Systematic equity models are often built by combining different signals, such as value, momentum, and quality, into a single score. In most cases, this is constructed in a linear way, where each signal contributes independently and proportionally to the final result. While this approach is simple and effective, it assumes that each signal works the same way regardless of the broader context. In reality, this is not always the case.
As part of the statistical learning research (Alentorn & Mrazik, 2025) within our academic programme, we have been exploring how non-linear interactions between investment signals can be captured more effectively using, for example, decision trees and neural networks as well as more traditional techniques. In this instance, we have opted for a model that captures non-linearities while still being interpretable, and where we can validate the investment insight being implemented.
Non-linear interactions, or state dependence, allow the model to capture situations where the effectiveness of one signal depends on another. Instead of treating signals in isolation, the model can recognise that certain combinations of characteristics are particularly important.
The benefit of this approach is that it allows the model to focus on the parts of the market where signals are most meaningful. Rather than spreading predictive power evenly across all stocks, non-linear approaches help identify specific conditions where returns are more predictable. This can deliver a more targeted and informative signal, better aligned with how markets actually behave. In essence, non-linear relationships provide a way to extract additional insight from existing signals, without needing entirely new data, by understanding how those signals work together rather than separately.
Non-linear interactions can capture patterns that evade linear models.
Source: Jupiter. For illustrative purposes only.
Combining behavioural and informational effects
Our new non-linear signal falls within our Company Management stock selection strategy, which seeks to assess the quality of a management team through the data that comes out of their decision-making. The new signal blends existing signals into an interaction term designed to enhance the information already present in each component. Specifically, it combines the informational effect of management signalling with the impact of price behaviour.
The key insight is that the interdependencies between these two effects can be more informative than either signal on its own. It can capture meaningful divergence or convergence between the external perceptions of market participants and the internal conviction of management. Divergent cases may be more likely to reveal mispricing, for example.
By explicitly modelling subtle structural relationships, our new signal is able to identify cases where divergence or convergence are most meaningful. The model can better capture situations where market expectations and underlying fundamentals may be misaligned. This leads to a richer and more economically intuitive signal. For example, weak price behaviour combined with poor management signalling (convergence) might reinforce expectations of a negative outcome; whereas weak price behaviour combined with strong management signalling (divergence) might in some cases point to a mispricing and potentially a positive outcome.
Benefits of the new enhancement
The addition of the new management signalling interaction within the model’s Company Management stock selection strategy represents a meaningful extension. By capturing conditional relationships and economically intuitive functional dependencies between signals, it provides access to incremental alpha that is not fully captured by linear frameworks. When implemented carefully, this can:
- enhance model diversification by incorporating a new uncorrelated signal that operates on a different dimension from existing signals
- improve risk management, especially in the handling of price behaviour information which can experience periods of sharp reversals, particularly in stressed market environments
- strengthen the model’s ability to source alpha by producing a refined and more robust stock selection process.
Continually enhancing the systematic process
Central to the Jupiter Systematic philosophy is a continuous and disciplined research effort to ensure that our investment process improves over time. For over 20 years, this philosophy has resulted in a regular stream of evolutionary changes to our investment process, leading to improvements in our expected risk-adjusted returns over time. We currently have several other exciting research projects under way, including the development of further non-linear signals, and we look forward to bringing you more details of these in due course.
The team’s investment process has been implemented for more than two decades and is continuously refined. Recent enhancements are shown below. These enhancements follow extensive research undertaken by the team.
Recent enhancements
Introduced to each of the three parts of the investment process
References
James G, Witten D, Hastie T, Tibshirani R & Taylor J. An Introduction to Statistical Learning, 2023. Available at https://www.statlearning.com/
Alentorn A & Mrazik M. Fooled by noise? Why statistical learning, not hype, drives our process, Jupiter, 2025. Available at https://www.jupiteram.com/uk/en/professional/insights/why-statistical-learning-not-hype-drives-our-process/
The value of active minds: independent thinking
A key feature of Jupiter’s investment approach is that we eschew the adoption of a house view, instead preferring to allow our specialist fund managers to formulate their own opinions on their asset class. As a result, it should be noted that any views expressed – including on matters relating to environmental, social and governance considerations – are those of the author(s), and may differ from views held by other Jupiter investment professionals.
Important information
Marketing communication.
This document is intended for investment professionals* and is not for the use or benefit of other persons, including retail investors.
The value of investments and income may go down as well as up and investors may not get back amounts originally invested. Exchange rate changes may cause the value of investments to fall as well as rise.
This document is information only and is not investment advice.
The views expressed are those of the author(s) at the time of preparation, are not necessarily those of Jupiter as a whole and may be subject to change.
Every effort is made to ensure the accuracy of any information provided but no assurances or warranties are given.
Past performance does not predict future returns.
Forecasts are an estimate of future performance based on evidence from the past on how the value of this investment varies, and/or current market conditions. Forecasts are not a reliable prediction of future returns. What you will get will vary depending on how the market performs and how long you keep the investment/product.
Historical examples are for illustrative purposes only, may not represent current fund holdings, and are not a recommendation to buy or sell.
Where a benchmark is used for comparison, it is shown for illustrative purposes only and does not imply future performance.
No part of this document may be reproduced in any manner without the prior permission of Jupiter.
Issued in the UK by Jupiter Asset Management Limited (JAM), registered address: The Zig Zag Building, 70 Victoria Street, London, SW1E 6SQ is authorised and regulated by the Financial Conduct Authority. Issued in the EU by Jupiter Asset Management International S.A. (JAMI), registered address: 5, Rue Heienhaff, Senningerberg L-1736, Luxembourg which is authorised and regulated by the Commission de Surveillance du Secteur Financier. No part of this document may be reproduced in any manner without the prior permission of JAM/JAMI.
In Singapore, investment professionals means accredited and/or institutional investors as defined under section 4A of the securities and futures act (“SFA”) and in Hong Kong professional investors as defined under the Securities and Futures Ordinance (Cap. 571 of the Laws of Hong Kong).