Why Option Market Makers Fit Smooth Implied Volatility Surfaces
Summary
The document explains why an options market maker may fit a smooth implied volatility surface instead of quoting directly from individual observed strikes. A fitted surface can reduce noise in market observations and provide a consistent estimate of volatility across strikes and maturities. Because volatility is more naturally described by moneyness than by a fixed strike, the surface can also help guide quotes when the underlying price moves and contracts shift to different moneyness levels.
The discussion highlights additional difficulties in the wings: bid–ask spreads are wider, increasing observation noise, while vega is lower, making prices less responsive to volatility changes. These conditions make raw implied volatility readings less reliable. The answer distinguishes automated fitting from autoencoders, noting that a neural representation could be computationally burdensome for latency-sensitive market making. It offers qualitative rationale rather than a specific fitting algorithm, calibration procedure, or performance comparison.
Key ideas
- A smooth implied volatility fit can reduce noise in observed option quotes.
- Moneyness provides a more useful reference for volatility than a fixed strike when the underlying moves.
- Wing quotes are harder to interpret because spreads are wider and vega is lower.
- Latency constraints can matter when selecting a volatility surface fitting method.
Tags
Full text
# Option Market Making fitting the vol surface # Option Market Making fitting the vol surface As an option Market Maker why do you need an auto-fitter for the vol surface, and why using the mid in the wings is not a good pratice? My guess would be that you need to fit the vol surface to be able to quote? So for example you observe that the implied vol of call with strike $K$ is $\sigma$ and the implied vol with strike $K+1$ is $\sigma_1$. Why would need to interpolate the vol surface between $\sigma$ and $\sigma_1$ for example? If you want to quote for the call with strike $K$ you don't need to know what is the implied vol for the strike $K+ \epsilon$ right? You are probably just going to take the implied vol $\sigma$ and skew it by a certain amount and that's it. That's why I don't really get why you need to fit the vol surface when being an option MM and also apparently, it's harder in the wings but I don't know why. If someone has any ideas that would really help my understanding. ## Answer by krkeane (score 1) https://quant.stackexchange.com/a/80747 > why do you need an auto-fitter for the vol surface[?] I don't understand your terminology. Do you mean an automated fitting process when you use the term "auto-fitter", or are you suggesting use of an auto-encoder for representing the volatility surface. An auto-encoder approach would be cool, but likely a bit "heavy" for a latency sensitive market making. Fitting a smooth surface offers a noise reduced estimate of a volatility. > Why would [you] need to interpolate the vol surface[?] - Volatility is not associated with strike as much as it is a associated with moneyness. Moneyness changes when the underlying changes. A smooth surface offers guidance after underlying price movement. - Its a noise reduction method. > it's harder in the wings Some issues you must endure in the "wings" include wider bid-ask spreads (observation noise) and lower sensitivity to volatility (lower vega), which makes a market descriptor fixated on volatility problematic.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.