Friends,
Last week in Breakpoints, the discussion was about measuring implied skew.
A common measure and the one we use in moontower.ai is normalized skew which computes the percent premium or discount of IV at the 25d strike vs the 50d strike.
It’s not a measure that lends itself to direct interpretation. If 50d IV is 30% and the 25d put is 36% that’s a normalized skew of 20%. It doesn’t mean anything on its own but it is useful to see if skew is relatively or historically high or low. You chart it as a time series or percentile the value on a 1 or 2 year lookback. You can compare skew cross-sectionally across correlated assets.
Skew, or any measure, can be attacked from any number of angles. Our single measure of normalized skew itself requires choosing tradeoffs. The last post addressed the biases of various breakpoints. Moneyness, standard deviation, and delta-relative are all common ways to fix the gridpoints.
Today, we’ll use an approach that many might find more intuitive — thinking about skew in terms of option premiums instead of implied vol. When we look at option chains we are looking at prices. When we trade options our p/l depends on how the premiums change. For many investors, premiums are a more natural way to think about options than IV.
We will use GME to demonstrate a number of ways to think about skew which are more tightly intertwined with how skew trades are expressed — through verticals and ratio’d verticals.
We can even turn the metrics into a simple oscillator based on arbitrage bounds. If I do my job right, this post will make the concept of skew more concrete and inspire you to track it in new ways.