finds that volatility-model rankings differ by horizon, with Markov-switching HAR performing best at short horizons and ARFIMA generally leading at the monthly horizon. [S6]
How policy uncertainty is presented separately
The European Central Bank describes its economic outlook as highly uncertain, with risks tilted upward for inflation and downward for economic growth. [S8]
What to keep distinct when reading the evidence
The ECB says it is not committing to a predetermined interest-rate path, so a policy outlook should be treated as changing context rather than as a direct ranking of volatility models. [S8]
What this means
Questions for the reader: What forecast horizon matches the decision being evaluated? Does the cited model comparison use that same horizon? Is a central-bank outlook being used only as contextual information, rather than as evidence that one volatility model will perform better? What incoming data would justify revisiting the assessment?
Limitations
This article is limited to the supplied excerpts. The Federal Reserve material discusses realized-volatility forecasting, while the ECB material describes monetary-policy uncertainty and decision-making. The excerpts do not establish that a model predicts future portfolio returns or that policy decisions determine asset prices. No investor-specific portfolio, time horizon, or risk tolerance is assessed.
horizon-specific volatility forecasting
Sources
- Linear and Nonlinear Econometric Models versus Machine-Learning Models: Evidence from Realized-Volatility Forecasting — 2026-09-02T15:30:00+00:00
- Monetary policy decisions — 2026-09-10T12:15:00+00:00