Multi-asset: a cycle without a textbook?

As is often the case at the year-end, when reflecting on what has happened over the past year, we can have the pervasive impression that the year was trickier than usual. This near-term perception of changing regimes and shifting paradigms is a natural human bias and, when it comes to asset management, it can have quite negative side effects, giving the impression that this time really is different and prompting changes to the sets of rules normally applied to investing money. If you know us, you will know how much we disagree with this. We remain steadfast to the three founding pillars that guide every step of our investment journey: we focus on diversifying risk, we are aware of economic cycles and we are systematic in all that we do.

These three pillars can be used when explaining our performance throughout the years. The Covid and post-Covid eras may have been testing for these pillars, even more so than the 2013 taper tantrum or the 2018 trade war, with sharply declining yields, rebounding markets and latterly yields at the long end gaining more than 400 bps. None of these situations has been experienced by us before or was even considered by our “backtests”! Yet, during this period of terra incognita, we have not been dissatisfied with the adaptability of our investment process. So, now that 2024 is at our door and, again, appears to be at odds with the past 40 years, we ask ourselves how different will this year be and what new challenges might we face?

This new edition of Quarterly Simply Put intends to walk our readers through our observations about the current environment. In short, we believe markets are probably ahead of themselves, with many eggs having already been placed in the soft-landing basket. With that, expected Sharpe ratios could be surprisingly low, reinforcing the importance of utilising dynamic rebalancing techniques and being equipped with the right overlays.
 

This cycle will remain unusual

There is a textbook approach to macro and market cycles. Figure 1 shows how their stories progress with both cycles usually obeying common characteristics:

• Late cycles tend to see strong nominal GDP growth, with contributions from both inflation and real growth. Profits progress less than GDP while equity valuation gains are in line with profits.
• Recessions (the term previously used for “landing” periods pre-artificial intelligence (AI)) usually face a contraction across all three of these factors.
• Recovery periods see equity prices rise at twice the pace of profits, with profits overtaking GDP growth as margins progress in a context of higher unemployment.
• Steady growth periods come with decent growth while profits grow in line with equity prices.

Traditionally, these four steps of the business cycle follow each other in a remarkably steady sequence and usually in the order listed. This time, however, things seem to be taking a different turn, probably because central banks are doing their best to deliver a soft landing: their hikes are meant to weigh on economic activity but not to the extent that unemployment deteriorates enough to force markets to plunge significantly. The right side of Figure 1 shows 2023’s data: equities have risen by 20%, which is consistent with a recovery; profit growth is close to 0% (flat), which is consistent with a recession; GDP growth at 6% is in line (in its breakdown) with a late-cycle period. None of these seems to agree on the regime we are in and this disconnect between the “macro” side of the picture (GDP growth), the “fundamental” side of it (earnings) and the “market” side (equity prices) has animated our discussions throughout the year.

As we start 2024, the current picture is mixed and this mixture is something that has (very) rarely been seen since the end of Bretton Woods. So where does this past and current situation leave us in terms of scenario(s) for 2024? Well, we ultimately focus on risk-adjusted performance and we now believe we are at the dawn of a low Sharpe ratios period. 

FIG 1. S&P 500, US corporate profits and nominal GDP changes per regime and in 2023 
 

Source: LOIM, Bloomberg. 
 

Welcome to the era of low Sharpe ratios

Quick reminder. We are not paid to deliver cash returns (this is the central bankers’ job!), but rather to deliver an excess return above cash, ideally by not taking undue risks. Sharpe ratios measure our ability to deliver on this target, measuring high excess returns by unit of risk (in this case volatility). Thus an environment of low Sharpe ratios may mean investors struggle to earn significantly more than a cash return for the additional risk taken. 

Now back to our topic of the day. We think the complexity of the current cycle could give rise to three different scenarios. Our baseline view is that of a soft landing, starting with a central bank pivot that may happen earlier than expected. In this scenario, most markets would be expected to deliver a return of some kind over cash, as the decline of deposit rates acts as a tide that lifts all boats. Our first alternative scenario is a hard landing: this time, rapidly declining cash rates would signal deteriorating macro prospects leading to negative performances for risk-on assets. Only duration as a risk premium would be rewarded. Our second alternative scenario, by order of likelihood, would be a no-landing situation: in this scenario, central banks would need to hike further to contain ongoing inflation pressures. Here, most assets would suffer and cash would be king once again – a repetition of 2022 albeit with a lower scale of magnitude. 

