Serving the Quantitative Finance Community

 
frolloos
Topic Author
Posts: 752
Joined: September 27th, 2007, 5:29 pm
Location: Netherlands

Why is SABR so popular in the rates world?

March 30th, 2017, 12:04 pm

Is it largely due to the Hagan et al formula for implied volatility that made SABR popular? If it is only for inter and extrapolation reasons why not SVI for example?
 
User avatar
Alan
Posts: 2958
Joined: December 19th, 2001, 4:01 am
Location: California
Contact:

Re: Why is SABR so popular in the rates world?

March 30th, 2017, 2:59 pm

Just guessing, but the SABR work (i) pre-dated SVI, and (ii) spoke directly to FI quants with examples from there. 

Also, based on many threads here, I am pretty certain the practical interest from traders in the model is almost purely, as you say, as an interpolation formula with some theoretical justification.  Since there is a model underlying the interpolation, this adds another point in its favor vs. SVI which is very nice but more ad hoc. 

Objectively, your point is well taken -- why bother with an interpolation formula that so easily breaks down (deep otm is not arbitrage-free) and is so much trouble (from the traders point of view) to correct. (Traders have little interest in the exact solns to the model, as far as I can tell, as they will tell you they don't take the model 'seriously'. What they really mean is "yes, the interpolation formulas have problems, but correcting them by solving the model exactly/numerically is way too much work!)  

Finally, which is the cooler acronym?  :D
Last edited by Alan on March 30th, 2017, 3:35 pm, edited 1 time in total.
 
frolloos
Topic Author
Posts: 752
Joined: September 27th, 2007, 5:29 pm
Location: Netherlands

Re: Why is SABR so popular in the rates world?

March 30th, 2017, 3:34 pm

Thanks Alan. Yes, as you pointed out there are some difficulties with the deep OTM options especially recently (last few years as far as I can tell). Which is why I started wondering why still use it. But most rates traders are comfortable with it or its tweaks like shifted SABR. Furthermore I am  Not sure whether SVI would have same issues with rates vols - never tried SVI for those.

And just now I found out that SABR is also popular in tennis: SABR conquers tennis. Waiting for Nadal to answer with an SVI move :-D
 
User avatar
Alan
Posts: 2958
Joined: December 19th, 2001, 4:01 am
Location: California
Contact:

Re: Why is SABR so popular in the rates world?

March 30th, 2017, 4:09 pm

Haha!

A related issue is the cognitive dissonance that comes from eschewing specific dynamical models -- which is what most traders do: they will tell you they do not "believe" in specific models -- any specific model. Once you adopt that point of view religiously, everything becomes ad hoc. This is true in both equity and fixed income. The consequence was/is a notable lack of dynamical models dealing seriously with a zero (or small negative) rate bound. 

Anyway, this is from the point of view of someone looking at rates models only casually and from the outside of banking.   
Last edited by Alan on March 30th, 2017, 4:26 pm, edited 3 times in total.
 
User avatar
gatarek
Posts: 213
Joined: July 14th, 2002, 3:00 am

Re: Why is SABR so popular in the rates world?

March 30th, 2017, 4:16 pm

.
Last edited by gatarek on March 12th, 2019, 7:40 am, edited 1 time in total.
 
User avatar
Alan
Posts: 2958
Joined: December 19th, 2001, 4:01 am
Location: California
Contact:

Re: Why is SABR so popular in the rates world?

March 30th, 2017, 5:30 pm

I'll bite. Two well-known problems with the small-T asymptotic SABR formulas are:

(i) negative implied vols at low strikes, and

(ii) an ambiguous prescription of what the underlying F "does" when it hits F=0 (or some other shifted bound).

Both problems are simple failures of an asymptotic approximation when pushed to a regime it can't handle. But, given those issues, how is it that it always 'works'?  
 
User avatar
gatarek
Posts: 213
Joined: July 14th, 2002, 3:00 am

Re: Why is SABR so popular in the rates world?

