Portfolio Selection Algorithm Research Articles

Integrating Weak Aggregating Algorithm and Reinforcement Learning for Online Portfolio Selection: The WARL Strategy

Integrating Weak Aggregating Algorithm and Reinforcement Learning for Online Portfolio Selection: The WARL Strategy

Novel online portfolio selection algorithm using deep sequence features and reversal information

Computational finance combines machine learning with financial needs to provide more efficient solutions for investment analysis and automated trading. In previous studies, traditional online portfolio selection (OLPS) algorithms were found to be overly reliant on artificially designed, subjective financial features. To address this issue, we propose a new predictive price tracking algorithm based on deep sequence features and reversal information (DSF-RI-PPT) for OLPS, extending a hybrid stock prediction algorithm to a multi-asset trading algorithm. We respectively employ the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN), principal component analysis (PCA) algorithms and long short-term memory (LSTM) network to perform decomposition, feature extraction and prediction on financial data. Further, we supplement the reversal information by modifying the predicted prices with a reversal indicator-rate of change (ROC). Finally, we introduce a fast error back-propagation algorithm to integrate the predictive information into the investment ratio using gradient projection. Through empirical comparison and statistic analysis of the DSF-RI-PPT algorithm, price-tracking algorithms with similar prediction models, and nine classic OLPS algorithms in nine portfolio data sets under three financial indexes, it can be found that the DSF-RI-PPT algorithm is profitable and generalizable.

RPS: Portfolio asset selection using graph based representation learning

Portfolio optimization is one of the essential fields of focus in finance. There has been an increasing demand for novel computational methods in this area to compute portfolios with better returns and lower risks in recent years. We present a novel computational method called Representation Portfolio Selection by redefining the distance matrix of financial assets using Representation Learning and Clustering algorithms for portfolio selection to increase diversification. RPS proposes a heuristic for getting closer to the optimal subset of assets. Using empirical results in this paper, we demonstrate that widely used portfolio optimization algorithms, such as Mean-Variance Optimization, Critical Line Algorithm, and Hierarchical Risk Parity can benefit from our asset subset selection.

A hybrid framework for mean-CVaR portfolio selection under jump-diffusion processes: Combining cross-entropy method with beluga whale optimization

<abstract><p>In this paper, a new hybrid meta-heuristic algorithm called CEBWO (cross-entropy method and beluga whale optimization) is presented to solve the mean-CVaR portfolio optimization problem based on jump-diffusion processes. The proposed CEBWO algorithm combines the advantages of the cross-entropy method and beluga whale optimization algorithm with the help of co-evolution technology to enhance the performance of portfolio selection. The method is evaluated on 29 unconstrained benchmark functions from CEC 2017, where its performance is compared against several state-of-the-art algorithms. The results demonstrate the superiority of the hybrid method in terms of solution quality and convergence speed. Finally, Monte Carlo simulation is employed to generate scenario paths based on the jump-diffusion model. Empirical results further confirm the effectiveness of the hybrid meta-heuristic algorithm for mean-CVaR portfolio selection, highlighting its potential for real-world applications.</p></abstract>

A novel approach to portfolio selection using news volume and sentiment

AbstractIn this study, we develop a novel approach to portfolio diversification by integrating information on news volume and sentiment with the k‐nearest neighbors (kNN) algorithm. Our empirical analysis indicates that high news volume contributes to portfolio risk, whereas news sentiment contributes to portfolio return. Based on these findings, we propose a kNN algorithm for portfolio selection. Our in‐sample and out‐of‐sample tests suggest that the proposed kNN portfolio selection approach outperforms the benchmark index portfolio. Overall, we show that incorporating news volume and sentiment into portfolio selection can enhance portfolio performance by improving returns and reducing risk.

Weak aggregating specialist algorithm for online portfolio selection

Weak aggregating specialist algorithm for online portfolio selection

A novel adjusted learning algorithm for online portfolio selection using peak price tracking approach

Online Portfolio Selection (OLPS) has attracted extensive interest in recent years. Accurate prediction of future prices and determining the optimal portfolio selection strategy based on the estimated return is a challenging topic in machine learning. We propose a novel adjusted learning algorithm based on peak price tracking for OLPS to tackle this challenge. The algorithm is based on an aggressive strategy with residual information and transaction costs. We first propose an adjusted online portfolio selection algorithm using Peak Price Tracking Approach (PPTA) to improve the accuracy of return prediction by introducing λ to adjust the impact of peak price and residual term on the predicted price. We then build the Net Profit Maximization (NPM) model with transaction costs. Finally, we integrate the PPTA and the NPM algorithms into a new algorithm called the PPTA-NPM to maximize the cumulative return. Extensive benchmark data experiment results and statistical analysis show that the PPTA algorithm significantly improves the accuracy in predicting the future price, and the integrated PPTA-NPM algorithm is superior to multiple classic OLPS algorithms.

