Abstract: Ask one hundred investors whether this is a bull market or a bear market, and you are likely to find their opinions split evenly down the middle. No one is really certain that the September 2001 low marked the end of the bear market and the start of a new bull market. But, this uncertainty is nothing new. As long as stock exchanges have existed, analysts and investors have always placed heavy emphasis on the difficult task of identifying the primary trend of the stock market. Everyone’s ideal market strategy is, at least in theory, to avoid the ravages of each bear market, and then to move aggressively into stocks after each important market bottom. To further maximize the benefits of a new bull market, time is of the essence. An investor should buy as close to the final low as possible. This is the ‘sweet spot’ for investors -- the first few months of a new bull market in which so many stocks rise so dramatically. But, theory and reality, especially in the stock market, are often entirely different matters. To bring this theoretical investment strategy to reality, an investor would need a time-tested method of identifying major market bottoms – as opposed to minor market bottoms – and would have to apply this method quickly, to capture as much of the bull market as possible. Traditional methods of spotting major turning points in the market often leave a great deal to be desired. The financial news typically remains negative for months after a new bull market has begun. The economic indicators offer little help since, historically, the economy does not begin to improve until about six to nine months after the stock market has already turned up from its low. Even some widely accepted technical indicators, such as 200-day moving averages or long-term trendlines, can sometimes take several months to identify a major turning point in the market. To spot an important market bottom, almost as it is happening, requires a close examination of the forces of supply and demand – the buying and selling that takes place during the decline to the market low, as well as during the subsequent reversal point.

Abstract: Using asset pricing models with structural breaks, we identify two structural breaks in the relationship between momentum returns and common risk factors used in the literature. After the second structural break, which occurred around the end of 2000, the profitability of the momentum strategy has disappeared. Our results are robust to different ways of constructing momentum portfolios.

Abstract: In this paper, we present an automated approach to discover patterns that can distinguish between sequences belonging to different labeled groups. Our method searches for approximately conserved motifs that occur with varying statistical properties in positive and negative training examples. We propose a two-step process to discover such patterns. Using locality sensitive hashing (LSH), we first estimate the frequency of all subsequences and their approximate matches within a given Hamming radius in labeled examples. The discriminative ability of each pattern is then assessed from the estimated frequencies by concordance and rank sum testing. The use of LSH to identify approximate matches for each candidate pattern helps reduce the runtime of our method. Space requirements are reduced by decomposing the search problem into an iterative method that uses a single LSH table in memory. We propose two further optimizations to the search for discriminative patterns. Clustering with redundancy based on a 2-approximate solution of the k-center problem decreases the number of overlapping approximate groups while providing exhaustive coverage of the search space. Sequential statistical methods allow the search process to use data from only as many training examples as are needed to assess significance. We evaluated our algorithm on data sets from different applications to discover sequential patterns for classification. On nucleotide sequences from the Drosophila genome compared with random background sequences, our method was able to discover approximate binding sites that were preserved upstream of genes. We observed a similar result in experiments on ChIP-on-chip data. For cardiovascular data from patients admitted with acute coronary syndromes, our pattern discovery approach identified approximately conserved sequences of morphology variations that were predictive of future death in a test population. Our data showed that the use of LSH, clustering, and sequential statistics improved the running time of the search algorithm by an order of magnitude without any noticeable effect on accuracy. These results suggest that our methods may allow for an unsupervised approach to efficiently learn interesting dissimilarities between positive and negative examples that may have a functional role.

Abstract: We propose a practical investment framework for Dynamic Strategic Asset Allocation (DSAA) across different economic regimes. The advantage of DSAA is that risk is constant across the economic cycle, whereas for static SAA risk tends to go up in bad times. In addition, a DSAA approach improves expected returns. Since a dynamic portfolio has time-varying portfolio weights we explicitly include transaction costs in the analysis. Using a sample consisting of US data from 1948 to 2007 we identify four phases in the economic cycle: expansion, peak, recession and recovery. We are able to predict NBER recessions and find that commodities and large growth stocks are attractive during expansions and government bonds and small value stocks during peaks. Cash and small value stocks are attractive during recessions and credits and commodities during recoveries. Finally, we show that the aim for stable risk across economic regimes leads to a different dynamic portfolio than the aim for stable outperformance.