Optimising predictive patterns for portfolios

investment viewpoints

Optimising predictive patterns for portfolios

 

As part of its on-going, quantitative research initiative, Lombard Odier Investment Managers invited Professor Semyon Malamud for an interactive session at the firm to meet our quantitative researchers from New York, London, Geneva and Hong Kong.
 

Professor Malamud is an Associate Professor at EPFL (École Polytechnique Fédérale de Lausanne), and a research fellow at the Centre for Economic Policy Research. He earned a PhD in mathematics from the Swiss Federal Institute of Technology, Zurich.

Professor Malamud presented a mathematical model that is designed to optimise portfolio construction. Following the presentation, the visiting academic discussed recent developments in the fields of trend-following and momentum research with the LOIM team.

Professor Malamud’s model aims to construct a generic portfolio with low risk and high returns, and find predictive relationships between signals and returns. The signals are built from historical data and can include momentum, carry, volatility or fundamental strategies. The returns belong to multiple asset classes such as bonds, currencies, commodities or stocks.

The model aims to construct a generic portfolio with low risk and high returns, and find predictive relationships between signals and returns.

In essence, the model attempts to optimise predictive relationships between models’ signals and future returns. Among other features, the model takes into account the rebalancing frequency and can have any number of different signals and returns streams. It can incorporate specific constraints, transaction costs, etc.

The ultimate aim is to build a portfolio with the following characteristics:

  • High expected returns.
  • High predictability – this involves finding signals containing significant and robust information about returns.
  • Low risk.
  • Robust (or free of so-called noise) – this means that the predictive relationship that is identified in the data must not be spurious and must be stable over time.
  • Looks for the ground state of the predictive relationship.

“I propose a mathematical methodology where return is predicted by a number of signals, not just risk. My method takes into account these five criteria (as above) and the general formula aims to detect multi-variate predictive patterns in data,” said Malamud. “It’s a natural way to identify weak vs strong signals, and it can be applicable to any data – stocks, futures, commodities, currencies or derivatives.”

“The aim is to build a portfolio that is the most resilient to the breakdown of the predictive relationships between signals. We want to find the ones that are the most robust.”

 

The spectral method to reduce noise

Malamud then moves beyond linear geometry, to Perron-Frobenius theory and historical physical probabilities, or what is deemed a spectral method in order to reduce noise. Perron-Frobenius imagines a discretized state space model – a state of the system follows a (high dimensional) Markov chain.

The objective is to find the non-linear ground state of the system for the largest Eigenvalues.1 It’s possible to construct a transition probability matrix that encodes information about the joint behaviour of signals and returns. But this matrix contains a lot of noise and Perron-Frobenius can be applied to extract the ground state.

Lastly, Malamud incorporates some of the theories developed by the US economist, Stephen Ross, in his 2011 paper entitled “The Recovery Theorem.2” The theorem takes multiple option prices and calculates, under certain technical conditions, the market consensus for the physical probabilities. Malamud applies this technique using historical underlying price data (instead of options) in order to find physical probabilities and thereby build optimal portfolios.

“I believe this Perron-Frobenius theory, this ability to identify predictive, but non-linear, predictive relationships from transition probabilities and prices – both physical and risk-neutral - is a very powerful tool that can be used to construct portfolios.” 

Finally, Malamud summarised his findings: “I use spectral analysis to find the ‘market ground state’: the most robust portfolio in terms of return predictability and risk. More generally, my approach allows me to decompose returns into a basis of ‘optimally predictable’ portfolios that is better suited for portfolio analysis than the more standard PCA (price component analysis)”.

 

Please find key terms in the glossary.

 

Sources.

1 Eigenvalues and Eigenvectors are at the core of PCA analysis and are used to reduce noise in data.

2 The Recovery Theorem by Stephen A. Ross. NBER Working Paper No. 17323, August 2011. Please see: https://www.nber.org/papers/w17323

 

important information.

For professional investor use only

This document has been issued by Lombard Odier Funds (Europe) S.A. a Luxembourg based public limited company (SA), having its registered office at 291, route d’Arlon, 1150 Luxembourg, authorised and regulated by the CSSF as a Management Company within the meaning of EU Directive 2009/65/EC, as amended; and within the meaning of the EU Directive 2011/61/EU on Alternative Investment Fund Managers (AIFMD). The purpose of the Management Company is the creation, promotion, administration, management and the marketing of Luxembourg and foreign UCITS, alternative investment funds ("AIFs") and other regulated funds, collective investment vehicles or other investment vehicles, as well as the offering of portfolio management and investment advisory services.

Lombard Odier Investment Managers (“LOIM”) is a trade name.

This document is provided for information purposes only and does not constitute an offer or a recommendation to purchase or sell any security or service. It is not intended for distribution, publication, or use in any jurisdiction where such distribution, publication, or use would be unlawful. This material does not contain personalized recommendations or advice and is not intended to substitute any professional advice on investment in financial products. Before entering into any transaction, an investor should consider carefully the suitability of a transaction to his/her particular circumstances and, where necessary, obtain independent professional advice in respect of risks, as well as any legal, regulatory, credit, tax, and accounting consequences. This document is the property of LOIM and is addressed to its recipient exclusively for their personal use. It may not be reproduced (in whole or in part), transmitted, modified, or used for any other purpose without the prior written permission of LOIM. This material contains the opinions of LOIM, as at the date of issue.

Neither this document  nor any copy thereof may be sent, taken into, or distributed in the United States of America, any of its territories or possessions or areas subject to its jurisdiction, or to or for the benefit of a United States Person. For this purpose, the term "United States Person" shall mean any citizen, national or resident of the United States of America, partnership organized or existing in any state, territory or possession of the United States of America, a corporation organized under the laws of the United States or of any state, territory or possession thereof, or any estate or trust that is subject to United States Federal income tax regardless of the source of its income.

Source of the figures: Unless otherwise stated, figures are prepared by LOIM.

LOIM does not provide accounting, tax or legal advice.

Although certain information has been obtained from public sources believed to be reliable, without independent verification, we cannot guarantee its accuracy or the completeness of all information available from public sources.

Views and opinions expressed are for informational purposes only and do not constitute a recommendation by LOIM to buy, sell or hold any security. Views and opinions are current as of the date of this presentation and may be subject to change. They should not be construed as investment advice.

For more information on Lombard Odier’s data protection policy, please refer to www.lombardodier.com/privacy-policy

No part of this material may be (i) copied, photocopied or duplicated in any form, by any means, or (ii) distributed to any person that is not an employee, officer, director, or authorised agent of the recipient, without Lombard Odier Funds (Europe) S.A prior consent.

©2019 Lombard Odier IM. All rights reserved.