Online portfolio selection with state-dependent price estimators and transaction costs

Artificial intelligence (A.I.) techniques have been applied to the online portfolio selection (OLPS) problem, a topic attracting increasing attention. In brief, OLPS is the task of sequentially updating the investment portfolio with the continuous update of assets’ prices. In this paper, we study the OLPS problem with transaction costs. First, we study the exact computation of the transaction cost and derive related constant upper and lower bounds, which allow us to take the transaction costs into account when deriving an optimal portfolio in each investment period. Second, considering that assets’ market states switch from time to time and their prices exhibit different behaviors in different market states, we propose the state-dependent exponential moving average method (SEMA), which can accurately predict assets’ returns based on historical return data and assets’ market states. Third, we construct the net profit maximization model (NPM) and the net profit maximization model with a risk parity constraint (NPMRP). Finally, we combine these three parts to build the state-dependent online portfolio selection algorithm (SOPS) for solving the OLPS problem with transaction cost. Our empirical results reveal that the proposed SOPS algorithm can outperform many state-of-the-art OLPS algorithms.

A New Boosting Algorithm for Online Portfolio Selection Based on dynamic Time Warping and Anti-correlation

A New Boosting Algorithm for Online Portfolio Selection Based on dynamic Time Warping and Anti-correlation

Empirical Analyses of OLMAR Method for Financial Portfolio Selection in Stock Markets

The OLMAR method, which stands for the on-line moving average reversion method, is reported to be one of the most powerful among portfolio selection algorithms in the stock markets. In this research, we use intensive statistical and simulation analyses of long-term data on stock market changes to uncover the secrets of why and when the superiority appears. We find that there have been long-lasting fluctuations in the stock markets and that the OLMAR method actively makes use of such characteristics. In this paper, we analyze long-term stock data from Japan and the United States. The analyses confirm the following points. 1) The OLMAR method yields superior returns. 2) By using the moving average divergence rate provided by the OLMAR method, it is possible to detect specific fluctuation characteristics in long-term stock data from Japan and the United States. 3) Superior returns cannot be obtained from data in which specific fluctuation characteristics have been corrected.

Asset splitting algorithm for ultrahigh dimensional portfolio selection and its theoretical property

The presence of a huge number of assets poses challenges to classical portfolio selection algorithms. Constrained l1 minimization approaches have been proposed to directly estimate effective parameters in the optimal portfolio. Linear programming method and alternating direction method of multiplier (ADMM) algorithm is used to solve the corresponding minimization problems. However, these two algorithms may fail due to the limitations of computing time and computing memory when a huge number of assets are considered in the portfolio optimization. This article proposes an asset splitting ADMM (AS-ADMM for short), a parallel computing algorithm, to tackle such challenges, and establishes the convergence property of the new algorithm. Furthermore, we develop a new regularization method for estimating the effective parameters with the folded-concave penalty and establish its oracle property. The local linear approximation (LLA) algorithm is used to redirect the new method to a weighted l1 regularization method. We conduct simulation studies to investigate the advantage of the proposed algorithm and regularized model in solving the high dimensional portfolio selection problems. A real data example is also included to demonstrate the applicability of the proposed algorithms and regularization methods.

Robotizing bond portfolio selection on the Russian debt market on the basis of a modified strategy of riding the yield curve

The modern global debt market features historically low average interest rates, convergence of yields on bonds with different maturities, an increase of yield curve inversion emergence frequency and a large-scale trend to automate financial decision making. The researchers’ attention in these fields is mainly focused on designing models that describe the state of the debt market as whole or its individual instruments in particular, as well as on risk management methods. At the same time, the specialized literature offers very few works concerning the topic of computer algorithms for bond portfolio selection based on traditional or advanced investment strategies. The aim of the present research is to create a modification of the existing algorithm of riding the yield curve strategy application, employing, first, average bond yield over the holding period instead of traditional bond yield to maturity; second, a developed algorithm for calculating the market spread on bonds; and, third, alternative risk evaluation indicators (compensation coefficients), which allow us to measure objectively price risk, liquidity risk, transaction costs risk and a general risk. The modification and the development of the algorithm for calculating the market spread were carried out using the direct measurement of the result technique, which entails application of the strategy to the data on bond issues received through the Moscow Exchange API. The selection of financial instruments was conducted in all sectors of the Russian debt market: public bonds, sub-federal and municipal bonds, corporate bonds. The modified algorithm enabled us to get extra yield for each selected bond issue, thereby proving the high effectiveness of the technique compared to the traditional strategy. Software implementation of the algorithm can be integrated into any robotized or semi-robotized stock exchange trading application.

