In a nutshell: Our macro nowcasters are increasingly issuing late-cycle signals and such periods pose a lot of questions for multi-asset investors. The current expensiveness of equities could be more of an issue for 60/40 portfolios than for extensively diversified solutions, such as ours. Should 2% real rates become the norm, history shows that dynamically rebalanced portfolios and the search for diversifiers can help deliver outperformance.

 

The simplest way to describe the current economic situation in a couple of words is (literally) “late cycle”. Market participants have been remarkably impatient in their assessment of this cycle (and so have we), expecting that the Federal Reserve’s (Fed) rapid tightening followed by that of the European Central Bank (ECB) would prematurely end the positive growth episode that has wrong-footed so many investors. But here we stand: the S&P 500 Index remains in the neighbourhood of 4,500 while the trend across manufacturing surveys has improved. Before unpacking how this late cycle could drive tactical asset allocation, there are two questions we want to tackle:

• A “late-cycle” scenario usually comes with higher rates and expensive equities. Will this jeopardise the expected returns of a risk-based strategy, such as ours?
• Should higher rates last for longer than expected, what should we expect from a risk-based strategy, especially in the context of more traditional 60/40 portfolios?

 

These questions are reflective of our recent interactions with clients and prospects, and our quarterly update is the perfect occasion to debate them, with our answers being informed by our multi-asset research. So, starting with these structural questions and followed by a comment on our tactical views, here is our take on why this late cycle could well be a land of opportunity for multi-asset and risk-based investing.

 

Late cycle, don’t be a stranger

It’s been an awfully long time since we were last confronted with a late-cycle scenario. The slowdown that was initiated in 2018 didn’t turn into the anticipated recession as the pandemic suddenly struck. The sequence was there, however: inflation was building up from 2016 onwards, central banks were hiking and a slowdown was burgeoning (with a trade war on top). The pivot from the Fed in Q1 2019 could be compared to how it shifted gears in late 2007 as the economy increasingly showed signs of struggle. But with the pandemic shutting down our economies in early 2020, an entirely different narrative emerged with a market recovery starting in late March. It is hard to draw any conclusions from that unusual episode. We would rather use the 2007-2008 period as a more typical example of an economic slowdown within a full and extended economic cycle.

As discussed in our weekly Simply put publications, a typical late-cycle period, such as 2007, can be characterised by key trends, including higher rates, emerging markets outperforming developed markets, positive commodity returns and a lower US dollar. We have not yet ticked all these boxes in 2023, but to us, evidence of late-cycle trends are slowly but surely building up. As you may know by now, we are investors who refuse to see the world through the lens of a unique scenario. The current situation may look disturbing if you only take a medium-term perspective. But if we consider it as the end of a long-term cycle, risky assets are expected to become expensive – valuation is usually a slow-moving metric. So, if your typical core engine (such as equities) is becoming expensive, medium-term investors may well be progressively worried about future returns.

Right now, expanding one’s investment universe should uncover better value that was or is building up in other market segments. Diversified, but adaptable, investment solutions are precisely engineered to capture the changing value of an expanded investment universe, as a variety of indicators progressively help identify currently or formerly unattractive risk premia as the diversifier(s) of tomorrow. Bonds are probably now one of the more attractive return sources, as long as they are considered and used to their full potential, including longer duration or higher notional exposure. Volatility is another example: this investment space has behaved as a “value play” for the last two years. Finally, commodities have become extraordinarily helpful in diversifying cross-asset portfolios over the past few quarters. The latter also reminds us of previous late-cycle episodes, even though the current one has also been magnified by geopolitical tensions and a slowly evolving structural background linked to the forces of climate change and the energy transition.

So, what is the value of increased diversification now? Figure 1 illustrates our answer, by showing our novel valuation signal across the constituents of different assets and investment solutions, including our risk-based asset allocation approach. Our valuation metric, which is discussed in Section 5, is based on a distance measure between prices and their trend. A score of 1 (resp. -1) indicates the maximum expensiveness (resp. cheapness). Using this metric, developed equities appear expensive and have consequently had a major impact on the valuation of more equity-driven allocations, such as 60/40 portfolios. Meaningfully increasing diversification can help capture the relative cheapness of other market segments or assets: our preferred risk-based strategy shows a similar expensiveness metric as a 60/40 portfolio at this stage, but it comes with higher diversification across a variety of cheaper risk premia.