Figure 2 depicts our Sharpe ratio estimates in relation to these three scenarios, obtained from our nowcasting indicators and combined with 5-year consensus return estimates. 2023 was a year with a low Sharpe ratio for bonds (but not negative as happened in 2022), but of a high Sharpe ratio for equities (about 1). Most of our projections for 2024, for each of these three possibilities, point to Sharpe ratios remaining quite contained for the year and also over the long run. In other words, our estimates suggest that beating current cash rates is going to be difficult as risk-taking could be poorly rewarded. Buy-and-hold positions in government bonds currently offer such appealing prospects that it could be extremely tricky for investors to find asset allocations that beat these performances. At best, bonds could deliver a 0.5 Sharpe ratio in a hard-landing scenario (but equities would post a -0.8 Sharpe ratio) while equities could achieve a 0.5 Sharpe ratio in a no-landing situation, but the required actions from central banks to end inflation could very well put the hard-landing scenario back on the table. In a nutshell, as complex as the current cycle may seem, given the elevated level of rates, current prospects in terms of Sharpe ratios – for long-only investments – are low on an ex-ante basis. 

FIG 2. Expected Sharpe ratios per macro scenario and over the long term, based on consensus forecasts
 

Source: Bloomberg, LOIM. 
 

Living by our principles

When considering this low Sharpe ratios scenario, we believe two salient features of our investment process should help us extract some performance from markets:

• First, systematic risk budgeting techniques can help exploit the advantages of leveraged and dynamic exposure. While we rely on a variety of indicators to build a diversified portfolio at all times, we use derivatives exposure to potentially leverage this diversification in order to calibrate exposure to the desired risk and return targets. These techniques can help preserve capital (or “drawdown management”) but can also lead to higher exposure at times. In our view, this dynamism will prove more helpful in dealing with lower-than-average Sharpe ratios than static and limited exposures. 
• Second, our carry, trend and macro overlays can further help navigate these tricky market conditions. Carry can help lift the “implied yield” in selected markets and therefore increase the excess returns of our portfolios. Trend can also contribute, by adding exposure to trending assets (or conversely limiting exposure to stressed assets) and improving our ex-ante Sharpe ratio. Finally, our macro overlay can help us modulate our so-called “40-40-20” implicit regime allocation to add exposure to assets with the highest Sharpe ratios in the likeliest regime. The integration of these overlays is motivated by their potential to improve the ex-ante Sharpe ratio of our solutions and so far, they have proven successful, irrespective of market context.

Yet, this does not prevent us from having doubts. A way we seek to reassure ourselves is by looking at the past performance of our solutions. The last three years have also seen unusual regimes. Figure 3 highlights how our strategies have fared compared to market benchmarks over this period. Key highlights are:

• Making a difference in terms of asset allocation was difficult during 2023, but our disciplined approach made it possible to deliver returns above cash while also containing risk (see Section 4).
• Since 2022, only total market exposure has made a difference. However, even our most highly leveraged solution did not underperform passive 60/40 portfolios.
• Since 2021, our solutions have clearly started to make a difference: the combination of overlays and risk management helped us capture a large part of the 2021 rally.
• Finally, since 2020 (or from the start of the Covid cycle), risk parity benchmarks underperformed cash but we have not, posting stronger returns than passive 60/40 portfolios. 

These achievements are a credit to the team. Yet, our research programme continues and Section 5 details some of the topics we are currently tackling – including how we can further improve this return profile. For now, we simply acknowledge that the current situation shows an apparently high level of complexity but we are assured that past complex periods have been accommodated by our investment process. So we will continue to do our best and look forward to seeing what comes next!

FIG 3. Total return cumulated performance over different periods


Source: Bloomberg, LOIM.
 

Over the past two years, a notable feature of our multi-asset strategy has been the explicit use of cash as an investment. We have consistently communicated on this topic, highlighting the significance of holding cash when diversification fails investors – a situation when no other asset class exhibits positive prospects, not even duration. The one thing that has dramatically changed in our allocation since the October lows is the amount of cash held in our portfolios. The rapid rise of markets, coupled with declining volatility in most risk premia and a renewed appetite for risk among investors, has naturally influenced our gross exposures which have dynamically adapted to this fast-evolving environment. This section outlines all the elements of our asset allocation dashboard and their implications in terms of investment. In short: our systematic process now signals that it is time to normalise exposures and end the prolonged cautious stance that has served us well thus far. Welcome to an era of pivots and positive (yet limited) Sharpe ratios.
 

