point

 

 Remember me

Register  |   Lost password?


Sign up here to let us know if you are interested in joining us for our Introduction to QuantLib Course later in the year.

 

MoneyScience Blog Header 2015 2


Paul Darbyshire and David Hampton on Hedge Fund Modelling and Analysis

Tue, 14 Apr 2015 07:35:00 GMT

Paul Darbyshire and David Hampton discuss their trilogy of books written for Wiley Finance.

You can read a sample chapter of their latest book Hedge Fund Modelling & Analysis using MATLAB here. To claim 30% off your copy order direct from Wiley here and enter our promo code MON30 at the checkout. Also available is Hedge Fund Modelling & Analysis Using Excel and VBA (Published 2012) with Hedge Fund Modelling & Analysis Using C# due to follow in 2016.

Jacob Bettany: Could you begin by introducing yourselves and telling us a little bit about your background?

Paul Darbyshire and David Hampton: We are both grounded mathematicians/computer modellers/physicists/engineers at heart – so a great match for the trilogy we have written in Excel/VBA, MATLAB and C# (due out 2016) around the core subject of how to analyse and model hedge funds and CTAs. We met at EDHEC Business School around 2007 where I (David) as Director of the Financial Economics Masters courses (currently rated in top 5 in Europe by the Financial Times) was recruiting lecturers for a course on C++ for Quantitative Finance. Paul’s CV seemed spot on so I hired him for the job. The first day we had lunch together, and he told me he was planning to write a book on hedge fund modelling for Wiley Finance in Excel/VBA. I (David) had been working uniquely in quantitative finance since my first job at Bank of America after PhD/MBA graduation with a particular interest in hedge funds/CTAs, their modelling and analysis using mathematics and Excel/VBA. I (Paul) had finished a PhD in Theoretical Physics and then went on to work in Quantitative Finance and Risk Management in the City and on Wall Street for various top-tier investment banks. So, it seemed like a great opportunity for a co-author partnership and our writing duo was born. It took us a few years to gather the data, models, approach and the first book of the trilogy was finished and subsequently published in 2012. The second book in the series, using MATLAB as the programming language, came out a couple of years later and was published in 2014. The current book, using C#, is due out in 2016.

JB: How did your books come about, why were they necessary? Who are they written for?

 PD & DH: I (David) had been lecturing quantitative methods and software programming in finance at the MSc level at Skema Business School in Sophia Antipolis for quite a few years before I was hired by EDHEC Business School in Nice to lecture financial modelling and analysis. I (Paul) had been using a variety of programming languages and mathematical models for developing exotic option trading systems and their risk management. There seemed to be a gap in the market for suitable student as well as practitioner based textbooks. We had the idea to create a series of textbooks which would highlight the most useful accepted theories and models used in both academia and industry with data and examples in each of the major programming languages.

I (David) had been doing original research on long short alpha estimation since 2000 and was able to publish the results in the first book – a kind of “standard model” for asset pricing of hedge funds. The books do not tell you how to create a trading strategy – rather they allow you the hedge fund manager or investor, to measure alpha – skill related useful returns which are worth paying for. Many a hedge fund manager will find they are not producing alpha if they use Hampton’s alpha for estimation, whereas Sharpe’s alpha or Modigliani and Modigliani’s versions would be positive. Hampton’s alpha represents the first “standard model” of alpha pricing in all moments, since it is the first to take the first and second co-moment of returns into account – something critical for modelling hedge funds due to their ability to take long and short positions from a basket of underlying instruments which are often positively correlated.

JB: How are the books structured, and what was your approach to writing them? Did you face any particular challenges?

"The books do not tell you how to create a trading strategy – rather they allow you the hedge fund manager or investor, to measure alpha – skill related useful returns which are worth paying for."

PD & DH: The books are all structured along a common theme which allows the reader to progress naturally in order to fully understand the subject matter e.g. introduction to the industry and data sources, statistical analysis, performance metrics, asset pricing and finally risk management. The key challenge for asset pricing was related to my (David) research of finding a pricing mechanism for alpha which would include all possible statistical moments, and in particular the leverage effect and the reduction of volatility caused by the correlations between long and short positions. The solution was included in book

one. For Paul, the main challenge was in integrating theory and computational analysis in a clear and concise manner which we believe comes across in the books.

JB: What supplementary materials are supplied with the books?

PD&DH: Full data sets of realistic hypothetical hedge fund returns data (to get a feel for modelling) along with real life global diversified futures prices, French stock prices and Hurst exponent data. All code examples, data sets and sample chapters are available for download from our website: www.darbyshirehampton.com. At this website, there are also further details on each book and contact details for any questions regarding the books and to get in touch to discuss any consultancy requirements. There is also a short You Tube video for the Excel/VBA book at: https://www.youtube.com/watch?v=Y7kDNJEpj4Y.

JB: Why did you choose to work with MATLAB? What makes it particularly useful for this kind of modelling?