 

Figure 1: Valuation measure for different asset classes and investment solutions (1 = expensive, -1 = cheap)

Source: LOIM, Bloomberg.

 

Higher yield, lower performance?

If late-cycle valuation challenges can be solved in part by increased diversification, will diversified solutions still be tested by the “higher for longer” central bank mantra, or simply because long-term yields could remain elevated for the foreseeable future? The 60/40 portfolio (and its close variations) benefited from the prevailing declining rates environment from 1985 to 2020, and was again helpful for investors as the pandemic knocked on the door. But not since then. Amid steadily rising rates, bondholders have not had the opportunity to profit from bonds’ reconstituted carry. At the time of writing, a global government bond index has delivered just 2% in 2023 (Bloomberg), less than its average carry. The missing piece has been absorbed by the negative duration impact. Moreover, most major yield curves remain flat or inverted, suggesting cash is king and bonds are not being well paid unless rates start to stabilise or go down.

So what happens to diversified risk-based solutions during a higher rate cycle? We think the key point is not necessarily yields, but rather real yields. The key feature of the third quarter has probably been the fact that 10-year real yields have anchored themselves closer to the 2% level, having previously hovered around 1.4%. We do not model expected returns in favour of monitoring the relationship between risk and performance in excess of cash. Figure 2 shows the empirical relationship between the Sharpe Ratio of our risk-based solutions and real rates. Empirically, this performance highlights a negative relationship: higher real rates come with a lower Sharpe Ratio 5-year forward, on average. With real yields at around 2%, the expected Sharpe Ratio sits slightly below 0.5, but with a significant dispersion: it ranges between 0 and 1.5, something that does not happen with negative real yields. On a standalone basis, we are not too worried about the current potential of diversified portfolios to deliver an excess return over a medium-term horizon, but we are just worried enough to continue enhancing our overlays with strategies that incorporate macro data, or introduce a value component or trend signals.

 

Figure 2: Realised 5Y Sharpe Ratio vs initial US 10Y real yield

Source: Bloomberg, LOIM. Based on US data. Data as at August 2023.

 

The relative merits of risk-based investing

On a relative basis, we feel more comfortable with the current higher rates environment thanks to the risk-based solutions that we run. When comparing our approach to a “benchmark”, investors would naturally confront us with a 60/40 allocation. In our view, a major failure of such a static asset allocation is the rigidity of its hedging component (and its lack of alternatives): it relies on the capacity of equities to deliver returns and on the potential of bonds to lower the impact of equity downturns, as though bonds were a riskless asset. But duration is a risk factor, and when it fails (such as in 2022, at the same time as other major assets) there is no alternative diversification source to be found in a 60/40 allocation. Risk-based investing, especially the “macro risk-based” strategy that we propose, offers a solution to that issue by including risk premia that are not present in a 60/40 portfolio, such as commodities, inflation linkers / swaps or long volatility strategies, on top of which systematic rebalancing mechanisms can add further diversification.

The challenge also lies in adjusting exposures in a way that makes it possible to benefit even more significantly from the increased diversification potential of any portfolio component. Figure 3 compares the Sharpe Ratio of a simplified version of our strategy with that of a 60/40 allocation during key periods. The 2011-2020 period did not play to our advantage and neither did the 1984-1998 period: both of these periods were a perfect fit for a 60/40 allocation, with hindsight showing bonds to be the top diversifier and equities driving returns. Other periods highlighted a shakier path for rates and interesting uptrends for commodities, leading to a clear outperformance for risk-based portfolios. It was particularly notable in the high rates period from 1971 to 1983 that higher rates can lead to outperformance for risk-based solutions should the investment universe be large enough and rebalancing techniques efficient. The 2021-2023 period shows that higher rates are not necessarily the investor’s enemy.