The unbalanced volatility normalisation

As our regular readers know, at the heart of our process is the consideration of risk. Figure 1 illustrates the recent evolution of our volatility metrics which guide our risk-based allocation across risk premia. H1 2023 told a broad story of a volatility decline across the board. Moving onto H2 2023 and a more nuanced story emerged, with two key observations:

• Trends: first (and foremost) most risk premia volatility exhibited a decline during the last two quarters, as depicted in Figure 1. The decline was somewhat less pronounced than in H1 2023, the average percentiled volatility experienced a more modest drop of approximately 1%.
• Discrepancies: notably, there remains a meaningful divergence between the volatility of duration and other assets. Firstly, volatility in duration remained elevated compared to its long-term history and more recently it has nudged slightly higher – a stark contrast to the steady decline observed in volatility across all other assets. This was especially true for risk assets, such as DM equities and credit, which saw volatility in Q4 dip below the long-term median and touch levels previously seen pre-Covid. 

Elevated volatility in government bonds likely reflects divergent investor opinions about the asset class, stemming from uncertainties surrounding the prospects for future inflation and growth. This factor primarily explains our underweight in duration compared to other risky assets. Our portfolio has never held this little duration throughout its 12-year track record. But this is not entirely due to elevated volatility; sentiment measures have also contributed to the situation. 

FIG 1. Risk premia volatilities


Source: Bloomberg, LOIM as at 18/12/2023. The red zone shows the first quartile of our risk measures and the dashed line the median.
 

A small patch of red in an ocean of blue

In addition to the overall normalisation of volatility, trend signals in risky assets also presented interesting twists. Figure 2 highlights that the contrast in trend signals between duration (ex-China) and riskier assets was again quite stark: note the dominance of red that represents the latter, while the prevalence of blue throughout 2023 denotes the former, with only a small patch of red during September-October marking an oddity in this sea of positivity. While commodity markets have also maintained a positive stance recently, the trend in commodities exhibited general weakness or instability, marked by several reversals, which tend not to bode well for trend-following strategies. 

As we transitioned into December, the blue wave intensified and, post the Fed pivot, even duration started to exhibit a positive trend signal. This exemplifies one of the strengths of our multi-signal approach: the ability to discern between periods of high volatility and negative trend for duration (observed in Q3 and the first two months of Q4) and periods of high volatility and positive trend (which was particularly evident at the end of Q4). This differentiation has, in part, cushioned our reduced participation in the bond upside following the Fed’s policy shift and can partly explain our overall re-risking, in tandem with shifts in risk appetite.

FIG 2. Trend following signals


Source: Bloomberg, LOIM. As at 18/12/2023. 
 

Bullish sentiment, faster re-risking

Risk appetite plays a key role in our dynamic drawdown management techniques. Figure 3 provides an insight into our Global Risk Appetite Index and its message is becoming increasingly clear: 2023 favoured a “risk-on” environment, with occasional dips into the neutral zone. Except for a brief episode in March 2023, our indicator showcased alternating periods of bullish sentiment (above 60%) and neutral sentiment (between 40% and 60%), while steadfastly avoiding a descent into bearish territory. This signal is one of the ingredients of our dynamic drawdown system. When we capture such a bullish signal, indicating a global market upswing, our re-risking process activates more swiftly. This aspect explains another part of the rapid global re-risking observed in our portfolios since the beginning of November. In the presence of such a confluence of favourable market conditions, our systematic investment process interprets this as a signal to amplify global market exposure. This dynamic approach has kept us actively engaged since the pivot, aligning with our commitment to capitalise on opportune market dynamics. 

FIG 3. LOIM Global Risk Appetite and its components


Source: Bloomberg, LOIM, as at 18/12/2023.
 

Moody macro 

The only aspect of our asset allocation process that has retained its cautious stance is our macro risk overlay. As detailed in the subsequent section, a central bank pivot typically occurs within the context of a deteriorating growth outlook. It is hard to anticipate a systematic macro risk overlay, which exploits historical macro patterns, in order to overweight risky assets in such circumstances – quite the contrary. Our ongoing recognition of lingering inflationary pressure is another reason behind our cautious approach towards bonds. Consequently, the macro risk overlay has maintained a slight underweight to risk-on assets and a neutral position in duration. Such a position reflects the typical scenario observed during periods characterised by slower growth, a central bank pivot and escalating inflation. Unsurprisingly, the contribution of this overlay is relatively modest compared to the other elements in our investment process.  However, it is crucial to acknowledge that circumstances can change, highlighting our adaptability to various scenarios and uncorrelated signals as we are not committed to a single outlook or uniform set of signals. 
 

