This paper presents an approach to index portfolio re-balancing, focusing on the median slice of asset performance instead of the more traditional focus on “winners” and “losers.” In the proposed approach, one constructs an equal-weight portfolio from the index’s median (by returns) component. We consider the Dow Jones Industrial Average (DJIA) as a case study and introduce a systematic re-balancing strategy that targets the middle third (“median”) segment of asset performers within the index. The proposed methodology provides significantly better returns and mitigates drawdowns compared with a passive “buy-and-hold” strategy while promoting a disciplined and simple portfolio re-balancing strategy. The study empirically evaluates the effectiveness of the median-based re-balancing strategy over historical data. We compare our strategy to a benchmark portfolio closely tracking the DJIA index and other popular re-balancing strategies focusing on winners and losers, including the Dogs of the Dow strategy. We provide a simple model to compare factor loadings offered by such strategies and show that the median strategy provides broad diversification across sectors. We also analyze factor exposures of the proposed strategy using the 3-factor Fama-French model. The analysis demonstrates that the “median” rotation strategy consistently outperforms the Dow Jones and the broad market index S&P-500, yielding higher returns and reduced drawdowns.

In this paper, a shorter and more publication focused version of our recent article “A Bottom-Up Approach to the financial Markets” ( Mahdavi-Damghani, & Roberts, S. 2019 .) is presented. More specifically we propose a new approach to studying the financial markets using the Bottom-Up approach instead of the traditional Top-Down. We achieve this shift in perspective, by re-introducing the High Frequency Trading Ecosystem (HFTE) model Mahdavi-Damghani, B. 2017 . More specifically we specify an approach in which agents in Neural Network format designed to address the complexity demands of most common financial strategies interact through an Order-Book. We introduce in that context concepts such as the Path of Interaction in order to study our Ecosystem of strategies through time. We show how a Particle Filter methodology can then be used in order to track the market ecosystem through time. Finally, we take this opportunity to explore how to build a realistic market simulator which objective would be to test real market impact without incurring any research costs.

We examine to what extent the GICS sector categorization of equity securities may be systematically reconstructed from historical quarterly firm fundamental data using gradient boosted tree classification. Model complexity and performance tradeoffs are examined and relative feature importance is described. Potential extensions are outlined including ideas to improve feature engineering, validating internal consistency and integrating additional data sources to further improve classification accuracy.

We propose a novel deep learning architecture suitable for the prediction of investor interest for a given asset in a given time frame. This architecture performs both investor clustering and modelling at the same time. We first verify its superior performance on a synthetic scenario inspired by real data and then apply it to two real-world databases, a publicly available dataset about the position of investors in Spanish stock market and proprietary data from BNP Paribas Corporate and Institutional Banking. 1 , 2

Over the past ten years, applications of machine learning research have steadily shown mastery of broader and more complex domains, and the tools to create advanced models have become ubiquitous and more democratized. During that same time, the resources for studying algorithmic finance have also greatly expanded with populous open source communities and more granular time series available to researchers. However, progress in applying machine learning to finance has mostly stagnated, and in some cases regressed. We not only see similar, routine mistakes to those made ten years ago, but the complexity of cutting-edge models and frameworks can easily hide those mistakes. Drawing on both referee reports and published papers, this editorial series will review frequent problems that occur throughout the ML research pipeline which lead to invalid results and unwarranted conclusions.1 The pipeline refers to the series of research stages that are common to all ML projects and independent of any specific machine learning technique or architecture.

In the last two decades, the hedge fund sector has experienced a spectacular growth, up to the point that it is currently estimated to move more than 50% of the daily volume of stock markets. In contrast to other financial institutions, hedge funds are subject to less restrictive regulations which , in particular, allow them to sell short. As they exploit the asset mispricings, their action is thought to contribute to market efficiency. In this paper we aim at studying the impact that short sales have on the informational efficiency of a financial market. This can be used not only to assess the effect that hedge fund actions have in financial markets, but also the consequences of regulatory measures such as short-selling restrictions or bans. Building on an agent-based market, the simulation results indicate that short sales are beneficial to market efficiency, although the market does not become completely efficient even when all the population can sell short.

The underlying stochastic nature of the requirements for the Solvency II regulations has introduced significant challenges if the required calculations are to be performed correctly, without resorting to excessive approximations, within practical timescales. It is generally acknowledged by practising actuaries within UK life offices that it is currently impossible to correctly fulfil the requirements imposed by Solvency II using existing computational techniques based on commercially available valuation packages. Our work has already shown that it is possible to perform profitability calculations at a far higher rate than is achievable using commercial packages. One of the key factors in achieving these gains is to calculate reserves using recurrence relations that scale linearly with the number of time steps. Here, we present a general vector recurrence relation which can be used for a wide range of non-unit linked policies that are covered by Solvency II; such contracts include annuities, term assurances, and endowments. Our results suggest that by using an optimised parallel implementation of this algorithm, on an affordable hardware platform, it is possible to perform the ‘brute force’ approach to demonstrating solvency in a realistic timescale (of the order of a few hours).

This paper presents an intuitively simple asset pricing model designed to predict stock returns and volatilities, when stock prices may follow a fractal walk rather than a random walk. The model utilizes similarity ratio of the return fractals as the basis for forecasting. We argue that a collection of past returns such as moving average statistics can be “expanded” to generate future returns through the similarity ratios by trading time multiples. We also argue that stock returns with fractal dimensions in excess of 1.5, the norm for market efficiency, may be prone to frequent jumps and discontinuity. The paper builds an econometrically testable model to estimate fractal dimensions to offer an alternative way to forecast return volatilities without having to estimate separately the bipower variation and the jumps in existing studies. We argue that fractional Brownian motion may be a more suitable description than the standard Wiener process when describing stock return behaviors. The paper also demonstrates the application aspect of our asset pricing model to high frequency algorithmic trading.

Not long ago securities were traded by human traders in face-to-face markets. The ecosystem of an open outcry market was well-known, visible to a human eye, and rigidly prescribed. Now trading is increasingly done in anonymous electronic markets where traders do not have designated functions or mandatory roles. In fact, the traders themselves have been replaced by algorithms (machines) operating with little or no human oversight. While the process of electronic trading is not visible to a human eye, machine-learning methods have been developed to recognize persistent patterns in the data. In this study, we develop a dynamic machine-learning method that designates traders in an anonymous electronic market into five persistent categories: high frequency traders, market makers, opportunistic traders, fundamental traders, and small traders. Our method extends a plaid clustering technique with a smoothing framework that filters out transient patterns. The method is fast, robust, and suitable for a discovering trading ecosystems in a large number of electronic markets.