Using judgement with automation to avoid another failure in Portfolio Risk Management

A decade ago, The Times of London published my article “How to Prevent Another Failure in Risk Management”. I stressed the need for a multi-pronged approach towards risk management, comprising a combination of qualitative and quantitative techniques, for regulators to detect early signs of an imminent crisis.

Since then it has generally been accepted that that one of the main factors precipitating the Credit Crunch was an over-reliance on risk models such as VAR (Value at Risk) by Central Banks, Regulators and the Rating Agencies. This methodology assumed that asset prices were normally distributed and grossly underestimated the chance of extreme events. It also assumed that the correlations between asset prices assumed stability based on historical data.

A decade or so later, we are in a much better position now to implement the preventative approach that I outlined in my article of 2009, due to advances in technology and availability of new types of diversified funds.

However, with fears of another potential financial crisis looming, it is surprising to find that very few portfolio managers are applying the rigorous, albeit very intuitive, approach needed to reduce and manage the risks that could severely impact their investors’ portfolios.  In the ongoing Covid-19-inspired market correction, we have seen many so-called “Balanced” and even “Cautious” funds returning losses in excess of -20%.  The risk of fear and panic spilling over to the credit markets should not be overlooked.

The Fed’s “super bazooka” has calmed market tensions for now but it would be naïve to dismiss a potential spill-over as we saw US unemployment spike last week alongside those of other nations.  The western populous is deeply indebted with added stress being placed upon personal loan and mortgage repayments.  Loan delinquencies, as they are known, could easily create a banking crisis unless indicators are scrutinised closely, and remedial actions taken where necessary.

There is much talk in the press about using technology where it can surpass human capabilities, and about harnessing it at every step by human judgement to check assumptions and produce information to facilitate decision making. However, as far as managing multi asset portfolios in complex financial markets is concerned that is still very far away, despite the claims of many algorithmic trading systems.

The drawback with heuristic algorithms is that they are trained on past financial market conditions and are therefore unable to cope with completely new paradigms. Essentially, they fail due to model error and changes in regime that the model cannot capture due to the constraints imposed by model specification.

The alternative, purely qualitative / fundamental portfolio management, has difficulty synthesising the myriad of changing complex financial relationships which can provide valuable clues about asset price movements. This renders it unable to forecast market events, and misprices downside risk and the negative impact of the ‘herding’ mentality and investor overconfidence.

Clearly, a smart combination of the two approaches is needed – one that deploys quantitative techniques for complex technical and mathematical analysis, the results of which can be qualitatively analysed by experienced portfolio managers.

Such an approach is second nature in other professions, such as medicine, where stethoscopes, X-rays, scanning devices and chemical tests are all part of the automated tool kit, used with judgement, to evaluate risks and reach diagnoses.

Below is an example of what this approach can consist of drawn from the process employed by Alpha Beta Partners.

The first step consists of using automated risk questionnaires together with client interviews and determination of risk categories.

Next, the impact of market factors – macro, fundamental, technical and geopolitical – on all the permitted asset classes is measured using quantitative models which generate signals with which the portfolio manager can formulate a few high conviction views and confidence levels. The results of the quantitative models used can provide further insights into market risks.

These selected views can then be analysed using an automated implied views calculator to generate views on markets on which outlooks have not yet been formulated. These throw light on the inter-relationships being assumed between asset classes and can lead to revisions before being finalised.

The complete set of views is next used by an automated optimisation model to generate portfolios for each risk category.  Critical analyses of these is followed by scenario analysis where each portfolio is stress tested in worse case scenarios, such as the credit crunch scenarios, or the dot-com crash. This enables the manager to use judgement to finalise decisions after understanding risk from yet another perspective.

Finally, the portfolios selected are monitored on an automated real time basis and the models that generated them are constantly reviewed as are deviations beyond limits set, of the performances and risks of the portfolios. If these are not due simply to transient market anomalies, the models can be corrected, and the portfolios rebalanced where appropriate. This last process can often highlight changes in market paradigms.

Due to the availability of low-cost automated systems, and low-cost diversified funds for asset allocation (ETFs and index funds), the above intuitive and transparent steps can be used to generate high performance, low cost and risk-controlled portfolios – but only if used by experienced portfolio managers.

Just as medical surgeons use their experience to harness the powers of medical technology, so highly skilled portfolio managers need to use judgements based on many years of experience and training to provide high quality portfolios for investors and reduce the chances of exposure to another failure in risk management.

Read more articles on our Insights page

Shahid Chaudhri, Investment Director

Shahid Chaudhri is a highly qualified and accomplished Investment Director, with broad experience that includes working at a range of top tier banks (investment bank, central bank and private bank), fund management companies, hedge funds and regulators across the world.

His professional track record includes collaborating with Fischer Black and Bob Litterman on the Black-Litterman Asset Allocation and Risk Management Model, Pioneering Asset/Liability Management at Salomon Brothers, and managing $16bn Central Bank reserves.

Shahid’s management experience includes Executive Director at Goldman Sachs, Director at Salomon Brothers, Head of Client Solutions at RBS/Coutts and Chief Investment Officer at Capita Financial Group.

As a result, he is well versed in qualitative and quantitative fund management techniques, and aware of the regulatory issues that impact investment funds.

The information, materials or opinions contained on this website are for general information purposes only and are not intended to constitute legal or other professional advice and should not be relied on treated as a substitute for specific advice of any kind.

We make no warranties, representatives or undertakings about any of the content of this website (including without limitation any representations as to the quality, accuracy, completeness or fitness of any particular purpose of such content, or in relation to any content of articles provided by third parties and displayed on this website or any website referred to or accessed by hyperlinks through this website.

Although we make reasonable efforts to update the information on this site, we make no representation warranties or guarantees whether express or implied that the content on our site is accurate complete or up to date.