Bullish re-risking

The outcome of these asset allocation mechanics is illustrated in Figure 4, which draws a distinction between the asset allocation formula (rebased to 100%) and how much of this formula is deployed into markets (our total gross exposure). As we conclude the quarter, two key points emerge:  

• First, our asset allocation mix continues to maintain an underweight position in bonds and an overweight position in other risk asset classes, in stark contrast to our long-term average. While this bias was even more pronounced in September, it moderated in December, influenced by trends and macro factors closing their underweights. However, the relative volatility situation persists.
• Second, our market exposure has surged meaningfully. For example, gross exposure for our balanced profile increased from a low of 40% at the end of October to approximately 110% at the time of writing. This rapid capital redeployment is a result of the factors mentioned earlier: improved sentiment, positive market performance, a reduction in drawdowns and a further decline in the volatilities of risk premia.  This strategic move leaves us well-positioned to meet our return objectives, even amid the challenging environment expected for Sharpe ratios, as discussed in Section 1. 

FIG 4. All Roads Strategy positioning rebased to 100% (left) and the evolution of market exposure (right)


Source: Bloomberg, LOIM, as at 18/12/2023.
 

With the Federal Reserve (Fed) having announced its pivot and the European Central Bank (ECB) committing itself to the exact opposite in December, the topic of central bank pivots is dominating the headlines. Pivots are a relatively rare thing in economic history, which can add further confusion. A pivot is the moment a central bank changes the stance of its monetary policy – be it with words or with deeds. Those pivots can be either dovish or hawkish: January 2019 was a dovish pivot (meaning a switch from hawkish to dovish) while 2022 saw a hawkish pivot (from dovish to hawkish, remember “transitory inflation”). This quarter, we are focusing on what the Fed has just done – a dovish pivot, where it has switched from inflation-fighting rhetoric to an inflation-propelling one. The question to ask now is “why?” What makes central banks pivot, what impact can it exert on financial markets and what could be happening this time? This section attempts to make the situation clearer – if possible...

What is currently expected

The key highlight for Q4 was monetary policy once more. Figure 1 shows how rapidly monetary policy expectations have evolved since July 2023 for both the Fed and the ECB. The quarter opened amid the “high for longer” rhetoric but “longer” has turned out to be around nine months as, since the Fed’s pivot, markets are now anticipating five rate cuts for the year, the first of which could occur in March or April. This is a very significant change in monetary policy stance and even those markets that were hoping for this change were surprised by it. Despite the ECB’s December speech and its outright reluctance to even discuss a pivot, a similar scenario is expected from the ECB, with a first cut forecast to happen in August – the typical six-month lag between the US and the Eurozone. These pivots are now clearly priced in, creating the question of whether it is reasonable for the Fed to be moving towards it and for the ECB to dispel it. 

FIG 1. Monetary scenario evolution (Fed on the left and ECB on the right)
 

Source: Bloomberg, LOIM (Fed is derived from Fed Fund Futures and ECB from the OIS yield curve).

On the road to pivot-land

Before we delve into the long-term time series to try and break down the code behind such pivots, let us just refocus our attention on what has just taken place in the US and the Eurozone in terms of the macro situation. The beauty of the way we look at macro – through a systematic lens – is that we can dissect the evolution of macro forces on a daily basis. Figure 2 shows the daily evolution from June 2022 to December 2023 of our nowcasting indicators, gauging the evolution of growth, inflation surprises and monetary policy shocks. Their recent evolution shows how:

• Growth has slowed down during the period of tightening financial conditions, without declining below the 30% threshold which usually separates growth periods from recessions. This is what is currently being coined as a “soft landing”. 
• Inflation pressures collapsed during the period, particularly before March 2023 when they started rising again in the US (but not in the Eurozone). This is disinflation: the level here matters more than the direction (for now). 
• Monetary policy has shifted from a clearly hawkish territory to a neutral zone that started in December 2022 and ended in August 2023 after the last Fed hike. Since that moment, our indicator has been consistent with a dovish Fed, with the ECB remaining neutral.
   