 

Figure 3: Difference in Sharpe Ratio between an equity-bond-commodities risk-based solution and a 60/40 allocation

Source: Bloomberg, LOIM. Data as at March 2023.

 

So this quarter is the perfect occasion for us to again observe the different parts of our strategy and see how they behave during a treacherous transition period. We see value in our overlays, which are forcing us to increase risk (see Section 2). The outcome will naturally depend on how right we have been in our assessment of this late-cycle phase – our macro views tend to corroborate this scenario without being overly reliant on it (see Section 3). Then, with a possible rise for commodities and equities, our dynamic risk budgeting techniques are likely to prompt a further increase in risk – Section 4 discusses “how” and “by how much” should that uptrend unfold. Finally, Section 5 reviews our research endeavours.

 

In a nutshell: Duration remains our main underweight at the moment. It would take a sharp normalisation of volatility (and risk budget) to restore this pillar of our portfolio to its historic norm Our solutions have added reflation assets of late, notably commodities and inflation linkers In light of the appearance of late-cycle dynamics, we have also become more positive towards risky assets, albeit temporarily

 

We entered Q3 with the dominant scenario being a deteriorating macro situation would bring about a difficult end to the summer; actually, the opposite happened: our risk appetite and trend signals continued to push for higher equity and credit exposure, and finally got the upper hand. What remains from our early Q2 positioning is a meaningful underexposure to bonds, but that is no longer the only tactical position our suite of indicators is directing us to. Our macro signals started with conflicting messages (bullish market signals being at odds with bearish macro signals) but finally united and are also aligning with the persistent message from our trend signals, leaving us with what increasingly looks like a set of late-cycle positions: underweight duration, overweight emerging market assets and commodities. Let’s take a walk down the aisle of our tactical signals.

 

Duration remains volatile

One thing that has not moved on our radars over the past three months is the situation for fixed income volatility. Figure 1, as shown in previous quarterly reviews, might give the impression that little has changed. As highlighted by the charts, the one area where volatility has not yet normalised from its 2022 levels is duration. Over the long run, duration volatility is around 2.5%. In 2022, as during Covid, that number topped 6% (more than double the long-term level) and, at the time of writing, duration volatility is still around 4%, or its 85% percentile versus history, showing it is still quite elevated. For the other asset classes we track, volatility has returned to or below its long-term average and so this remains a bonds-specific issue. This fact alone is reflected within our portfolio, where we have adopted a strongly defensive position in this risk premia. Now, a key feature of a “late cycle” period is the associated rise in interest rates, at least at first. We see the progression in real rates in Q3 as an element of consistency with the duration message from our risk model.

 

Figure 1: Risk premia volatilities

Source: Bloomberg, LOIM

 

A growing late-cycle trend

Remaining on the late-cycle topic, our trend strategy shares the view that is implicit in our risk measures: summer momentum has led our signals to position an increasing underweight in bonds and this signal remains firmly in place at the moment. This is not the only late-cycle trade that emerged during the summer: our trend signal for commodities also gained in strength, combining with short duration into a reflation position. The recent uptrend in commodities could push headline inflation numbers higher, adding to upward pressure on rates. This is probably the first of our tactical positions entering into Q4. Added to that, the persistent bull signal regarding credit and developed market equities remains positive. In the case of developed market equities, the signal is stronger for US stocks than for the rest of the world. Emerging market equities are favoured by our trend strategy and here it disagrees with our macro signals, which we will discuss later.

 

Figure 2: Trend-following signals

Source: Bloomberg, LOIM. As at 20 June 2023.

 

Sentiment remains positive…

Sentiment has probably been the one factor that has most surprised investors in 2023. If equities at the beginning of the year were undervalued and under-owned, sentiment has helped normalise these anomalies and investors have progressively increased their equity exposure. Whichever way you look at risk appetite, it recovered significantly over Q1 and Q2 and remained elevated throughout Q3. Figure 3 shows this, with returns momentum (“Regime Switching”), hedging demand (“Market-based”) and risk-reward across assets (“CAPM”) all agreeing on the positive sentiment animating markets. If the first three weeks of August exhibited some signs of exhaustion, September has enabled our gauge of risk appetite to remain elevated.

