Signals and noise: blending expert judgement and quantitative insight for superior multi-asset outcomes


   Serge Pizem, Global Head of Multi-Asset

The complexity of today’s financial system means that investors have more information to consider than ever before. Serge Pizem explains how blending judgemental views with quantitative insights can lead to a better-constructed multi-asset portfolio.

In January 2018, the VIX index rested calmly around 10, a level it had maintained for much of the prior 12 months. It was not sending up any distress flares. Then, suddenly, this ‘fear gauge’ shot above 50 in early February as prices crashed across different asset classes.
The VIX is a useful tool for investors to see what is happening in the markets, but ultimately offers no insight as what to might happen next.
“The VIX has been distorted as central banks have created a low-volatility environment,” explains Serge Pizem, our Global Head of Multi-Asset.
“That’s why sometimes you see those spikes of volatility. No model could have fully anticipated the explosion of volatility we saw in February, which is why investors tend to underestimate such ‘long tail’ risk events. Therefore, we try to find other signals that truly show the risk in the market – things that are less skewed and less impacted by external factors than traditional macroeconomic or market indicators.”

-> Find out more about our multi-asset expertise:


Knowing where to look

At the moment, Pizem is able to draw on 150 such signals for his multi-asset portfolios – and is always looking for more. He incorporates these into his investment process, which combines deep data analysis with human insight. For Pizem and AXA IM, the best outcomes for clients in multi-asset investing are achieved by strengthening high-conviction judgements with robust quantitative inputs and analysis.

Those investment signals have been developed by the multi-asset portfolio management team, who continually aim to develop new, additional signals. This investment team is also supported by portfolio engineers within our ‘Quant Lab’ when it comes to risk analysis aspects and performance monitoring.

So what are these 150 signals? Individually, each is a specific factor that has an impact on asset prices; by monitoring each signal, Pizem and his team can anticipate the movement of different markets.

“In practice, the predictive power of some data sets doesn’t hold up when you back test them. We therefore back test each of our proprietary signals over the full cycle of the past 10 years. If you are disciplined in the way you back test your model and you avoid biases, you can generate models that statistically have a good hit rate in determining market movements. So for a signal to go in our database, it has to be back tested and proven statistically significant.”
Serge Pizem, Global Head of Multi-Asset

All of these 150 signal therefore give the portfolio management team an informational edge in their analysis of macro and market data and help them to make decisions and navigate complex financial markets for their investors.



Anticipating turbulence

Two signals epitomise the insights that can be brought to Pizem’s attention through this process.

  1. One is AXA IM’s proprietary turbulence index, which tracks the evolution of parameters based on volatility and correlation. Any spikes in this index can imply the market is entering a risk-off mood, which can lead the multi-asset team to lower their equity exposure.
  2. The other is a proprietary ‘surprise gap’ index. This macroeconomic indicator compares the one-year order forecasts of purchasing managers, based on survey data, with the actual size of those orders when they are placed. This can suggest shifting economic dynamics that can influence the portfolio management team’s asset allocation decisions.

These signals are truly forward-looking and shed light on multiple asset classes, which stands in contrast to more widely cited metrics like the VIX, which is merely a proxy for the stock market’s aggregate expectation of volatility as expressed through S&P 500 index options.

“Everybody knows about volatility", remarks Pizem."The turbulence index does something different; it tells us about the evolution of stress in financial markets and helps us to see when matters are becoming dangerous. When the turbulence index starts to increase, it means there is stress building up in the market.That stress is like a spring that is becoming tense: it can steadily take on more and more tension over time, but can deploy its energy very suddenly with little provocation, so you have to be careful. If the index goes up and we don’t have any clear explanation for that, we would logically reduce the proportion of risky assets in the portfolio to avoid large drawdowns. But tension can also go on building up for months until it reaches a critical level, so you have to be careful and you have to use it in combination with other signals.”



The mosaic effect

But that is not the end of the story: Pizem and his team also use artificial intelligence tools and machine-learning techniques to combine these signals and optimise their analysis. Deploying this massive computing power on the data identifies relationships and correlations that might not be noticed by the human eye alone.

“There is no magical crystal ball,” says Pizem. “It’s more like a mosaic. You put together little pieces of stone, but if you look too closely you can’t see anything. If you pull back a little bit, however, all of a sudden you see something. We’re trying to look at as many stones as possible, so that we can analyse the full mosaic and read the markets more clearly. That’s our goal.”

Human interpretation of all these results is nevertheless still essential. “I have 150 signals,” remarks Pizem. “But what combination of them has the most explanatory power? Machine learning can give you some answers you wouldn’t have thought of, but then it’s up to you as a portfolio manager to decide on whether to act on that information or not.”

