Dealer Gamma (Institutional Options Market Making)

[joederp]

Agree with 'You', very interesting info that Real Money and Same Lazy Element have laid out, and the depth of discussion is greatly appreciated. I first took notice of this topic through Charlie M. with Nomura, and have been browsing through the links that R.M. provided.

@Real Money - aside from the basis spreads, are you doing a cumulative analysis of option trades by logging each update in the option chain from Thinkorswim, or another source?

From SqueezeMetrics's white paper on the GEX, it seems they just take the most recent value of gamma per-strike, multiplied by its OI - see attached.

Whereas I think with charts I've seen from spotgamma and Nomura, they've done some cumulative distribution by-strike (or spot).

 

[Kevin Schmit]

My research indicates that "dealer gamma" is very real, and it is a factor for the models/tools I use to trade.
https://spotgamma.com/why-gamma/
https://squeezemetrics.com/download/white_paper.pdf

Of the two sites you reference, the first, SpotGamma, seems like a complete scam site. All their charts are lifted from somewhere else; they either don't have any historical "Dealer Gamma" data of their own, or they don't know how to create a scatterplot in Excel. I would avoid that site.

The second site, SqueezeMetrics, is a little better, but their White Paper is misleading. Mostly because they include the GFC in their analysis. Nearly all the negative Dealer Gamma days were in the GFC period, so the actual link between Dealer Gamma and next day vol is much weaker than they claim, and the relationship between Dealer Gamma and signed next day returns is non-existent. Only six percent of SPX days in the last 10 years have hadeven slightly negative Dealer Gamma. Strongly negative readings are just a proxy for "we're in a bear market so vol is higher than usual."

I have attached a scatterplot of SPX dealer gamma (calculated with the SqueezeMetrics method) over the last 10 years. Notice is looks a lot like the scatterplot in their White Paper but with the entire left quadrant (GFC period) missing. There is no visible relationship with forward signed returns, but a reasonable relationship (r^2 ~ .15) with forward vol (next day's squared returns). Also the information in SqueezeMetrics version of DealerGamma vis-a-vis forward squared returns appears to be at least partially independent of the information in implied SPX vol or recent SPX realized vol.

 

[Real Money]

@Real Money - aside from the basis spreads, are you doing a cumulative analysis of option trades by logging each update in the option chain from Thinkorswim, or another source?

I don't really use anything except basis spreads, rate spreads, and index spreads. This is all an effort to see what the size players are doing. It is very obvious to me that the BD guys are running the show in the bond spreads, and therefore in the basis.

This makes sense because they have the largest client books and they make markets for the biggest players.

However, they also happen to be the biggest dealers in vol in addition to the off exchange, global, and OTC market making.

It all makes sense if you just assume that the BD guys are running the show. Dealer gamma, and the motivation by OMMs to manage their delta, is consistent with everything that I see on a daily basis in these markets. How or if you use it is up to you.