Practitioner Research

Alongside academic work, I spent two decades producing research for practitioners and putting research to work.

If you are a J.P. Morgan client, you can still find many of these papers on their website, covering a range of topics in quantitative strategies (searching the titles may also work, as some universities have used them in MBA courses). Some of the titles:

• Risk Premia in Volatility Markets: Exploiting Volatility Spillover and Clustering, 2012. Abstract

  Selling implied against realized volatility earns a premium in calm markets but takes large losses when volatility spikes. The paper builds econometric models that forecast the probability of a positive volatility risk premium in each market, exploiting clustering (high volatility persists) and spillover (volatility shocks transmit across related markets). A strategy that rotates short-volatility positions across US, European, and Japanese equity markets, and stands aside when the premium is not clearly positive, beat a constant short S&P 500 volatility position by roughly 5% a year since 2001. The same framework extends to currency volatility, where equity-market information improves a short FX volatility strategy.

• Commodity Equities or Futures?, 2011. Abstract

  Investors wanting commodity exposure can hold futures, with their roll costs, or commodity producer equities, which pay dividends but track spot prices less closely; the paper asks when each is the better hold. A monthly rule compares the carry of each commodity's futures curve (its slide) with the yield of the matching commodity-sector equities and holds whichever is higher, with a variant that shorts a global equity basket to strip out the excess equity beta. The long-short version earned a Sharpe ratio of 1.75 since 2002 and the long-only version 1.02, against 0.05 for the S&P GSCI over the same period. The gap is attributed to mispricing in the commodity curves and segmentation between commodity and equity markets.

• Economic and Price Signals for Commodity Allocation, 2009. Abstract

  Can macro activity data time commodity index exposure, and how do such rules relate to price-based momentum and slide signals? The signals are global industrial production growth, the global manufacturing PMI, and PMI components such as new orders, applied as long-short or long-neutral rules on the S&P GSCI with monthly rebalancing. A rule based on the level of PMI new orders returned around 20% a year with a Sharpe ratio near 0.9 over the prior decade, while long-only GSCI exposure was close to flat. Economic and price signals agree in normal times but diverge at major turning points, so combining them improves risk-adjusted results.

• Longevity Risk and Portfolio Allocation, 2009. Abstract

  Pensions and annuity books want to shed longevity risk; the paper asks what investors gain from taking the other side through instruments like mortality forwards and longevity swaps. It measures correlations between changes in US, UK, and Japanese mortality and equity and bond returns, examines tail events back to 1900, and uses mean-variance analysis to back out the return a short longevity position needs to earn its place in a portfolio. The correlations come out close to zero and statistically insignificant, so shorting longevity diversifies a portfolio even at a near-zero expected excess return. It also hedges pandemic scenarios, which hurt equities while paying off short longevity positions.

• Profiting from Slide in Commodity Curves, 2009. Abstract

  The shape of a commodity futures curve says surprisingly little about where spot prices go, and the paper asks whether the slope can nonetheless be traded profitably. It tests long-short strategies that buy the most backwardated and sell the most contangoed commodities or contracts, choosing where along the curve to position so that gains come from contracts sliding down an unchanged curve toward expiry. Simple versions deliver information ratios around 1.5 over the prior decade, and a curve-slide strategy trading two maturities per commodity posted a Sharpe ratio of 3.5 in 2008. The returns look like a risk premium, with index flows, over-anticipated mean reversion, and hedging demand as candidate explanations.

• Commodity Prices and Futures Positions, 2009. Abstract

  After the 2007–08 price spike raised suspicions that financial speculation had become the main driver of commodity prices, the paper uses the CFTC's newly released disaggregated Commitments of Traders data to test that claim. It relates weekly price changes to changes in the net positions of money managers and swap dealers, adds economic controls such as the dollar and inventories, and estimates VAR models to trace the full dynamics. Positions and prices do move together week to week, but the link fades at monthly to quarterly horizons and largely reflects both responding to the same supply and demand news; funds and banks were in fact cutting oil longs while oil rose from $80 to $145. Position changes unrelated to fundamentals have only a small, short-lived price impact, arguing against speculation as an independent driver of price levels.

• Combining Directional and Sector Momentum, 2009. Abstract

  Can a market-level trend signal improve global equity sector momentum? Each month the MSCI World is compared to its own 12-month average: above it, the strategy runs long-only sector momentum (top 3 of 10 sectors by past 12-month return); below it, it switches to the long-short version, or shorts the index itself when shorting sectors is impractical. The conditional strategies produced Sharpe ratios around 0.9 to 1.0 since 1996, against 0.03 for the MSCI World, and the long-short sector strategy earned 24.3% in 2008. A dynamic Markowitz version that re-estimates inputs monthly under a 10% volatility cap performs similarly with more stable risk.

• Volatility Signals for Asset Allocation, 2008. Abstract

  Does scaling exposure to an asset by its recent volatility beat holding a constant weight? The rule is simple: estimate volatility from exponentially weighted daily returns and size the position each day so the portfolio targets the asset's long-run volatility, deleveraging into cash when markets turn turbulent and releveraging when they calm. Since 1990 the approach added about 2.7% a year over the S&P 500 and 1.3% over the JPMCCI commodity index, while cutting kurtosis and improving skewness across equities, bonds, credit, and commodities. Because volatility-controlled indices carry less tail risk, options written on them are also cheaper, a more efficient way to buy upside exposure.