So here we are, facing a disinflationary soft-landing scenario with a unique point of differentiation between the US and the Eurozone as central bank policy has only pivoted on one side of the Atlantic, but the other side will probably pivot in the coming two quarters. From that perspective, the December situation is not entirely surprising, yet we fail to understand the genuine reason behind it. Maybe a look at the long-term hard data will clarify things.

FIG 2. Growth, inflation surprises and monetary policy shocks nowcasting indicators


 Source: Bloomberg, LOIM.

What makes the Fed pivot

The nowcasting indicators, by construction, only employ survey data, aka “soft data”. At the moment, the soft data – such as the ISM surveys or the European Commission surveys – are sending a macro message that can be quite different from what the “hard data” are showing, such as industrial production, GDP or retail sales. From a hard data perspective, what makes a central bank pivot? If inflation is always the reason for a hawkish pivot, growth is usually the reason for a dovish one. Table 1 shows the average departure of key time series from their recent trends around pivots. For instance, when unemployment stands high enough above its “normal” level (its recent trend, average past four years of data), then it should prompt the central bank to pivot and hint that its next move will be a cut. Table 1 shows the average departure from trend of the unemployment rate, core inflation, headline inflation and GDP growth that prompted the Fed to pivot during the period from 1945 until 2022. The answer is simple: when the unemployment rate, and core and headline inflation were on trend, and when growth was about 60 bps below trend. Applied to today, this suggests an unemployment rate of 4.5%, inflation in the neighbourhood of 4% and GDP growth at about 2%. The table shows the spread between these targets and today’s value. 

In all fairness, from that perspective, the Fed was not far away from being in its usual pivot environment – only unemployment remains a bit too low for that. But how should a central bank act to achieve a soft landing if it doesn’t react slightly ahead of time to avoid a big increase in the unemployment rate? If you look carefully at the bottom of table 1, it shows that “after” a pivot unemployment increases to sit 0.4% above trend while inflation falls below trend and growth continues its decline. What the Fed orchestrated in December was to simply pre-empt this usual pattern, in the hope that by acting in advance of its usual pivot it could deliver the undeliverable: a soft landing. 

Table 1. Average macro situation around pivots in terms of inflation, growth and the unemployment rate (in excess of trend)
 

Source: Bloomberg, LOIM.

Markets in a time of pivot

So why is the Fed being so careful about pre-empting the usual macro moment to pivot? Most likely, this is out of concern for employment and growth, but financial market consequences are probably also being considered. As the reader may already know, two of the channels for monetary policy to transit through to asset prices are the wealth effect and risk appetite. An important part of achieving a soft landing is to limit the decline in consumption, which comes with a negative wealth effect, and the decline in investment that originates with a strong surge in risk aversion. If you are not convinced that a pivot usually coincides with declining prices and weaker risk appetite, Figure 3 provides further evidence. It shows the performance of the S&P 500, the change in AAA yields and BBB spreads before, during and after pivots, again using data since 1945. It shows that just before a pivot, equities usually advance, but decline during the month of a pivot and the quarter that follows (“after pivot”), while yields rise ahead of a pivot and then decline afterwards. Finally, spreads compress at the time of a pivot and then widen afterwards. It is essential to keep these numbers in mind to better understand what will happen during Q1 – should we really see an early pivot from the Fed, these negative side effects could be limited. This covers some of the financial consequences, but what about economic impacts?

FIG 3. S&P 500, rates and spread fluctuations around the pivot time


Source: Bloomberg, LOIM.

What could be different this time?

In terms of economic consequences, this is probably where the Fed intends to make a genuine difference. Figure 4 offers a classic view of monetary policy but probably provides the most informative perspective at the moment. The chart compares real yields (10-year, US) versus our estimate of “trend growth” in the US, that is the natural growth rate of the US economy, assuming it is not derailed by external shocks. At the moment, our estimate agrees with that of the IMF, being slightly above 2%. What it means in terms of the conduct of monetary policy is that unless real yields remain above that level, financial conditions are not restrictive enough to slow down the US economy. Recent history, and especially that of Q3, shows that the “high for longer” mantra led real yields to experience a stint above that level to reach 2.5%. However, the Fed’s pivot immediately caused real yields to decline sharply below 2%, reaching 1.7% at the time of writing. As made clear by the chart, every recession has coincided with real yields climbing above real potential growth – another name for trend growth. This was the case in 2001 and 2008: usually the economy cannot withstand more than a quarter or two of this expensive funding cost and activity starts to roll over. Paying close attention to what has happened of late: the Fed has let real yields exceed potential growth and shortly after that (about one quarter instead of the usual two) has decided to pivot. Here again, the visible hand of a central bank trying to achieve a soft landing is revealed. 

