13 Jan, 8 tweets, 2 min read
Let's talk FX correlation...

While FX correlation pricing and trading is probably a topic for a longer post (probably somewhere down the road will do a write-up...), I think grasping the basics of fx correlation is crucial in in understanding the dynamic and effect on fx vol
so let's dive in...

the fx vol business is structured in a way that each desk on the fx vol side in in charge on a "bloc", so you have EUR,JPY,GBP, and commonwealth (AUD,NZD,CAD) blocs. Each desk is in charge of making price and running the risk of the currencies that fall under
that bloc. so far it all makes sense, right? but in FX almost any trade outside of the liquid pairs (mosly USDxxx and intra-bloc pairs, like EURGBP,AUDNZD,etc...) are essentially correlation trades, as the dealer in most cases need to synthesize the risk (vega/gamma risk) using
the liquid pairs. This is where it all gets very interesting...

Let's say that I ask my dealer for a price on AUDJPY 1m atm vol, but the AUD trader has no AUDJPY exposure on the book, so she needs to make a price (she can use the AUDUSD exposure on the book, and ask the JPY
trader for price in USDJPY).. how does she make a price?

the easiest way to make a price for illiquid fx vol is using a correlation triangle.

to understand how correlation triangle works, we need to take a trip down memory lane to high-school trigonometry lesson
given a triangle where we know two vectors (AB,AC) and the cosine (x), where x= the angle between AB,AC. we can solve vector BC - the illiquid pair's vol (or we can solve the implied correlation given the three quoted pairs)
now, as correlation is a byproduct of volatility, the implied correlation is going to reflect the dealer's price of future realized correlation between AUDUSD/USDJPY, which is far from trivial...

there are many ways to take a view on fx correlation (either via vanilla
strategies or via correlation swaps).. as these correlations tend to present many misprices and anomalies ( in G10 alone we have 45 ccy pairs, with about 30 regularly traded in the market, so this entire matrix is extremely complex to price perfectly at any point in time)

• • •

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

This Thread may be Removed Anytime!

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

# More from @VolQuant

12 Jan
Back-end rates vols finally start to show some signs of life....

After a very long period of flat term-structure/flat skew rates vols finally start to show some risk priced into the next few months, with steepening (as much as one can call a 0.2vol spread a steepness...)
and further steepness of the put skew (compared to pre-GA runoff)...
Given the the rates narrative has changed significantly over the last few sessions, we could see some fast-money accounts trying to chase the TY lower (either via options or via hard delta)

I like playing the put skew via spreads, as I don't see a case for an acceleration
9 Jan
So it's the weekend and seems like the perfect time to touch on one of the most deliberated and discussed subject in option trading - the weekend effect

Any option trader knows that the weekend has some kind of strange effect on implied volatility and there is not one truth
when it comes to how we should treat the weekend in vol pricing... this is mainly b/c of the fact that option pricing uses actual days (i.e., 365.25 if we want to accurate), while in fact there are roughly 250 trading days/year (ex. weekends and holidays)
this mismatch between act days/trading days creates two effects :
1. it requires us to discount non-trading days when pricing volatility ,
2. it creates a discontinuity in price dynamic, which we should account for (weekend gap on Monday)
7 Jan
If you want to know how FX is being driven by the move in long dated rates, you just need to look at the substantial repricing of the back end of the entire FX market (mainly EM, and few DM to some extent).

In the "post-corona" long dated rates have been sold across the world
to an historically low levels, as everybody and their mother was piling on to long duration. But.... when everybody piling to a carry trade, it almost always end in a squeeze...

And so we got a nice squeeze in EM 5y5y fwd rates , which seems to be weighting on EMFX today
in G10 we also see a move in the back end of the curve, but to a lesser extent... one thing I do look at is the AUD,NZD 5y5y (they High-yielders of G10), which have been moving higher since basing around mid-oct
6 Jan
Few thoughts about US long dated rates this morning....

Obviously the long dated rates moves was on the card for quite a while (lagging behind the forward-looking inflation measures, like 5y5y forward inflation)... it didn't take too much for rates to start moving post GA
results. Obviously we still need a confirmation with regards to who won, and who will control the senate, BUT, the move in rates looks imminent regardless...

Few things that caught my eyes:
1. TY put skew has steepen (which makes sense given how high rates can go from here
if the 10y clears the 1% hurdle momentum buyers will eye 1.2% or 134-25 in TY).

Honestly I don't think the move, if happens, is going to be violent, but given that the absolute level of vol is trading well below average it makes the play nice in risk-reward term.
2 Jan
In financial market we tend to think that carry beats everything... being long stocks, short vol, long EM.

Almost every macro/volatility trade is essentially a carry trade. When markets are calm and nothing happens we get compensated for holding risky assets.
I recently had an interesting discussion with my friend @amirwe about carry in EMFX.. He claimed that If i ran a strategy that PAYS 1yr USDTRY outright forward at the beginning of every year since 2010 I would have made a killing... That sounded quite counterintuitive to me
As I was taught that carry is the king, and we make money by being long high-yielding currencies... so I had to check if what he said is factually...

Well - he was spot on! being LONG usdtry (with carry well against you) made almost 900% return (buying at the beginning
30 Dec 20
One of the biggest flaws in financial time-series analysis is looking at the data as a function of time.

Obviously it's the easiest way to analyze data, as we can plot it nicely, calculate the volatility, etc... but when we analyze data in a dimension of time we usually get
biased results. why? because we are extremely sensitive to our selection of sampling frequency/length of sample window.
Just think about the most useful tools in your quant toolbox : realized volatility and z-score...

Both are usually measured as function of time (and relative
to historical sample window), but who said that volatility evolves at t-fixed time intervals? the underlying fluctuation is not a function of time but the orderbook imbalance.
Also, let's say that I dynamically hedging my delta, am I really using fixed t-intervals, I guess not.