Raw SVI in plain language
Gatheral's raw SVI parameterization models total variance as a function of log-moneyness. Its parameters control the level, slope, location, and curvature of the smile. Once calibrated, the function can evaluate a smooth curve between observed strikes.
Log-moneyness is commonly written as a relationship between strike and a forward or spot reference. Using a consistent convention matters: changing the reference or mixing spot and forward inputs can distort skew and comparisons between expiries.
What SVI helps with
Readable skew
A smooth curve makes the change in IV across strikes easier to compare than disconnected quotes.
Compact parameters
A small parameter set can be stored with a timestamp and calibration error for historical analysis.
Trader context
Skew, curvature, and term structure can be compared across expiries and regimes.
What SVI does not guarantee
A smooth fit is not automatically an arbitrage-free surface. Calibration quality, quote selection, bid/ask width, expiry normalization, and calendar consistency still need checks. Extrapolated wings should be labeled, not silently presented as market observations.
IVExplorer rule: display the fit as a model layer, preserve the raw market context, and expose fit quality rather than hiding it behind a polished mesh.
From descriptive to predictive
Historical SVI parameters can become features for regime classification or surface-change forecasts. Those forecasts should be compared with persistence and rolling-average baselines before any advanced model is trusted.