FIG 4. Real rates vs. potential growth in the US
 

Source: Bloomberg, LOIM.

In essence, the Fed is trying to go against the usual patterns, trying to pre-empt its “normal” way of doing monetary policy. It could refrain from cutting short-term rates during Q1, since by just using words it has already caused real yields to revert to an accommodative stance. The Fed is buying itself time to see how the macro situation plays out. The only guarantee we have is that this disinflation process could last even longer than anticipated – the ex-Fed Chairman Arthur Burns gave it a try in the ‘70s, hoping for a rapid economic reaction to avoid what Paul Volcker later had to do triggering two consecutive recessions. Let’s hope the Fed does not make this same mistake and retains its independence from political forces. At the moment, the ECB is reluctant to go down the same path and openly prefers a recession to end inflation for good (as in China) rather than experience pervasive above-target inflation. Call it ghosts from the past, but this will be a key difference to monitor in Q1.
 

The year put our investment philosophy to the test again. Our solutions rely on three distinct pillars: a systematic investment approach, a core principle of disciplined risk management and a tactical understanding of macro and market cycles – all of which proved to be essential in 2023. As we often explain when introducing our solution to clients and prospects, our portfolio construction process is composed of an asset allocation step – the asset mix, rebased to 100% – and a sizing step – deciding how much risk we need to deploy in the markets and, when necessary, how much should remain in cash. Over the past two years, both of these decisions have had meaningful consequences on our portfolios. 2023 has proved challenging to active investors: many allocation choices would have resulted in an underperformance versus passive investment mixes or, even more annoyingly, cash. Against this backdrop, our investment strategies have delivered close to their cash+ targets, and importantly, in line with their risk budget. Let’s review how this was done. 

The cost of making choices

As in most years, 2023 involved specific allocation decisions, within both equities and bonds, as well as between equities and bonds. Usually, both decision levers (within and between asset classes) exert an influence on overall performance, but the allocation between equities and bonds is typically the most important: how much equity risk can be borne. Over the past two years, strange things have happened: in 2022, regardless of asset class implementation and irrespective of equity-bond allocation, the resulting performance was the same (deeply negative). This past year delivered a different puzzle: this time it was not whether you preferred equities or bonds that mattered, but which type of equities and bonds. In other words, intra-asset class decisions had an unusually meaningful impact. 

Figure 1 shows the performance year-by-year of different allocation schemes, using different asset class implementations – an equal weight and a market-cap weighted S&P 500 Index and similarly weighted (sovereign) bond indices. As shown, 2022’s performance would have been the same regardless of the allocation decision and the investment support. This past year was an entirely different story. With equal-weight solutions, the allocation decision mattered little: performance in excess of cash would have been about the same and near zero. With market-cap investment vehicles, the outcome is sharply different: the performance in excess of cash varies between 2% and 15%. In other words, 2023 was not only a matter of how much equity risk you were willing to take, but also (and even more so) what kind of risk you were taking within your portfolio. As primarily derivatives-based investors, this had a limited impact on our performance. 

FIG 1. Performance in excess of cash of various multi-asset solutions using market cap (left) and equal weights (right) instruments (USD)
   

Source: Bloomberg, LOIM.

The least concentrated year since 2005

Another issue with 2023 was that there were many wrong ways to be exposed to markets, and underperforming cash was a real possibility. How can this be illustrated? A Principal Component Analysis can help us examine how much of our multi-asset investment universe can be explained by its first risk factor: a market factor that loads positively on all assets within our portfolio. The explanatory power of that factor is a natural measure of “beta” opportunities. When beta is low, “alpha” opportunities dominate the investment universe, and with alpha, as has been apparent this past year, there are many ways to underperform when trying to time such opportunities. Figure 2 shows the portion of market risk that can be explained by the “beta” factor. From that perspective, 2023 has been a year with the lowest risk concentration, a level not seen since 2005. Another relevant comparison point is 2013, a year of two halves when “taper tantrum” declines were followed by an end-of-year rally. This chart also highlights the years that were more difficult for multi-asset portfolios to capture the right beta exposure.

