The show featured a line up of speakers that were going to feature at the conference later in the week:
Dr. Marco Fasoli- Managing Partner, Titian Global;
Robert Passarella- MD, Dow Jones;
Con Keating*- European Federation of Financial Analysts;
Simon Kerr- Enhance Consulting/Hedge Fund Journal;
Bartt Kellermann- Organiser, Battle of the Quants
There was some great too-ing and fro-ing between guests as topics were debated - both sides seemingly with conviction. It was the best edition of the show I have been on myself.
Bartt Kellerman, Rob Passarella, and guest announcer
and "Dazed & Confused" Editor Rod Stanley squeeze into the studio.
and Con Keating prepare to respond to a tough question.
If you would like to listen to this show you can hear it via this download link.
My second round of Battle of the Quants was the full conference held on Thursday the 16th June, at which the Keynote Speaker was Dr.Paul Wilmott, Ph.D, publisher of the eponymous quant magazine and website. His professional background includes trading volatility at a hedge fund, so he has seen and applied quantitative methods as used in finance in practice as well as in theory. He addressed his audience of quants and investors in quant funds on some of the problems of doing this. He cited calibration and market completeness as particular problems.
An example of calibration is the the estimation of volatility for the pricing of derivatives. Models have to be fit for purpose and reflect the world as is. According to Paul Wilmott re-calibration of a model by changing parameters is a form of model risk. He was aghast to inform his audience that regulators of financial activity like to see re-calibration of banks' models used to price products and estimate risk, sometimes to the point of enforcing re-calibration.
Dr.Wilmott sees the concept of market completeness as somewhat dangerous. Markets are incomplete in reality - even fast moving markets with very large volumes have gaps in price, as fx markets show over and over, and the flash crash in the S&P showed last year. The quant maven asked rhetorically "Why then is the idea of market completeness popular?" It is because it enables market participants to use ideas of risk neutrality and a certain type of mathematics. The danger is in the fact that risk neutrality in a state of market completeness is a special circumstance, and not the governing mode of markets.
The Keynote Speaker had a theory about quantitative techniques as used in finance - that most people who use them don't much go beyond the tools of second year undergraduate mathematics. In particular, Paul Wilmott claimed that knowledge of fluid dynamics and the maths of mechanics would replicate how most people in the market address some very complex real world topics in finance. He gave an example of coming across a group in a financial institution attempting to model hurricane activity who assumed a log-normal distribution.
My own take on this is that the application of quantitative techniques to markets is no different than, say, a fundamentally driven approach: there are a range of abilities and resources being applied within the broad category. So for those that analyse industries and companies there will be some who rely only on street research; there will be those who carry out there own research, and those that employ expert networks. The depth of knowledge and understanding of a company or industry will vary a lot, and (relative) bet size should vary with the size of the edge (and risk/reward).
A good example of how different market participants cope with the shortcomings of modelling is in the pricing of traded options. All market participants (apart from the retail punters) are aware that the distribution of market returns is not log-normal. If you like, if you have a Bloomberg screen you will know that there is a smile in option pricing in OTM puts. If you are running money at a long only institution and use options you will be aware that those OTM puts are not necessarily expensive, but reflect a higher probability of a large fall in the underlying asset than is reflected in a log-normal distribution of returns. For the long only investor, who would hedge or take risk using options over a period of weeks-to-months, the assumption of the distribution of returns is flawed but sufficient for the purpose. For a market maker, who is modelling a three dimensional volatility surface and managing risk through constructing a risk book with a positive gamma, and has a time frame of intra-day and over-night risk, the assumption of log-normal returns is not adequate. No market maker uses pricing software that assumes that naive distribution of returns. So the different utility functions of the types of market participants will feed into the willingness to operate with a pricing model that is known to have shortcomings.
*Con Keating told a good story off-air about getting through US Immigration quite a few years ago. Although UK passport holders benefited from a visa waiver scheme when travelling to the US, Brits still had to fill in the customs form. Keating completed his and waited-in-line to be seen by the customs/immigration officer. Many will know that the US immigration procedures for an alien can be tough - they don't hesitate in sending visitors back on the next flight.
Finally Con Keating got over the yellow line and handed over his papers; the officer raised an eyebrow on reviewing them: "Mr. Keating you know we take immigration matters very seriously here. You don't appear to be doing the same: it says here under 'currency and valuable material' being brought into the United States that you are bringing in $223million. Is that correct?"
Keating confirmed that it was, and the officer went to have a consultation with his superior, eventually deciding that it was okay for the would-be visitor to enter the United States. Keating was much relieved as although he was there for a series of important business meetings he also had the job of delivering $223m of bearer bonds on behalf of his employers!
