Modern portfolio theory (MPT) is dead, we just don’t realise it yet. Exactly four years on from a global crash that effectively demonstrated a portfolio of well diversified assets is not necessarily any protection against eye-watering losses, MPT's assumptions, zombie like, continue to lumber on.
Although academia has regarded the underlying assumptions of MPT as suspect for some years, it has yet to formulate a simple and widely accepted alternative.
Justified suspicion of quantitative, trend-driven and systematic trading systems (or, indeed, anything not easily explained to anyone untrained in degree-level mathematics) among managers and clients has further delayed the search for alternatives among discretionary managers.
‘All of the major [portfolio] software providers will have its assumptions underlying them,’ notes Robert Jukes, global strategist at Collins Stewart. ‘If it is off the shelf, it has MPT in it.
‘The single biggest problem is the assumption that volatility and correlation are constant. That seemingly innocuous assumption was what directly led to large losses [in 2008]. The problems have been obvious to economists for some time – but not how to fix them.’
Jukes, alongside colleague Ed Smith, has been attempting to plug this knowledge gap. Initially brought in by Collins Stewart to analyse why traditional investment models failed so dramatically in 2008, their investigation was rapidly transformed from an interrogation of what went wrong into an experiment in how to set it right.
Three years down the line and an undisclosed sum of money later, the risk management system they have designed is a primary input into the company’s private client portfolios, and in its uncut form, the engine behind the company’s Remap (Risk enhanced multi asset portfolios) service.
As of June, the company’s ‘conservative’ equivalent Remap model had made a rolling 12-month return of 6.0% versus the Apcims Conservative Index’s 3.7%, with maximum discrete drawdown of -1.2% and -4% respectively. The Remap volatility stood at 2.9% versus Apcim’s 5.9%.
‘A lot of the mathematical modelling has actually brought us back into the realm of common sense, which is comforting. We are not a hedge fund: we need to make sure the framework is transparent and that we can explain it to clients,' Jukes says.
‘The bit which is the hard part – and where we are patting ourselves on the back a little – is how we apply it to portfolio management. Using ex-ante [forward looking] measures allows us to take de-risking measures much earlier [than is possible using traditional MPT].’
The company is tight-lipped about how many advisers have signed up for the Remap portfolio service – charging 0.8% a year below £1 million and 0.55% above – since launch six months ago. But to a large degree, the process underpins the £9 billion the company runs on behalf of private clients, although the management team has some discretion about the divergence they run from the model.
The approach – if not unique in the world of discretionary management, then certainly pioneering – has won endorsement outside the business, with Trinity College Cambridge economics fellow Steve Satchell describing the methodology as ‘leading the field’.
‘Their intuitive framework is one that can be explained to interested clients with relative ease due to its methodological transparency,’ he added in an academic critique of their work.
At the heart of the process is a move away from the mean reversion principles of MPT with its location on an efficient frontier of the outcome of a diversified portfolio of assets, toward an attempt to anticipate the likely market outcomes of a particular index.
The managers track several dozen indicators and their related probability of given outcomes for a dozen key asset classes, then use this to formulate a market stress indicator, which is applied to portfolios on a sliding scale of five ‘stress levels’, used to determine exposure to risk assets.
Primarily, these are volatility and intra-asset class correlation. Jukes notes: ‘Of course, there are other indicators of market stress, but our proprietary tool is the only indicator that accounts for both determinants of multi-asset portfolio risk: volatility and correlation.'
At extremes of market volatility and moments of inflection, this will take portfolios almost entirely out of risk assets. Because it is intended as an investment rather than a trading model, the process has a high threshold for switching between stress levels: both market peaks and market bottoms get missed out by design.
‘Risk is inherently boring. And it is difficult,’ says Jukes. ‘If you look at the dictionary definition then it says “entailing a probability of loss or harm”.
'And that is interesting – that we think of it as asymmetric. In finance it is actually symmetric: there is an opportunity for gain or loss, which is quite a mathematical definition of risk.’
Broadly, the emphasis on probability of outcomes and the tailoring of asset exposure means the process aims to exclude the most extreme outcomes in any given market scenario. While this means that the process will avoid extremities of drawdown, it will also exclude extremities of upside.
‘As humans we are loss averse, but in modelling risk we are symmetric about it. The return distribution [that we aim for] also caps some of the upside. Giving up some of the losses means giving up some of the gains.
‘We are giving up investment risk but partially replacing it with client risk – the danger that clients will leave us because other people are adding more of the upside.
'What has been comforting introducing this to [Remap] clients is that once they understand it and become used to it they frequently become evangelists for it.’
The ‘utility’ threshold applied to expected upside return when adding additional risk into portfolios, means, for instance, that the guidance remained at a ‘high stress’ level through much of the early rally this year and has remained largely consistent over the course of this summer.
‘Although we remained underweight equities [through early 2012], we added a little more risk at the end March, a move that hurt portfolio performance in April,’ the company said in an investment update.
'In mid-May our [Market Stress Indicator] flashed stressed for the first time in 2012, as the European solvency crisis resurged. We immediately reallocated over 10% from equities to bonds and cash.
As of mid-summer the company’s middle risk-rated Remap portfolio comprised 85.2% allocated to fixed income, 4.4% allocated to equity, 1.6% allocated to alternatives (a single position in the ETF Securities Brent One Month ETF) and 8.8% in cash.
The fixed income allocation was dominated by 55% of portfolio assets in two to 15-year dated gilts, around 10% each allocated to the M&G Corporate Bond and Invesco Perpetual Corporate Bond funds and 8.5% in the M&G Strategic Bond fund.
‘Many people are surprised that we continue to hold gilts, but the fact remains that when markets become stressed, gilt yields continue to head lower,’ the managers said in their update.
‘Our successful preservation of capital last August re-cemented this view.
‘Furthermore, while longer-term drivers of yields (such as debt servicing as a percentage of GDP and the amount of foreign-owned debt) point towards an upward trajectory, shorter-term drivers such as growth and inflation are more benign.’