The ETF correlation matrix is the web's first 'live' correlation matrix that
gives us the ability to understand the cross-correlations on various markets,
sectors and asset classes. It is a 37x37 matrix that utilizes a two year daily
return series and is thus based off 1 million plus unique data points. A new
addition is the Top and Bottom 25 paired correlations. A high pairwise
correlation is a high Sharpe ratio proposition, while a low pairwise correlation
scenario is a high Alpha potential. The availability of large-scale computing
power and the web as the new platform for information dissemination enables
us to put this on the web, something that until recently was the preserve of
specialized knowledge providers.
Now we are making available the ETF Variance-Covariance matrix, based on
daily returns. We think that a longer interval period (such as monthly) results in
information loss, especially in todays fast paced trading environment. Legacy
issues probably constrain the generation of high frequency Var-Cov matrices,
we hope to provide unique value added by operating outside such parameters.
These correlations* can be utilized to structure an optimal asset allocation
framework, prevent redundant diversification, structure market-neutral spread
trades, and control risk in a portfolio.
* Correlations are subject to the well-known phenomenon of inter-temporal non-
stationarity and sign inversion. As such they are best utilized as an indicative
and explanatory factor, rather than a predictive factor.
Other research on a sample of advanced finance and mathematical topics is
also presented here, for the interested viewer. The Bootstrapping routine
allows for repeated sampling to estimate a normal distribution from a non-
parametric or small-size sample - a fully functional downloadable file written in
Excel VBA is attached. Such methodologies are utilized in our work to generate
distributions for ETF's that have short trading histories. This is essential for
fast paced hedge funds and traders who cannot afford to exclude an asset just
because it is a new issue.
These are the views of Advanced Portfolio Solutions and are not to be
construed as investment advise of any sorts. This work is purely for the
purposes of research and understanding financial market dynamics. contact
Pankaj Agrrawal, Ph.D. for clarification on any technical aspects of this
(global markets,sectors, bonds, gold)
With Top and Bottom 25 Pairwise Correlations for efficient Hedging
Multi-market, multiple frequency (daily, weekly, monthly betas for S&P 500, MSCI-EAFE,
Emerging Markets and Russell 3000 ETF's) - Based on Institutional Investor Journals
published paper (2007)
ETF Variance-Covariance Matrix
in double precision based off daily (as opposed to monthly) returns
Sector Deviation Analysis
(extreme divergence from mean path) - Sample output from proprietary
US and Japan Bubble Dynamics
Liquidity Driven Economic Cycle
Annualizing Volatility & Beta Calculation (Download File)
Black Scholes Options PM (dynamic sheet)
Compound Linking of Returns (Beta Multiples) - Puzzle?
Bootstrapping a Non-Parametric distribution: Excel VBA application (with
fixed/variable Seed option)