R/Finance March Meetup – Exploring high dimensional asset dependence through Vine Copulas

Join us from 4.30pm on Thursday, 23 March at Rise Cape Town when Hanjo Odendaal (Data Scientist – Eighty20 Analytics) will be giving a talk on copulas and their growing usage in the fields of Finance and Economics.

For an introduction to the topic you can view Hanjo’s lightning talk at the recently held R conference in Cape Town, satRday here.

Copulas are a popular framework for both defining multivariate distributions and modeling multivariate data. A copula characterizes the dependence—and only the dependence—between the components of a multivariate distribution; they can be combined with any set of univariate marginal distributions to form a full joint distribution. Consequently, the use of copulas allows us to take advantage of the wide variety of univariate models that are available. The primary financial application of copula models is risk assessment and management of portfolios that contain assets which exhibit co-movements in extreme behavior. For example, a pair of assets may have weakly correlated returns, but their largest losses may tend to occur in the same periods. They are commonly applied to portfolios of loans, bonds, and collateralized debt obligations (CDOs).

The concept of copula has received growing attention in finance and economics in recent years. From the early days of use in finance over copulas finding their way to Wall Street in a mass market of credit derivatives, this episode of quantitative modelling of markets was also one of euphoria, exaggerations, misconceptions and debates. Bringing tools from the “hard-sciences” like mathematics or physics to a “soft-science”, i.e. to use formulas and models for financial markets or the behaviour of market participants always bears a certain portion of model risk. But the complexity and dynamics of financial markets makes it necessary to employ those tools and thereby improve existing methods. The case of copulas in finance is a typical example of how generally appealing and sensible models may cause unforeseen problems and can lead to debates on the applicability of quantitative methods.

Registration is from 4.30pm with the presentation set to begin at 5pm. The talk should last roughly 40mins after which Hanjo is willing to take some questions and the rest of the time until 6.30pm is available for networking.

PLEASE NOTE: Seating is limited to 40 people and therefore subject to RSVP.

Thanks to Rise Cape Town for hosting us as well as for their generous hospitality!

Is it Real? Or is it Random? A Financial Turing Test of the JSE

R/Finance November Meetup

Join us from 4.30pm on Thursday, 24 November at Rise Cape Town where Stuart Reid (Quantitative Strategist at NMRQL and blogger at Turing Finance) will be presenting his R implementation of the heteroscedasticity-consistent variance ratio test developed by Lo and MacKinlay in 1988 to test the Random Walk Hypothesis. Stuart will then test whether or not the variance ratio test can distinguish between random fictitious securities and those traded on the JSE. Stuart will end off with some theories which could explain the results of the test, including the controversial Stable Paretian Hypothesis presented by Benoit Mandelbrot in 1963.

Some more to whet your appetite from Stuart…”In his now legendary Turing test of the weak-form Random Walk Hypothesis esteemed professor Burton G. Malkiel had his students flip a coin and record an increase or decrease in the price of a fictitious random security. The resulting equity curve was presented to a chartist who exclaimed that it was a massive buying opportunity. To quote Nassim Taleb: “the chartist was fooled by randomness”. The problem with this test is that is doesn’t actually imply that security prices are random, it only implies that at least some chartists can’t tell the difference between what is real and what is random.”

Registration is from 4.30pm with the presentation set to begin at 5pm. The talk should last 45mins after which Stuart is willing to take some questions and the rest of the time until 6.30pm is available for networking.

PLEASE NOTE: Seating is limited to 40 people and therefore subject to RSVP.