Personalized direct indexing with tax-loss harvesting has traditionally been limited to high net worth investors, primarily due to the cost and effort associated with managing custom portfolios at scale. Since many of the portfolio management tasks associated with direct indexing are operational, AI has the potential to help. In this first part of our AI in Portfolio Management series, we discuss our experience using AI in our Portfolio Monitoring system. Even though Portfolio Monitoring is just one component of the daily portfolio management process, the overall efficiency gains we witness are significant, 82% savings in portfolio manager time and 63-85% savings in net computational costs, on average. Thanks to recent advances in large language models, our system is able to effectively generalize to new conditions with minimal or no in-context data. Forthcoming notes in this series will demonstrate how the use of AI in other areas of the portfolio management process can potentially lead to further efficiency gains.

Tax-loss harvesting has become an important part of many financial advisors’ toolkits. When implemented systematically in direct index portfolios, it can add significant value to clients through reduction of overall capital gains taxes – whether the overall objective of the client is to improve after-tax returns or to diversify away from a large appreciated stock position. In this paper, we discuss how traditional tax-loss harvesting strategies can be further enhanced through long/short extensions.

American depositary receipts (ADRs) offer a convenient way for investors in separately managed accounts to gain exposure to international equities. However, existing ADR indices can be expensive to trade because liquidity in an ADR can be much lower than expected based on the firm’s market capitalization. The Nasdaq-Brooklyn ADR Index presents an innovative solution that creates a liquidity-adjustment factor to optimally utilize liquidity across ADRs while maintaining broadly diversified international exposures. This significantly reduces trading costs for investors.

Since wealth tends to be created by entrepreneurs who embrace concentrated risks, prospective clients of financial advisors often hold appreciated stock positions that they wish to sell down in a tax efficient way. In this white paper for investment advisory professionals, we will discuss our approach to managing the tax-efficient diversification of such outsized positions toward a desired target portfolio.

The case for diversifying out of concentrated positions is even stronger than most investors realize. Outsized positions in individual stocks not only increase portfolio volatility, but usually they also reduce portfolio return. Over the past century, the median ten-year return on individual U.S. stocks relative to the broad equity market is –7.9%, underperforming by 0.82% per year. Stocks with strong prior performance have underperformed even more, by 1.94% per year.

We review opportunities for taxable investors to improve the outcomes of their overall portfolio.

How Can Tax Loss Harvesting and Custom Direct Indexing Enhance Portfolios? A Conversation with Brooklyn’s Chief Investment Officer Erkko Etula and Head of Equities Antti Petajisto, Q1 2023.

We document that the prices of exchange-traded funds can deviate significantly from their net asset values, especially in funds holding international or illiquid securities. We introduce a novel approach to control for stale pricing of the underlying assets and confirm that the mispricings are real. Cost-conscious ETF investors should be aware of these potential hidden costs.

We find that the index premium for the S&P 500 and Russell 2000 has historically averaged +8.8% and +4.7% for additions, respectively, and -15.1% and -4.6% for deletions. We introduce a new concept that we label the index turnover cost, which represents a hidden cost borne by index funds (and the indexes themselves) due to the index premium. We estimate its lower bound as 21-28bp annually for the S&P 500 and 38-77bp annually for the Russell 2000, which is much larger than the management fees of many index funds.

We use Active Share and tracking error to sort active mutual funds into style groups. We find that the most active stock pickers have outperformed their benchmark indices even after fees and transaction costs, while closet indexers have been reliable underperformers after expenses. Closet indexing increases in volatile and bear markets and has become more popular over time.

Standard academic factor models produce economically and statistically significant nonzero alphas even for passive benchmark indices such as the S&P 500 and Russell 2000. We track down the issues behind this, propose alternative models and show that they outperform the standard models in common applications such as performance evaluation of mutual fund managers.

We find that a simple global long-short strategy betting on high earnings quality produces a higher Sharpe ratio than the overall market or similar strategies betting on value or small stocks. Because the earnings quality portfolio has a negative correlation with a value portfolio, an investor wishing to invest in both exposures can achieve significant diversification benefits.

We present broad-based evidence that the monthly payment cycle induces systematic patterns in liquid markets around the globe. We document temporary increases in the costs of debt and equity capital.

Financial intermediaries trade frequently in many markets using sophisticated models. Their marginal value of wealth should therefore provide a more informative stochastic discount factor than that of a representative consumer.

Strategic asset allocation is arguably one of the most important, yet least advanced, aspects of investing. We present a new approach to strategic asset allocation that leverages the idea that long-term investment returns derive from multiple distinct sources.

We show that the risk-bearing capacity of U.S. securities brokers and dealers is a strong determinant of risk premia in commodity derivatives markets.

We decompose the U.S. dollar risk premium into components associated with macroeconomic fundamentals and a component associated with financial intermediary balance sheets.

We describe use cases of blockchain technology in finance.

We develop a machine learning method for automatically generating models of dynamic decision-making that both have strong predictive power and are interpretable in human terms.

We show that the fundamental legal structure of a well-written financial contract follows a state-transition logic that can be formalized mathematically as a finite-state machine. By conceptualizing and representing the structure of a financial contract in this way, we expose it to a range of powerful tools and results from the theory of computation.

We explain how computable contracts, coupled with automation, can drive innovation in the insurance business.

We present the first computational model of human behavior in repeated Prisoner’s Dilemma games that unifies the diversity of experimental observations in a systematic and quantitatively reliable manner.

Climate-contingent finance is a fresh approach to addressing catastrophic risk, building a bridge between long-term funding needs and financial risk management. It can be generalized to any situation where multiple entities share exposure to a risk where they lack direct control over whether it occurs (e.g., climate change, or a natural pandemic), and one type of entity can take proactive actions to benefit from addressing the effects of the risk if it occurs (e.g., through innovating on crops that would do well under extreme climate change or vaccination technology that could address particular viruses) with funding from another type of entity that seeks a targeted financial return to ameliorate the downside if the risk unfolds.

We define Environmental Impact Bond mechanics, elucidate the difference between EIBs and Green Bonds, and propose a common vocabulary for the field.

We present a simulation model as a computational test-bed for climate prediction markets. Traders adapt their beliefs about future temperatures based on the profits of other traders in their social network.

We built an open-source AI tool to produce short-term forecasts of vegetation health at high spatial resolution, using data that are global in coverage.

Our simulation models show that a farmer using seasonal forecasts has more diversified crop selections, which drive increases in average agricultural income.

We review models and literature regarding climate change adaptation, and present an argument in favor of using various computational modeling tools.