PD & DH: MATLAB is particularly useful due to its inherent use of matrix manipulations as a scientific computational language. We made use of all possible toolbox extensions, including statistics, optimisation and financial toolboxes. It also has an Excel link, so it can be used seamlessly with Excel/VBA so that both the MATLAB and Excel/VBA books complement each other nicely.

JB: Which features of hedge fund modelling are unique to hedge funds? What makes these features challenging?

"We did not want to cannibalise existing texts, for example the Chartered Alternative Investment Analysts (CAIA) guides or textbooks by Lhabitant etc. Rather, we wanted to complement these texts…"

 PD & DH: For me (David), an asset pricing quant, the long only models of Sharpe, Jensen and Merton were a step in the right direction since they were based on statistical mechanics. However, they lacked the leverage invariance needed to correctly measure a hedge fund manager’s skill – since leverage is not skill. This was rectified in the Modigliani and Modigliani and Graham and Harvey models all of which are included in the MATLAB book with complete code and worked examples. I had the chance to be involved with EDHEC Risk which was involved in hedge fund modelling with

interesting models and studies developed by Amenc, Garcia, Lhabitant, Martellini, Miffre and Ziemann at that time. My contribution to the “standard model” of hedge fund alpha estimation came when I was able to finish it off by including the first and second co-moment into the equation (from Markowitz) – modelled by the average long short correlation coefficient and modelled in theory by the H-function introduced in book one (Excel/VBA).

JB: Are there any areas which you didn't cover in the books but which deserve attention elsewhere?

PD & DH: We did not want to cannibalise existing texts, for example the Chartered Alternative Investment Analysts (CAIA) guides or textbooks by Lhabitant etc. Rather, we wanted to complement these texts with something that could be used at the graduate level by students or industry practitioners alike who wanted to understand the theory, underlying statistics, asset pricing and risk management, and who also wanted to develop their own models using various industry standard computing languages.

JB: The hedge fund industry represents a hugely varied and disparate range of firms and strategies. To what extent can a book like this prepare a reader for whichever scenario they encounter?

PD & DH: Yes there are a vast number of hedge funds but they are largely based on only a handful of underlying building blocks e.g. equity, fixed income, currencies, commodities and their derivatives. So just as physics is about reducing all known physical phenomena down to basic underlying principles as advised by Newton (in his book Principia), so these thousands of hedge funds can be reduced to one standard model of asset pricing, and their return probability distributions reduced to one Extreme Value Theory (EVT) risk model including fat tails of which the Gaussian case is a special case.

"…computation is a great way to learn numerous mathematical and financial concepts, as seeing a particular model take shape in a computer is by far the best way to understand the underlying fundamentals."

 JB: What, in your view, is the difference between a good model and a great model?

PD & DH: The difference is in the model’s specification. To be a great model, the model must be highly accurate or well specified in the global sense. A great model should take into account all the moments of the return distribution as well as other outside performance factors, and not just the main market factor. Good models tend to do more harm than good but are better than nothing and can be used as a fundamental building to more useful and realistic models.

Consider the current state of physics at the moment, one gets the feeling that the subject is in dire straits and need many of their fundamental models updating to become great models. For example, what if there is no dark energy or matter and their effects as seen at the cosmic level are due to having inexact “good” models at the subatomic quantum level, but not great models? We both believe physics needs a new “great” standard model. Einstein was perhaps only partially right when he concluded that space-time warped with mass and such a warp caused gravitation in space-time since he assumed a gravitational field was identical to being in an accelerated inertial field. Einstein’s theory is a good model but the fact remains where does mass come from? Is it really the Higgs mechanism or something else? What causes increasing entropy in space? What about fractal emergence at all scales as pointed out by Mandelbrot? When we know these answers and integrate them all seamlessly, we will then have a great standard model of physics.

Read more author interviews on MoneyScience...

Thammarak Moenjak author of Central Banking: Theory and Practice in Sustaining Monetary and Financial Stability

Dark Pools and High Frequency Trading For Dummies with Jay Vaananen

JB: Computational finance is a discipline in the ascendancy. Do you have a view on this and the extent to which students are being prepared for these challenges?

PD & DH: As I (David) was a key architect of the range of MSc programmes on offer at EDHEC Business School, I made sure computation finance was on the menu for all finance masters students – with courses on Excel/VBA, MATLAB and for the more advanced market based Masters courses, C++. Computation is the other half of the mathematics and statistics sphere and any graduate course which does not teach such techniques is incomplete in our opinion. In fact, computation is a great way to learn numerous mathematical and financial concepts, as seeing a particular model take shape in a computer is by far the best way to understand the underlying

fundamentals. Indeed, many cornerstones of financial theory such as advanced risk analysis, forecasting, option pricing and portfolio optimisation rely almost entirely on computational methods.

Hedge Fund Modelling and Analysis using MATLAB is out now published by Wiley. You can get 30% off your copy by ordering direct from Wiley here and entering our promo code MON30 at the checkout.