Data mining may reveal that the Japanese equity market, for example, is positively correlated with the orange crop in Florida. “At the end of the day, would you trust that relationship to remain the case in future seasons?” Pizem asks. “Of course not. We have to be open enough to say that we didn’t think about that relationship and it may make sense, but we also have to understand that correlation and decide for ourselves whether to invest according to that relationship.”

“The signals and data are unemotional,” he continues. “They can help you take a step back and analyse coolly what’s happening in the market. But ultimately we always want to interrogate the validity of these signals and manage portfolios through human judgement, because we’re not ready to leave everything to the computer.”



Risk assessments

Never basing decisions on isolated signals is just one aspect of the multi-asset team’s commitment to risk management. Indeed, the team’s portfolio managers can count on the support of dedicated portfolio engineers within the Quant Lab to help them ensure that their convictions are both aligned with their investors’ long-term interests and are efficiently implemented at the portfolio construction stage. The portfolio managers can also ask these portfolio engineers to help them calibrate their trades, by entering them in a model portfolio with pre-defined stop-loss and profit taking levels. All this can help the team to guard against future unexpected events.

“We can ask the Quant Lab what the impact on the portfolio would be if the events of the 2008 crisis happened in exactly the same way again today,” relays Pizem. “What about 1994 and 1997? What is the downside risk to the portfolio? That’s where they help us. It is very valuable and time efficient for us, because with their technical skills they can answer our enquiries about risk in the portfolio extremely quickly.”

The portfolio engineers within the Quant Lab have their own proprietary software, developed and tailored to their own internal needs to track – among other things – past correlations between different asset classes and back test portfolios against previous crises and market environments. “The essential points are looking not only at how much risk we have in the portfolio overall but how that risk is divided between our strategies on the long-only equity side, our long/short strategy, our strategy on the fixed income side, and our strategy on the currency side,” Pizem says. “Are all those risks well diversified and uncorrelated? If we have 12 strategies in the portfolio, you don’t want 90% of the risk in just one of them.”


Ongoing learning

A final core component of this multi-asset process is the wealth of experience within the portfolio management team. Among Pizem’s colleagues, there are those with up to 30 years’ worth of experience of markets and experts recruited with deep expertise in quantitative investing.

These experts in their fields all contribute to the continuous development of the database of signals and its application within the multi-asset portfolios.

“We don’t want to be too confident in any single signal, because in a couple of years it may cease to be useful and we will have to find new ones“.
Serge Pizem, Global Head of Multi-Asset


“It is very important that we always question ourselves, ask whether certain signals only work in certain environments, and remain humble. Our process is therefore very motivating for us all, as we continually try and find new ways to understand and anticipate what is happening in the market. It’s a great challenge, but with the assistance of big data and our own experience and insights we are focused on delivering better multi-asset outcomes for our investors.”


Not for Retail distribution: This document is intended exclusively for Professional, Institutional, Qualified or Wholesale Clients / Investors only, as defined by applicable local laws and regulation. Circulation must be restricted accordingly.
This promotional communication does not constitute on the part of AXA Investment Managers a solicitation or investment, legal or tax advice. This material does not contain sufficient information to support an investment decision.
Past performance is not a guide to current or future performance, and any performance or return data displayed does not take into account commissions and costs incurred when issuing or redeeming units. References to league tables and awards are not an indicator of future performance or places in league tables or awards and should not be construed as an endorsement of any AXA IM company or their products or services. Please refer to the websites of the sponsors/issuers for information regarding the criteria on which the awards/ratings are based. The value of investments, and the income from them, can fall as well as rise and investors may not get back the amount originally invested. Exchange-rate fluctuations may also affect the value of their investment. Due to this and the initial charge that is usually made, an investment is not usually suitable as a short term holding.
Due to its simplification, this document is partial and opinions, estimates and forecasts herein are subjective and subject to change without notice. There is no guarantee forecasts made will come to pass. Data, figures, declarations, analysis, predictions and other information in this document is provided based on our state of knowledge at the time of creation of this document. Whilst every care is taken, no representation or warranty (including liability towards third parties), express or implied, is made as to the accuracy, reliability or completeness of the information contained herein. Reliance upon information in this material is at the sole discretion of the recipient. This material does not contain sufficient information to support an investment decision.
Issued in the UK by AXA Investment Managers UK Limited, which is authorised and regulated by the Financial Conduct Authority in the UK. Registered in England and Wales No: 01431068. Registered Office: 7 Newgate Street, London EC1A 7NX.
In other jurisdictions, this document is issued by AXA Investment Managers SA’s affiliates in those countries.