The Computational Finance in R meeting, held December 4th at Columbia University in New Your City, was a great success. Over 100 people attended the meeting from Columbia University, Wall Street, Chicago and even further afield. The meeting was organized by Columbia University and sponsored REvolution Computing.
Many thanks go out to the organizers for putting together such a fine event, in particular Jan Vecer and Libor Pospisil from Columbia University and Krishna Kumar from Barclays. Thanks too to Linda Heinig from REvolution for her help setting up the event -- and especially arranging the drinks and delicious snacks during the reception afterwards! I was very proud to represent REvolution as the emcee of the event. Given that REvolution was founded with the mission of promoting the use of R in commercial environments, it was really great to see so many members of the financial community turn out in such numbers and to participate in such great discussions.
But the lion's share of the thanks goes to each of the seven speakers for delivering such excellent presentations. If you missed the event, the presentations are available for download from the REvolution Computing and Columbia University sites, and I've provided brief summaries below. (These are based on my brief notes taken during the sessions, so apologies for any errors.)
Whit Armstrong gave a presentation about discount curve construction in R and presented several packages that the maintains in R for the financial community, including fts (fast operations for time series, an interface to tslib, a c++ library for time series operations), RLIM (an R interface to the LIM financial time series data source), RFincad (an interface from R to the fincad library of financial functions), and RAbstraction (C++ abstractions for R objects). Whit works at KLS Diversified Asset Management, a hedge fund in New York City. Whit previously was employed at Highbridge Fixed Income Opportunities Fund as a quantitative analyst where he designed relative trading models for emerging foreign-exchange and fixed incomes.
Anthony Brockwell gave a very helpful introduction to the secretive hedge-fund industry, discussing amongst other things risk incentives for hedge fund managers and the type of investments they purchase for their funds. He also talked about the process of developing strategies for funds using statistical arbitrage. Anthony is a Senior Analyst at Horton Point LLC, a hedge fund management group based in New York City, where he develops and implements automated quantitative trading strategies. Anthony is also an Associate Professor of Statistics (currently on leave from Carnegie Mellon), and has be published articles in such journals as SIAM Journal on Control and Optimization, Journal of Time Series Analysis and Journal of Computational and Graphical Statistics.
Bryan Lewis gave an introduction to REvolution Computing and gave several examples of the performance benefits available in REvolution R, on single-core and multi-core workstations, and distributed clusters (with ParallelR). Bryan is Director of Systems Engineering at REvolution Computing. Formerly the CEO of Rocketcalc and with a Ph.D. in applied mathematics, Bryan has extensive experience in high-performance computing and its use in industry and science.
Scott Payseur's presentation highlighted how R can be used to make quant funds truly use unique estimation methods, as opposed to the common practice today where most funds use the same off-the-shelf optimization techniques based on the same covariance data (such as available from BARRA). Using the techniques Scott described, funds can actually develop their own techniques for estimating and incorporating multidimensional correlation into their models. Scott is a Quantitative Analyst at UBS Global Asset Management working on High Frequency Volatility Estimation. Scott previously worked at Insightful Corporation, where he was a Senior Financial Engineer. He was also an economics instructor at the University of Washington in Seattle, and a Senior Software Engineer at Sharebuilder.
Brian Peterson and Peter Carl have an entertaining joint presentation on the issue of performance management of funds. While many of the presentations focused on the tools available to an individual manager to forecast and optimize their own returns, Brian and Peter tackled the topic of comparing a universe of funds and establishing metrics for comparing their performance. They illustrated their presentation with examples created using the PerformanceAnalytics package they created and maintain. Peter Carl started his career with O'Connor and Associates in 1986 where he worked for 15 years through mergers with Swiss Bank Corp. and UBS. During that time, Peter worked closely with The Prediction Company to create a proprietary statistical arbitrage business. He left UBS and helped start a hedge fund doing Tactical Asset Allocation. He currently works with a Hedge Fund of Funds based in Chicago doing manager selection as a member of the investment committee. Peter's colleague Brian Peterson is a senior quantitative analyst and researcher who has published in venues including Risk and the Journal of Risk.
Jeff Ryan gave a visually arresting presentation of his Quantmod package, including many graphical presentations of financial data. (This directly addressed a wish that Anthony Brockwell expressed during his talk, that R have flashier graphics for time series.) Jeff is the owner of Insight Algorithmics, which specializes in custom quantitative software solutions for small to mid-sized proprietary trade desks and hedge funds. Jeff previously worked as an equity options market maker. He is the organizer or on the organizing committee of the Chicago R-Finance Conferences and Rmetrics 2009.
Comments