Key takeaways
  • Options-derived spot volatility adds forward-looking signal
  • Rough Heston beats Heston, Bates, SVCJ, and VIX
  • HAR gets better daily and month-ahead forecasts
  • Deep learning speeds large-options estimation

If you need to know whether tomorrow’s market swings will be wilder than today’s, options prices may give you an edge. This paper shows that information hidden in traded options can sharpen forecasts of realized volatility, the day-to-day volatility measured from high-frequency prices. The authors feed option-derived spot volatility, a model-based estimate of the market’s current turbulence, into the Heterogeneous Autoregressive (HAR) model. They infer that spot volatility under a rough stochastic volatility model and use a deep learning surrogate to speed estimation across large options panels. Against the Heston, Bates, and SVCJ stochastic volatility models, plus the VIX index, the augmented HAR-RV-RHeston model improves daily realized-volatility forecasting accuracy. The gain does not stop at one day: the model stays ahead across horizons reaching one month. The takeaway is straightforward: traded options contain forward-looking information that can complement the history embedded in realized volatility.

The market hides two clocks. One clock keeps yesterday's swings. The other lives inside option prices. This study asks which clock helps more. The answer matters if you care about tomorrow's risk. HAR, short for Heterogeneous Autoregressive model, tracks past volatility. It mixes daily, weekly, and monthly history. The surprise is that options add more than history alone. That edge beats the VIX index, the market's fear gauge. It also lasts beyond one trading day. The gain reaches horizons up to one month. That is a big deal when a calm morning can end in a wild close.

When options beat the market's memory

The best forecast comes from a rough Heston input. Rough Heston is a volatility model with jagged, fast moves. The model feeds its spot volatility read into HAR. Spot volatility means the market's current turbulence. That mixed model beats Heston, Bates, SVCJ, and VIX. It improves daily realized volatility forecasting accuracy. It also keeps the lead at longer horizons. The edge lasts through forecasts as far as one month. So the new signal does more than help on the next session. It stays useful when risk managers need a wider view.

1 monthforecast horizon

superior performance

daily realized-volatility forecasts
  • Heston gives a plain stochastic-volatility benchmark.
  • Bates adds jumps to that benchmark.
  • SVCJ adds both jumps and changing volatility.
  • VIX supplies a market-made volatility gauge.

“the augmented HAR-RV-RHeston model improves daily realized volatility forecasting accuracy”

From the abstract

How the options signal is built

The pipeline starts with traded options. Each option price carries a bet on future moves. A big options table gives the model many such bets. The rough-volatility fit turns that table into spot volatility. An iterative two-step loop updates the fit and the forecast. Then HAR takes the new signal and blends it with past volatility. A deep learning surrogate speeds the fit. That surrogate is a fast AI stand-in for a slower calculation. The result keeps the forecast grounded in market prices, not guesswork.

VIX gives a useful summary. It still compresses many option prices into one gauge. This route asks for a richer read. The rough-volatility route looks across the options surface. That surface means prices across strike levels and expiry dates. The model-based spot estimate can capture more shape than one index. That seems to be where the extra forecast value comes from. The lesson is simple. A headline gauge can miss detail that a model can keep.

Why the result matters

For a volatility desk, the win is practical. It can keep the simple HAR baseline. Then it can add an options-driven signal on top. That signal improves the daily call. It also helps over the next few weeks. The point is not that VIX becomes useless. The point is that model-based option reads can do more. They turn traded prices into a better forecast input. That helps risk planning when swings matter.

What to watch next

The surprise points to a clear consequence. A desk can feed option prices into HAR instead of relying on history alone. The result holds for daily forecasts and for horizons up to one month. The win comes from a model-built spot volatility read. That makes the options screen more than a hedge board. It becomes a forecast tool. This is the sharpest practical shift. Forecasting still starts with history. Now it can start with history plus the market's forward-looking bets.