• Timing Carry in US Municipal Markets, 2008. Abstract

  Muni carry trades earn steady income but were punished in the 2007–08 crisis; the paper asks whether the carry can be kept while sidestepping the periodic losses. The base trade receives percentage Libor in a constant-maturity 10-year BMA/Libor ratio swap, exploiting the muni curve's steepness relative to Libor, and exposure is then scaled by three signals: the level of carry, monetary-policy expectations, and momentum. The static trade earned 6.1% a year (Sharpe 0.43) over 1995–2008 with a 53% maximum drawdown; timing on the carry level alone lifts the Sharpe ratio to 0.73, and combining signals reaches 1.35. The dynamic versions also show low correlation to major asset classes.

• Cross-Momentum for EM Equity Sectors, 2008. Abstract

  Are emerging-market equity sectors better traded on their own past performance or on that of the same sectors globally? The strategy ranks EM sectors by 12-month returns, long the top group and short the bottom with monthly rebalancing, taking the ranking signal either from local EM sector returns or from global sector returns. Standard EM sector momentum earns an information ratio of 0.64; the cross-momentum version ranked on global sectors reaches 1.16, worth about 5 percentage points a year. The same improvement appears in Latin America, BRIC, Asia ex Japan, and Emerging Europe, consistent with slow diffusion of information into segmented markets.

• Momentum in Global Equity Sectors, 2008. Abstract

  Does past performance predict returns across global equity sectors, and is sector momentum better traded globally or region by region? The rule ranks 18 global sector indices by 12-month returns, buys the top third and shorts the bottom third, rebalancing monthly on data back to 1974. The market-neutral version earned 8.1% a year over cash (Sharpe 0.80), and a Markowitz-optimized version lifts the Sharpe ratio to 1.4, against 0.40 for the MSCI World. Global implementations beat local ones: lagged global sector returns forecast local sector returns better than local returns do, consistent with home bias and slow information diffusion across regions.

• Optimizing Commodities Momentum, 2008. Abstract

  A follow-up to Momentum in Commodities: can portfolio theory beat equal weighting? The paper applies dynamic mean-variance optimization to GSCI single-commodity indices, feeding in past 12-month returns as expected returns and a rolling covariance matrix, market-neutral with a 25% cap per commodity. Optimization raises the Sharpe ratio of the long-half/short-half momentum strategy from 0.76 to 0.99 over 1991–2008, and versions adding a volatility cap and seasonality push Sharpe ratios above 1.2. The optimized strategy also keeps portfolio risk stable as commodity volatilities and correlations shift, both of which turn out to be persistent and forecastable.

• Hedge Fund Alternatives, 2008. Abstract

  With institutional money flowing into alternatives, the paper weighs direct hedge fund investment against its cheaper substitutes: funds of funds, investable indices, replication, and rule-based investing. It reviews what each actually delivers, from the fee layers and database biases that flatter reported hedge fund returns to the lower cost, liquidity, and transparency of systematic products. A tactical combination of replication and momentum outperformed fund-of-funds indices over 1998–2007. Hedge funds and their alternatives serve distinct purposes and can coexist in an efficient portfolio, with the split depending on an institution's size, resources, and appetite for illiquidity.

• Markowitz in Tactical Asset Allocation, 2007. Abstract

  Mean-variance optimization is standard in long-term strategic allocation but rarely applied at tactical horizons; the paper asks whether it should be. It combines a cross-market momentum signal with a Markowitz efficient frontier built from the past six months of daily returns, risks, and correlations across ten asset classes, picking the frontier portfolio at a fixed volatility cap. The resulting strategy earned a Sharpe ratio of 1.37 over 1994–2007, against 1.13 for momentum alone and 0.77 for an equal-weight portfolio. The gains come from persistence in returns, volatility, and correlations, plus more stable total portfolio risk.

• Equity Style Rotation, 2006.
• Momentum in Commodities, 2006. Abstract

  Passive commodity index exposure had returned just 2.9% a year over cash since 1991; the paper asks whether rule-based momentum can do better. The benchmark rule ranks 24 GSCI single-commodity excess-return indices on past 12-month performance, buys the top half, shorts the bottom half, and rebalances monthly, so the signal combines price trends with the roll yield from contango and backwardation. That long-short rule earned 12.2% a year at 13.6% volatility over the same period, with little correlation to the direction of commodity prices. Relative momentum (commodity versus commodity) beats absolute and aggregate variants, and combining the three does better still.

• Exploiting Cross-Market Momentum, 2006. Abstract

  Most tactical asset allocation rests on value judgments; the paper asks whether past performance alone can guide the choice among asset classes. The rule ranks ten asset classes, from US and international equities to bonds, real estate, commodities, and hedge funds, by trailing 6-month returns, buys the top five equally weighted, and rebalances every six months. From 1994 through 2005 this added about 4% a year over holding all ten classes equally, with the best relative performance in periods of high volatility and high dispersion. The effect is attributed to underreaction, overreaction, and performance-chasing flows, strongest across dissimilar asset classes where relative value is hard to compare.