Projects
Research in Progress
-
Pattern shift recognition with time-varying parameters
Patterns in data come and go. Even a string of random numbers can exhibit a pattern for a short while and then it disappears. Our research in financial risk management, strategic asset allocation, and quantitative portfolio models depends critically on identifying when old patterns die and new patterns are born.
We work with several different pattern identification tools, and our current research is exploring the use of time-varying parameters (TVP) applied to several different financial market questions. Our version of TVP is essentially a statistical multi-factor regression. We time decay the data using an exponential process and a specific half-life to emphasize recent data over older data. We work through a time series of data one step at a time, allowing the TVP process to re-estimate the coefficients at each step, so we can track their stability and observe when a pattern break occurs. we have applied this approach to interest rates, equity indexes, commodities (from metals, to energy, to agriculture), and even cryptocurrencies.
Download Bitcoin and Equities using TVP analysis
-
Federal Reserve Forward Guidance
Back in the day, Federal Reserve (Fed) Chairs, such as Paul Volcker or Alan Greenspan, did not provide much in the way of forward guidance in terms of telegraphing to market participants what might happen at the next Fed FOMC (Federal Open Market Committee) meeting to set interest rate policy. Ben Bernanke was appointed Fed Chair in 2006, and he embraced forward guidance, at least in terms of what might happen at the next FOMC meeting. Chairs Yellen and Powell continued the forward guidance policy, at least until late 2025 when the Fed became more divided and embroiled with the White House over the appropriate course of short-term interest rates.
We use federal fund futures price data from CME Group. Federal funds futures were introduced in the mid-1980s, and became extremely popular both for speculation and for risk management. There is a monthly maturity curve available with prices for each forward-looking month from the current month through several years into the future. Our research focuses on the upcoming FOMC meeting as well as a meeting abut 12-18 months into the future. In our preliminary research one can easily identify the start of forward guidance by a material error reduction in how well federal funds futures anticipate interest rate policy, but the improvement occurs only for the upcoming meeting. Further out meetings are still subject to large errors, which is consistent with the Fed's embrace of data dependency in making its decisions. More recently, errors in the prediction of upcoming meetings have started to increase as the FOMC has become more divided about the appropriate interest rate policy.
-
Portfolio construction
We have been involved in portfolio construction research for several decades, starting in the early 1990s with global macro investing projects for the Chase Manhattan Bank, and later with Bankers Trust, the New York asset management subsidiary of Caisse des dépôts et consignations (CDC, Paris), Nikko Asset Management, among others. Current research focuses on commodity markets and how to construct trading systems that are largely uncorrelated with equity markets for use as a possible asset class addition to traditional equity-focused portfolios. The research with commodities is in its early stages yet appears highly promising.
We are using our time-varying parameter methods (see above) to simultaneously estimate one-period-ahead trading strategies involving metals (gold, silver, copper), energy (oil and natural gas), and agriculture (corn, soybeans, and wheat). All of the commodities in our selected universe have active and extremely liquid futures and options markets operated by CME Group. We estimate our multi-factor forecasting equations using a seemingly unrelated regression (SUR) process incorporating TVP. The SUR process allows us to continually update our portfolio risk analysis in terms of both specific commodity risks and cross-commodity correlations. Correlation relationships between the pairs of securities in the portfolio can be notoriously unstable. By dynamically updating our risk and correlation forward-looking estimates in a manner consistent with our return estimates, we have been able to make material improvements in long-term performance and in drawdown reductions (i.e., tail risk).
Our commodity portfolio research is a work in progress. We expect to issue a white paper describing our progress in a few months.
-
Unconstrained Market Risk-Return Probability Distributions
We believe that risk-return probability distributions for most financial assets as well as commodities are often highly skewed (i.e., elevated tail risk) or even bi-modal reflecting two competing and distinctly different narratives circulating among market participants. Most risk models and processes that rely on volatility metrics, such historical or implied options volatility, often embed hidden biases toward bell-shaped probability distributions that regularly and disastrously under-estimate tail risk or the elevated probability of large, abrupt price gaps. Our research aims to identify periods in which tail risk and probabilities of large price gaps (up or down) are unusually elevated. We use data from options, looking at all the strikes to spot asymmetries or anomalies in volume and open interest patterns. We also incorporate information from intra-day trading, especially where the high-low price spread intra-day is significantly elevated relative to past history.
We are looking for a sponsor for our risk distribution research. It is interesting to show how probabilities for a given outcome change over time. It is even more interesting to show visually how probability distributions evolve over time. This research has many potential applications for animated websites on risk, uses in risk management of quantitative or judgement-based portfolios, etc.
White paper coming soon.
-
Intersection of prediction markets, event contracts, and sports betting with traditional financial trading
The worlds of traditional stock trading and sports betting are coming together, for better or worse, in the form of prediction markets and event contracts. The regulatory landscape is in flux and unsettled. Sports betting (gambling) in the US is regulated at the 50-state level, as are insurance companies. Stocks, options, and futures trading are regulated at the national level by either the Securities and Exchange Commission (SEC) or the Commodity Futures Trading Commission (CFTC). We are available for expert consultation on how these markets are intersecting and evolving.