Little piece of vol quant history for you, which is essentially correct. Lots of lessons here - some were learned 20 years ago, others not.…

TL;DR - Goldman isn’t big in exotics, here’s why.
My memory of the dates is that this happened earlier and was done by 2003 but the story fits, I priced and risk managed these and we couldn’t get near the GS price. I think they lost 100M all told, back when that was a lot
There was a nice 6 month period when the street figured out what GS had done and nobody told them. (They were loved, even back then)
One lesson learned was that local vol has strong limitations on it for forward starts (cliquets). Lorenzo Bergomi’s monograph is the best source for why (and a great, great book overall). This shitshow explains why he spends 100 pages shooting down local vol
GS never found another way they trusted to price exotics, and they are still right.
The Europeans thought this was hilarious, didn’t learn the same lesson, and spent 2002-2008 building stochastic vol models. Hence Bergomis book.
Unfortunately the tradeoff with the new models was as follows: you’re gamma hedged so you have neutralised your exposure to realised moves. This is good as they are hard to predict.
In their place you have exposure to a bunch of Greek letters that you have some intuition on but not much. You hope they don’t move much. You try to find ways to offset that risk even though you know it’s likely model dependent
This all makes sense in stable markets, your volzomma parameter won’t move and you can keep taking margin. Unfortunately the global equity market has become more leveraged and prone to explode every 3 years. It’s not gone well.
This episode is also why JP doesn’t have big exposure to retail exotics. They never got comfortable with the game - if GS could get it wrong anyone could. They kept it simple.
One other lesson was that the customer will buy whatever you sell them. They can’t tell whether you can hedge or not. They’re morons and that’s actually dangerous for the seller - you’re on your own.
This is topical as after losing again in 2020 despite huge investment in ‘risk management’ a lot of European banks are finally coming round to the US view - some products shouldn’t exist.
Equivalently that trade off between realised vol exposure and volzomma may not be a good one in the long run.
Maybe they’ll actually go do some time series analysis and look at what they’re trading. That would seems to be the way forward.

• • •

Missing some Tweet in this thread? You can try to force a refresh

Keep Current with MultifractelFarol

MultifractelFarol Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!


Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @Mephisto731

1 Feb
OK so: summary. Is it me or is this nonsense over now? Feels like it to me.

It's not a pandemic or LTCM, or Europe 2011. It's more like XIV day, or a quant quake, or the flash crash, or flooded Japanese reactors, or all the other idiosyncratic fun we've had in times past.
RH has the extra cash it needed, this seems to keep them going. I do believe their brand is OK and they will survive, as I do believe they are good at finding another sucker. And yes Citadel still loves them.
Melvin is alive but badly injured losing near 2 years (30% p.a. before this). I thought they were dead but no. They'll raise more money. Why? Because they weren't formally academically, 'wrong' - they just got mugged by Potato_In_My_Ass.
Read 9 tweets
30 Dec 20
1. When we are all awash in information, as all market players now are, the defining feature of each of us is what we choose to ignore - or better, what we choose to forget and how fast.
2. ( Apologies for the Naval tweet )
3. "How can you just buy calls like a moron?" is like asking "Don't you know what can happen?", i.e. "How can you have forgotten already?"
Read 10 tweets
15 Dec 20
Rant mode on. I don’t care about your big data. Increasingly I care about small data. If you can talk sense having very little data then you have a chance when more data arrive. The converse is not true
We saw screens go wide or empty in March. How did your cutesy machine learning algo do then?
Here’s a vol quant question worth thinking about. I give you the vols at 40 delta put and 40 delta call. Nothing else. Build me a smile that’s arb free.
Read 16 tweets
3 Nov 20
Threads to take to your mind off things, something more interesting I hope. Part 2.
Following on from minority games in part 1, if you start caring about actually modelling how market players react to information then you end up coding what's called an agent based model (ABM).
In an ABM there are lots of actors receiving information, and reacting according to some preset reaction functions (noise traders playing some sort of trend following, contrarians playing mean reversion). These are based on a physical model called the Ising model.
Read 11 tweets
3 Nov 20
Threads to take to your mind off things, something more interesting I hope. Part 1.
Tail risk, done cleanly, is pure convexity and no beta. You’re looking for cheap optionality. 2 ways this can arise. First the option looks underpriced compared to payoffs in a few realistic scenarios. Second there are extreme scenarios that look more realistic than usual
Example of the first is just usual relative value trading: vol and skew is sometimes just cheaper than usual
Read 10 tweets
14 Oct 20
I've been looking at variance swaps (again), SPX skew termstructures (again), the CBOE $SKEW index and a couple of beautiful papers by Neuberger that made me smile - there's a nice trick.

Here's a puzzle: var swap replication uses option prices at maturity (and 30 days for $VIX), nothing in between now and then.
But a variance swap is the sum of squares of daily returns. How do you get from implied volatility in the option prices to daily realized variance?

There's a flip in the dimension from a sum over strikes to a sum over days. Its objectively weird but rarely discussed as such.

Read 12 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!