Further news regarding Alternative Reference Rates

A couple of days ago, the US Alternative Reference Rates Committee has published an announcement regarding its "preferred alternative reference rate". The announcement can be found on the New York fed web site at https://www.newyorkfed.org/arrc/announcements.html.

The proposed rate is based on repo transactions. It is certainly a useful new benchmark and a useful reference for "new USD derivatives". A discussed previously in this blog in post on similar subjects, and as implied by the name of the committee, this is an alternative reference rate, not a replacement of the LIBOR rate. The repo rates have a different credit implication than the inter-bank transactions underlying LIBOR. The risk management and valuation features of the two benchmarks will be different. The new benchmark will lead to "certain new U.S. dollar derivatives". Implicitly, this means also new markets, new clearing infrastructure, and new master agreements and CSAs.

The alternative benchmark is only proposed at this stage; it is not yet finalized and not yet published. Consequently there is no historical data available for analysis and no scrutiny on how it behaves on a daily basis. The author of the announcement acknowledge that its robustness is only a "likely robustness" and not a proven one.

The committee will propose "transition plans"; this is an important part of creating an alternative benchmark. Some LIBOR rate derivatives have maturity of 30Y or longer. The transition should be over that period. Any attempt to shorten that period will create risk management and valuation challenges for existing users. Solving them will require bilateral agreement specific to each existing contract; a committee imposed approach would certainly meet legal challenges.

The description of the new indices to be created can be found on the NY fed page

The exact definition of the benchmarks has not been finalized yet and the indices themselves have never been published. The proposed benchmarks are only Overnight rate; as such they are not a replacement for term LIBOR. Would the derivative market move to an overnight reference rate only market? Somehow I doubt this would be very practical for end users. Would corporate want to have daily fixing for their loans? What about floating rate mortgages?

The committee plan to publish its final report later this year. To my opinion, it will certainly take a fair amount of time, to be measured in years, before this alternative market is stabilized. If I take as reference the posts on this blog, the first one on the change of benchmark was published in 2014: Change of benchmark overnight index is a difficult task.  Almost three years later, the discussion has progress but very little has been happening in the actual derivative market regarding the benchmark.



Running Wall Street

The Wall Street Journal (WSJ) ran a piece this week-end titled "The Quants Run Wall Street Now". Some pictures of me are in the WSJ library; they took them at a presentation I gave at The Thalesians last month. They used one of those pictures to illustrate "the quants".

The picture illustrating the article is
Figure 1: Me with the first step of AAD printed on my face and the recursive formula on my shirt.

Following the article, I feel the need to clarify a couple of "details". First of all, I'm not running anything or anybody at Wall Street. You may thing that this is a pity for human society, but personally, I feel very good about it and I don't want to run anything or anybody; nevertheless I don't mind speaking to people who want to listen to me willingly. This is the case for practitioners, regulators, academics, and journalists.

Now about the content of the article. It discusses the increasing importance of "quants" in finance. It if important to clarify what that "quant" can be many things in finance; this is not done in the article and leads to confusion, like illustrating the article with a picture of me. There are several types of quants: big data quants, high frequency quants, derivative quants, risk quants, balance sheet quants (and probably others, but I stop there).

The article is more about big-data quants and a little bit about high frequency quants. So the illustration of the article by a derivative - in all sense of derivative, given the algorithm depicted in the picture - quant - me -, may not be the most appropriate. But I have to confess that using a title containing "Algorithmic Differentiation" in a seminar on derivatives and balance sheet, may have confuse people looking for algorithmic trading.

I described myself above as a derivative quant. I have to add that this has been the case and is still partially the case, but I have slowly transformed into a balance-sheet quant. I'm planning to write more about that in a forthcoming blog - which I have started to write some weeks ago. Derivative pricing is about replication; replication is creating a small balance sheet composed of the option, some cash and some underlying and creating a strategy for it to maturity; the balance-sheet of a bank is simply a grow-up contingent claim. In that sense, evolution from derivative quant to balance-sheet quant is a natural evolution when growing-up.

To my subjective point of view today, I would say "The balance-sheet quant will run Wall Street Tomorrow!" Maybe that will be an article in The Wall Street Journal in a couple of years, and then the picture will be correct.

The slide that you see on the picture will be Table 2.3 in my forthcoming book Algorithmic Differentiation in Finance Explained. The very last corrections of the proofs are with the editor as I write. I'm still hoping for an availability in libraries before the summer holidays. Thanks to the WSJ for running such an advertisement campaign for my book.


Book's proofs

I have just received the first proofs of my forthcoming book "Algorithmic Differentiation in Finance Explained". After a couple of months of editorial lethargy, the matter is moving again.

Hopefully the book will be available before the summer holidays.

The book is already announced on Amazon!


Profit explain explained

I was once asked to explain the P/L explain report. Something must be wrong in my story telling if I have to explain the explanation. After the failure of my story telling career, I decided to move to story writing. This note is an attempt at starting my new career.

The full text is available as PDF:

Profit explain explained


The Thalesians seminar

I will be speaking at The Thalesians seminar on

SIMM and SA FRTB: double AD

Seminar starts at 6:30pm on Wednesday 19 April 2017 at

City University Club

Champagne will be served before the seminar, around 6:00pm.

Register at Meetup:


Algorithmic Differentiation (AD) has been used in engineering and computer science for a long time. The term Algorithmic Differentiation can be explained as ``the art of calculating the differentiation of functions with a computer.
Over the last 5 years, AD has made its road to quantitative finance. The most straight forward use of AD is to compute the sensitivity of PV to market inputs. In the frame of SIMM and SA FRTB computation, those sensitivities are the main input and having an efficient way to produce them is important.
Once the IM/Capital number is computed, there are a lot of potential analysis which are handy, like marginal IM and IM attribution. Those analysis also require some form of differentiation, this time with respect to the positions.


SIMM and sensitivity based FRTB: double AD
  • Algorithmic Differentiation and computation of sensitivities
  • First AD: fast inputs for SIMM/FRTB
  • Second AD: sensitivity of the IM/Capital itself w.r.t. sensitivities
  • Second AD applications: attribution and marginal IM/Capital