To see how “alpha” forces can dominate markets, Figure 2 also shows the same concentration metric through the lens of alternative risk premia strategies. In this case, a higher-than-usual concentration typically means such investment styles were able to bring more value to investors, i.e., that systematic alpha was exceptionally rewarded (if on the right side of the trade of course). In 2023, alternative risk premia exhibited higher levels of concentration, particularly in CTAs and carry strategies. To put things simply, traditional risk premia opportunities were hard to capture in 2023, while opportunities for alternative risk premia were more aligned. From that standpoint, 2005 offers a relevant comparison point.

FIG 2. Risk concentration within traditional and alternative risk premia per year
 

Source: Bloomberg, LOIM.

How do we address this? Part of the solution is sizing market exposure correctly…

Against this backdrop, our risk-managed multi-asset solutions performed well, and more so when the previously listed headwinds are considered. But what does “well” mean? Two different, but essential, things. First, it means that our solutions delivered a performance that was in line with their respective excess return targets¹. Given the high level of cash rates, that was not an easy task. Second, it means that on a risk-adjusted basis, the performance of our solutions also held up against passive allocations. 

There are three different reasons why our solutions fared well during this unusual period. The first is shown in Figure 3 and is tied to the idea of having enough diversified risk. This chart compares the performance of the Bloomberg Risk Parity Index for different risk targets, ranging from 3% annualised volatility to 10%, and compares that performance with the return for cash and the Bloomberg 40/60 Passive Index. The 3% volatility risk parity strategy did not stand a chance versus cash over the past five years, let alone versus the traditional 40/60 portfolio. Its underperformance is not an issue of asset allocation – the smoothness of its path remains attractive, but it simply shows that less risk over the long run leads to lower returns. The solution to that is well known: by using derivatives, one can reach a risk level that makes it possible to deliver a performance higher than cash and closer to the 40/60 strategy. As the 40/60 strategy has a historical volatility of around 10%, there is little wonder that the 10% risk parity solution was the closest match. 

FIG 3. Rebased performance of the Bloomberg Risk Parity Index for four different target risk levels

 

Source: Bloomberg, LOIM.

… but market exposure is not enough

This point about target risk was a contributing factor to the success of our portfolios in 2023, but not the only one. As shown, leveraging risk-balanced solutions was not enough to beat the 40/60 portfolio. The other two factors that helped us meet our targets were the “macro” cycle view within our risk-based allocation and our tactical overlays. The issue with risk parity is that, at its core, it implies an unusually high frequency of regimes for which bonds are the right hedging asset. In simple terms, the bond allocation in a risk parity portfolio is fantastic for economic cycles with 50% of recessions. But recessions typically occur far less frequently – thank goodness! For a cycle that involves a higher frequency of growth regimes and inflation regimes, such an allocation would be sub-optimal, in our view. At the core of our allocation, our 40-40-20 mantra has served us well this past year, helping us have a larger exposure to cyclical assets, such as equities, credit and commodities, which added value this past year. Finally, as mentioned previously, alternative risk premia were an interesting play during the year, especially carry strategies. Using them as an overlay contributed the final brick to our performance for the year. Running enough risk, maintaining an understanding of macro regimes at the heart of the beta exposure and complementing that with alpha overlays were the three pillars that enabled our solutions to achieve their performance objectives this past year – all the while being deployed systematically. 
 

It is widely accepted that financial and economic time series are influenced by alternating regimes, which change the behaviour of assets risk premia. A common assumption is a two-state representation – which can then be sampled in finer time periods – between good times (economic expansion, bull markets) and bad times (bear markets, recessions). However, this recurring pattern may be disrupted by infrequent, rarely observed, distinct regimes that are difficult to model and can have undesired effects on portfolios. 

I have no memory of this place: a new regime or an unusual observation?

Rare adverse regimes are, by their very own nature, difficult to handle when constructing a portfolio. On the one hand, incorporating such regimes into a statistical representation of the investment universe (i.e., the unconditional distribution) may distort the portfolio; trying to hedge a regime that is rarely observed can be detrimental to the performance of the most frequent regimes – the famous mantra that “one needs to stay invested”. On the other hand, discarding such information is also debatable as it may result in large drawdowns when such a regime does occur. 