UNIVERSAL RISK BUDGETING

I juxtapose Cover’s vaunted universal portfolio selection algorithm ([Cover, TM (1991). Universal portfolios. Mathematical Finance, 1, 1–29]) with the modern representation of a portfolio as a certain allocation of risk among the available assets, rather than a mere allocation of capital. Thus, I define a Universal Risk Budgeting scheme that weights each risk budget, instead of each capital budget, by its historical performance record, á la Cover. I prove that my scheme is mathematically equivalent to a novel type of [Cover, TM and E Ordentlich (1996). Universal portfolios with side information. IEEE Transactions on Information Theory, 42, 348–363] universal portfolio that uses a new family of prior densities that have hitherto not appeared in the literature on universal portfolio theory. I argue that my universal risk budget, so-defined, is a potentially more perspicuous and flexible type of universal portfolio; it allows the algorithmic trader to incorporate, with advantage, his prior knowledge or beliefs about the particular covariance structure of instantaneous asset returns. Say, if there is some dispersion in the volatilities of the available assets, then the uniform or Dirichlet priors that are standard in the literature will generate a dangerously lopsided prior distribution over the possible risk budgets. In the author’s opinion, the proposed “Garivaltis prior” makes for a nice improvement on Cover’s timeless expert system, that is properly agnostic and open to different risk budgets from the very get-go. Inspired by [Jamshidian, F (1992). Asymptotically optimal portfolios. Mathematical Finance, 2, 131–150], the universal risk budget is formulated as a new kind of exotic option in the continuous time Black–Scholes market, with all the pleasure, elegance, and convenience that entails.

Portfolio selection in non-stationary markets

We consider the task of portfolio selection as a time series prediction problem. At each time-step we obtain prices of a universe of assets and are required to allocate our wealth across them with the goal of maximizing it, based on the historic price returns. We assume these returns are realizations of a general non-stationary stochastic process, and only assume they do not change significantly over short time scales. We follow a statistical learning approach, in which we bound the generalization error of a non-stationary stochastic process, using analogues of uniform laws of large numbers for non-i.i.d. random variables. We use the learning bounds to formulate an optimization algorithm for portfolio selection, and present favorable numerical results with financial data.

A hybrid algorithm for portfolio selection: An application on the Dow Jones Index (DJI)

With the developing technology, investors have started to direct their investments by using computer software. The fact that these software do not take investor’s views into account causes all investors to achieve the same expected return at the same risk level. In this study, in order to find a solution to this problem, a two-stage portfolio selection model, which takes investment data and expert opinions into consideration, is proposed. In the first stage, the weight of the criteria in the portfolio selection problem was determined by the Constrained Fuzzy Analytic Hierarchy Process method. In the second stage, the fuzzy linear programming (FLP) problem created using the weights of the specified criteria is solved by fuzzy logic approach. These two methods in the literature use triangular fuzzy numbers (TFNs) in the solution process of the problem. In this study, methods are developed by creating mathematical models that makes the triangular fuzzy number-oriented theoretical infrastructures of the two methods used suitable for the use of trapezoidal fuzzy numbers (TrFNs). As a result, a new hybrid portfolio selection algorithm is proposed in which the methods developed for TrFNs in the first and second stage are used together. In the application section, portfolio distribution was made with the proposed hybrid model based on TrFNs by using the return rates of 30 stocks included in the Dow Jones Index, financial ratios of the stocks and decision makers’ opinions between January 1, 2018 and December 31, 2020. In the conclusion section of the study, as a result of the portfolio distributions obtained from each method, the total return amounts obtained when the investment was made in January 2021 according to the rate of return of the stocks were compared.

Adaptive online portfolio selection with transaction costs

As an application of machine learning techniques in financial fields, online portfolio selection has been attracting great attention from practitioners and researchers, which makes timely sequential decision making available when market information is constantly updated. For online portfolio selection, transaction costs incurred by changes of investment proportions on risky assets have a significant impact on the investment strategy and the return in long-term investment horizon. However, in many online portfolio selection studies, transaction costs are usually neglected in the decision making process. In this paper, we consider an adaptive online portfolio selection problem with transaction costs. We first propose an adaptive online moving average method (AOLMA) to predict the future returns of risky assets by incorporating an adaptive decaying factor into the moving average method, which improves the accuracy of return prediction. The net profit maximization model (NPM) is then constructed where transaction costs are considered in each decision making process. The adaptive online net profit maximization algorithm (AOLNPM) is designed to maximize the cumulative return by integrating AOLMA and NPM together. Numerical experiments show that AOLNPM dominates several state-of-the-art online portfolio selection algorithms in terms of various performance metrics, i.e., cumulative return, mean excess return, Sharpe ratio, Information ratio and Calmar ratio.