March 30th, 2017, 5:54 pm

.
Last edited by gatarek on March 11th, 2019, 5:59 pm, edited 1 time in total.
 
frolloos
Topic Author
Posts: 752
Joined: September 27th, 2007, 5:29 pm
Location: Netherlands

Re: Why is SABR so popular in the rates world?

March 30th, 2017, 8:42 pm

You're probably right Gatarek, But haven't Andreasen and Huge written some quite recent papers on what they claim to be arbitrage free and extremely fast finite difference methods for local stochastic volatility models for short times to maturity? If true, then I suppose that could be an alternative to SABR?
 
User avatar
gatarek
Posts: 213
Joined: July 14th, 2002, 3:00 am

Re: Why is SABR so popular in the rates world?

March 30th, 2017, 8:58 pm

.
Last edited by gatarek on March 11th, 2019, 5:59 pm, edited 1 time in total.
 
User avatar
mtsm
Posts: 78
Joined: July 28th, 2010, 1:40 pm

Re: Why is SABR so popular in the rates world?

March 31st, 2017, 1:38 am

I am not familiar with SVI, but somewhat with SABR from an operational point of view as well as quant point of view. 

All the shortcomings of SABR are as you say. You can somewhat work around the low strike problem by shifting as far as I remember, but pricing of low or negative strikes will still be bad. It takes a different sort of approach for these strikes. So on the sell-side, pricing bid-asking needs to be handled ad hoc. Not sure about the system. At high strikes it's problematic also, so for CMS people typically use a cutoff for the replication to mitigate overly fat tails. Either way. 

I was also wondering why most people aren't willing to try purely parametric skew or even preferably a Bayesian approach maybe, but it seems that while no one really finds SABR dynamics trustworthy, all the same people are very reluctant entirely giving up a dynamic model for the underlying and vol. My understanding that is that elsewhere such as equities that's much more tolerated. In rates there is this belief that you'd still loose much if you gave that up. While intrinsic SABR greeks aren't very useful for hedging purposes, people still very much like the down-to-earth interpretations of beta, volvol, rho and sigma0 (depending how you run the model). It's convenient to look at the dynamics through tangible parameters with a nice interpretation even if the dynamics implied by constant parameters is wrong. Parameters get remarked anyway. Also it's nice to be able to express the skew exposure in terms of parameters. In principle it's also possible to express views through parameters, such as views on skew or smile, etc... 
 
User avatar
Martinghoul
Posts: 188
Joined: July 18th, 2006, 5:49 am

Re: Why is SABR so popular in the rates world?

March 31st, 2017, 10:05 am

I agree with mtsm...  I think it's a combination of two things: a) relative ease of implementation; and b) "intuitiveness".  Both of those aspects appear to make SABR, with all of its flaws, sorta "optimal", IMHO.
 
User avatar
gatarek
Posts: 213
Joined: July 14th, 2002, 3:00 am

Re: Why is SABR so popular in the rates world?

March 31st, 2017, 12:23 pm

.
Last edited by gatarek on March 11th, 2019, 5:59 pm, edited 1 time in total.
 
User avatar
mtsm
Posts: 78
Joined: July 28th, 2010, 1:40 pm

Re: Why is SABR so popular in the rates world?

March 31st, 2017, 1:37 pm

what is this supposed to mean? of course there are alternatives. the market isn't a function of the model, it's the other way round. 
 
User avatar
gatarek
Posts: 213
Joined: July 14th, 2002, 3:00 am

Re: Why is SABR so popular in the rates world?

March 31st, 2017, 3:07 pm

.
Last edited by gatarek on March 11th, 2019, 6:00 pm, edited 1 time in total.
 
User avatar
riskneutralprob
Posts: 23
Joined: December 16th, 2005, 5:16 pm

Re: Why is SABR so popular in the rates world?

April 4th, 2017, 5:18 pm

"While intrinsic SABR greeks aren't very useful for hedging purposes" 
What are intrinsic SABR greeks?   Are the Bartlett corrections for Delta not useful?