It could be argued that the past two years have witnessed one of these adverse regimes. Positive correlation between bonds and equities, coordinated drawdowns and equity volatility remaining unreactive while sovereign bond volatility spiked are all markers of odd market conditions. When faced with unusual points in the data distribution, one has to decide between three options:

• The simplest one is to discard these odd points, trimming those points that are deemed uninformative from the overall distribution. Of course, this simple solution is open to criticism as it is somewhat arbitrary to ignore observations. 
• A second option is to simply allocate these odd data points to one regime. This is a way of limiting the number of regimes, but risks diluting the regime-specific information. 
• A third way is to add a new regime to the data, but this risks overfitting the model with unrealistic regimes that may never be repeated.

In the case of rare regimes, the treatment is difficult because the observations of this regime are scarce. Simply blending it into an existing regime should not significantly change the specific distribution for which the odd regime is added, it should only increase its tails. For that reason, the two-regime model is often favoured as it allows us to easily distinguish between risk-on and risk-off periods. 

Risk parity or regime parity? 

Although our focus on regimes may appear disproportionate – indeed this is a topic we often discuss in this section – their importance for risk-based investing is paramount. Regimes directly impact the conditional distribution of returns (the behaviour of financial assets is not the same in bear or bull markets), meaning they distort the unconditional distribution (the behaviour of financial assets independent of the regime) and have a tremendous impact on the sampled covariance matrix. Ignoring regimes thus results in risk-based portfolios that are inept at capturing diversification opportunities, often overloading the portfolio with defensive assets. Among risk-based portfolios, the Equal Risk Contribution (ERC) is prominent as it brings a natural answer to the risk-concentration problem of capital-based allocation. 

The derivation of the ERC portfolio is based on the decomposition of the total risk of the portfolio, denoted  where denotes the capital allocation, into individual risk contributions:
 

.

The ERC portfolio is thus the portfolio such that and .

In a recent white paper², we introduced a new risk-based portfolio that explicitly uses regime-level information to build a regime-resistant long-term allocation. This Regime Parity (RP) portfolio preserves the risk-diversification features of the risk parity portfolio by relying on a regime-specific ERC solution but renders explicit the embedding of regimes by weighting the local ERC portfolios by the probability of occurrence of each regime. Thus, the regime parity portfolio is a linear combination given by:

where  denotes the number of regimes,  the frequency of each regime , and  is the local ERC portfolio calculated on the multivariate distribution conditionally of being in regime .

Ignorance is bliss (or is it?)

By construction, if the regimes used in the Regime-Parity portfolio are uninformative, the portfolio should coincide with the standard ERC portfolio as the conditional distribution should be similar to the unconditional one. However, when an adverse regime is ignored, the regime-specific distribution will be disturbed and the regime-parity portfolio will artificially diverge from the ERC, suffering from a misspecification in the structure of the conditional distribution.

To illustrate this issue, we created theoretical portfolios based on statistical assumptions. For simplicity, we consider a two-asset, three-regime case, with a frequent regime representing the good times when the risky asset yields high returns; a “bad” regime where the risky volatility of the asset spikes while the hedging asset performs well and allows for good portfolio diversification; and a “rare and adverse” regime where both assets are correlated in a drawdown amid lower volatility than in the “bad” regime. The parameters used in this theoretical exercise are presented in Table 1.

Table 1. Parameters used for the conditional distribution in each regime
 

Because we know that the three regimes in our simulated example are well defined, a Regime Parity portfolio should be able to perform better than the standard unconditional ERC portfolio, which is not conditioned for the regimes underlying the asset distribution. However, what happens when the rare regime is ignored and is mislabelled in favour of the frequent regime? In this misspecified two-regime scenario, we developed a portfolio that is different from both the standard ERC portfolio and the three-regime Regime Parity portfolio. 

FIG 1. Theoretical Sharpe ratios of the ERC portfolio, the misspecified 2-regime RP portfolio, and the 3-regime RP portfolio


Source: LOIM. 

Figure 1 highlights two interesting features of the Risk Parity portfolio. If the rare adverse regime is ignored, the Risk Parity portfolio yields an allocation that is worse than the unconditional ERC portfolio in terms of risk/return balance. However, the well-specified three-regime Risk Parity portfolio overperforms the ERC portfolio, emphasising the actual benefit of filtering out the rare regime from the unconditional distribution of returns. 

In summary, regimes underlying financial time series are paramount to portfolio construction as they enable the sequencing of risk-on and risk-off periods and build coherent portfolios that account for these alternating regimes. However, the occurrence of short-lived, rare and adverse regimes is difficult to model and is often discarded in favour of the “good time/bad time” consensus. Such regimes should not be ignored, however, as an adequate weighting of this specific information may prove useful when building long-term risk-based portfolios.