Assistant reference point guided evolutionary algorithm for many-objective fuzzy portfolio selection

Portfolio selection is dedicated to assigning limited capital to available financial assets to achieve an optimal trade-off between profit and risk. This work is targeted at a portfolio selection problem simultaneously considering five optimization objectives namely mean, variance, skewness, kurtosis, and entropy of return. An expert’s acknowledge-based fuzzy number variable, rather than a random variable, is used to model the return of a risk asset, such that a more accurate estimation of the return of a risky asset is achieved. An assistant reference point guided evolutionary algorithm (ARPGEA) is proposed to solve this five-objective portfolio problem. Assistant reference points are generated for the non-dominated solutions in all dimensions of the objective space. The generated reference points and a comprehensive fitness evaluation method are used to guide the search towards promising regions leading to well-distributed and well-converged solutions. ARPGEA is tested on publicly available data sets collected from six market indices including up to 1,203 assets. Experiment results demonstrate the efficiency and effectiveness of ARPGEA in comparison with other five state-of-the-art many-objective evolutionary algorithms.

The Importance of Measuring Liquidity Risk with Smart Financial Applications

Over the past few days, alarm bells have been ringing for the risk of recession in the world’s leading economies (Germany, United Kingdom, Italy, Brazil and Mexico). Deceleration is affecting several regions in the world and might even become more widespread, exacerbating investor mistrust and financial market instability produced by the U.S.-China trade war, among other uncertainty factors. Investors have transferred part of their investment portfolios to safe haven assets, such as sovereign bonds, with investor mistrust at record lows in recent years. Managing your investment portfolio when an economic cycle is drawing to an end can be tricky. Even though some experts recommend rebalancing asset allocation, buying treasury bonds, focusing on raw materials, commodities or real estate investment, the most important thing is to learn how to manage risk. In the current context of uncertainty and adverse market conditions, assessing investment portfolio performance optimization, as well as liquidity risk, is crucial, as explained in the chapter “Theoretical and practical foundations of liquidity-adjusted value-at-risk (LVaR): optimization algorithms for portfolio selection and management” of the recently published book Expert Systems in Finance. Smart Financial Applications in Big Data Environments (Routledge, Taylor & Francis Group, 2019).

Innovative deep matching algorithm for stock portfolio selection using deep stock profiles.

Construction of a reliable stock portfolio remains an open issue in quantitative investment. Multiple machine learning models have been trained for stock portfolio selection, but their practical applicability remains limited due to the challenges posed by the characteristic of a low signal-to-noise ratio (SNR), the nature of time-series data, and non-independent identical distribution in financial data. Here, we transformed the stock selection task into a matching problem between a group of stocks and a stock selection target. We proposed a novel representation algorithm of stock selection target and a novel deep matching algorithm (TS-Deep-LtM). Then we proposed a deep stock profiling method to extract the optimal feature combination and trained a deep matching model based on TS-Deep-LtM algorithm for stock portfolio selection. Especially, TS-Deep-LtM algorithm was obtained by setting statistical indicators to filter and integrate three deep text matching algorithms. This parallel framework design made it good at capturing signals from time-series data and adapting to non-independent identically distributed data. Finally, we applied the proposed model to stock selection and tested long-only portfolio strategies from 2010 to 2017. We demonstrated that the risk-adjusted returns obtained by our portfolio strategies outperformed those obtained by the CSI300 index and learning-to-rank approaches during the same period.

A framework for online investment decisions

ABSTRACT The artificial segmentation of the investment management process into silos of human operators can restrict silos from collectively and adaptively pursuing a unified investment goal. In this article, we argue that the investment process can be accelerated and be made more cohesive by replacing batch processing for component tasks of the investment process with online processing. We propose an integrated and online framework for investment workflows, where components produce outputs that are automatically and sequentially updated as new data arrives. The workflow can be further enhanced to refine signal generation and asset class evolution and definitions. Our results demonstrate that we use this framework in conjunction with resampling methods to optimise component decisions with direct reference to investment objectives while making clear the extent of backtest overfitting. We consider such an online update framework to be a crucial step towards developing intelligent portfolio selection algorithms that integrate financial theory, investor views, and data analysis with process-level learning.