Estimating Dealer Directional Open Interest from Options Trades

# Dealer's directional open interest


# Dealer's directional open interest












This question is related to the previous question I asked here. In one of the answers, the article from Squeezemetrics that discusses the effect of GEX on the spot price of an asset was pointed out. GEX is an interesting concept. If indeed we can successfully calculate GEX, then negative GEX means there is an accelerating effect of options OI on the spot price in the direction of its movement. If GEX is positive, then it tends to pull the spot price towards that strike price as if it has a compressing effect. Here's the catch. The article requires something called 'dealer's directional open interest' (DDOI) whose computation is given in the last page of the document here. I find that explanation very vague. Hence I thought of asking folks on this forum about how exactly can we find DDOI if at all possible. The questions are in two parts.

- Why does DDOI even differ from OI? If someone is long, someone else is short right? So why bother finding the OI of those people who are hedging their positions? Why would someone who is long an option hedge his/her position? Who in their right mind would short an option without hedging it? And how long can they hold such a position without going bankrupt? Long story short, I would think that anyone short an option should be hedging his position, and anyone going long an option does not hedge. Why then is DDOI even necessary to be defined?

- Assuming that there are some crazy traders who sell options without hedging their position with the underlying, or that there are people who long an option and hedge it for reasons they know best, it makes sense to define DDOI. But then the article does not clearly explain the steps in the calculation of DDOI. Let me concretely define the case and I will keep it generic so that you can answer it in some form of an algorithm based on the values of the variables. Suppose some stock or index has an option with OI of $oi_0$ and DDOI of $ddoi_0$ at the beginning of the candle (or before the tick update ... whatever the case maybe). Let it be $oi_1$ at the end of the candle. Let there be $b$ buy trades and $s$ sell trades in that candle for that option. How then do I find the DDOI, $ddoi_1$, at the end of that candle for that option?

## Answer by Bruno Reis (score 3, accepted)

https://quant.stackexchange.com/a/81137

> Why does DDOI even differ from OI?

Because DDOI focuses on one specific subset of market participants — "dealers", i.e. those who make markets on options — and tries to identify whether they're long or short. On the other hand, OI is just open interest: it doesn't differentiate among market participants.

> If someone is long, someone else is short right?

Correct: that's why OI has no sign.

> So why bother finding the OI of those people who are hedging their positions?

Because there are very significant and reliable dynamics that are a consequence of positioning.

> Why would someone who is long an option hedge his/her position?

Because they don't want the directional exposure that comes with that position. They may not have chosen to have that position, but rather were forced into that position: that's what happens with market makers.

> Who in their right mind would short an option without hedging it?

Someone who wants the directional exposure. E.g. it's a common strategy to sell a put to initiate a long position on some underlying.

> And how long can they hold such a position without going bankrupt?

If they know what they are doing, indefinitely. If you have at least 100*K usd, and you sell a put at strike K on some stock, you'll be able to pay for the stock if you're assigned, no matter what happens.

> Long story short, I would think that anyone short an option should be hedging his position, and anyone going long an option does not hedge.

This is likely because of lack of practical experience. There are lots of situations where shorts won't hedge, and longs will hedge, some described above.

> Why then is DDOI even necessary to be defined?

Should be obvious by now: if you can identify the (small) subset of market participants who generally hold very large option positions and don't have any intention of maintaining any directional exposure — i.e., "dealers" —, you can reliably count of their delta hedging activity.

> But then the article does not clearly explain the steps in the calculation of DDOI.

There's no way to calculate DDOI. Dealers won't share their positioning. You can only guess what it is, by making some assumptions. These are very common-sense: on indices, assume dealers are short puts and long calls; on stocks, assume dealers are short on both. That is: assume "users" of the options markets will be making "bets" on stocks, therefore buying options; and they'll be "always long" on the market overall so they'll buy puts on the indices to protect their portfolios while at the same time they'll sell calls to finance their puts.

This is a very gross approximation, but extremely useful and reliably accurate.

> Suppose some stock or index has an option with OI of [...]

Assume that whatever excess you have between `b` and `s` will land with a dealer. That will be the change on DDOI for that particular contract.