 

Figure 3. LOIM global risk appetite and components

Source: Bloomberg, LOIM, as at 20 June 2023

 

… while macro signals are now more constructive

Q2 ended with our macro signals hinting at an upcoming deterioration in growth conditions, while inflation pressures were receding. In terms of asset allocation, the former was a sign of caution, while the latter carried a more equity-friendly message. Surprisingly, Q3 saw an improvement in growth data, especially in the US, while inflation pressures started building again – the opposite of Q2. The key component of our growth signal showed that the recovery was mainly in manufacturing activity, which is the one sector that had been deteriorating previously. With that, our “Macro Risk Premia” signal adopted a bullish stance: growth is moving in a better direction, complementing the positive risk appetite signal. Inflation pressures moved from “low and declining” to “low and rising”, moving equity positioning from bullish to neutral. Finally, monetary policy remained in muted territory overall, allowing the positive signals for risky assets to gain the upper hand.  

 

The end positioning result

Combining these elements yields the positioning shown on the left-hand chart of Figure 4. Rebasing our allocation to 100% shows how our “asset allocation recipe” is affected by the different indicators compared to our long-term (neutral) allocation. High duration volatility explains half of our current underweight in bonds, the rest stems from trend and macro signals. Our reflation positioning has been expanded as our trend and macro overlays added to inflation-linkers and commodities. Amid the improved macro backdrop and stronger sentiment, our allocation now overweights risky assets, namely credit and developed market equities. The case for emerging market equities is more mixed, as trend signals an underweight and macro an overweight. Overall, as Q3 ends, this leaves us with a portfolio that reflects late-cycle dynamics: higher rates, higher commodities and a (last?) bout of risk-on performance. Q4 will help us fine-tune this position, notably as to when rates might need to come down as inflation stabilises or not. With the persistent shadow of past drawdowns, the right-hand chart of Figure 4 highlights that our exposure to markets is limited at the moment – should markets continue their recovery we could see our portfolio add market exposure globally. Section 4 focuses on this point.

 

Figure 4. All Roads positioning rebased to 100% (left) and evolution of leverage (right)

Source: Bloomberg, LOIM. As at 30 March 2023. Holdings and/or allocations are subject to change. Past performance is not a guarantee of future results. For illustrative purposes only.

 

In a nutshell: Our nowcasting indicators are consistent with a late-cycle period, signalling stabilising growth while inflation pressures progress again, albeit temporarily Past late cycles have tended to see higher oil prices – with central banks not hiking against that inflation – and manufacturing activity experience temporary improvements Of these common trends, the outperformance of emerging assets is likely to be the next stage in this not-so-abnormal soft-landing process

 

The lessons we drew from 2020 were wrong. Its rapid pace gave us the mistaken impression that cycles would last only two years. Since the pandemic ended, we have been hit by a wave of demand- and supply-driven inflation and rapid central bank hikes leading us to expect a recession in a couple of quarters. Yet, at the moment, our indicators highlight that the manufacturing sector is recovering – a recovery without a recession. More intriguing is the novel increase in inflationary pressures, this time around being driven by the rise in commodity prices. So is this environment at odds with historical experience? Not quite: what we are experiencing are simply standard symptoms of a late-cycle period. But what defines such a period? And, more importantly, what are the likely next steps in this cycle? This section reviews these different elements to help better identify what to expect should this diagnosis of a late-cycle phase be proven right. Here is late-cycle 101.

 

Growth stabilisation, rising inflation pressures and central bank normalisation.

The message from our economic indicators has changed during the third quarter in a meaningful way:

• Our growth nowcaster was low and deteriorating and has turned into a low but improving signal. As troubling as this signal may seem, it is essential to remember that these indicators are agnostic to the duration and shape of the cycle. A recovery signal without a recession could be a temporary phase of the final economic landing. The recovery is predominantly centred in the US currently, but China is stabilising while the Eurozone is decelerating at a slow pace.
• The inflation signal was set for negative inflation surprises, but now the data underpinning our inflation indicator shows an uptrend: inflation surprises could remain negative but be less frequent.
• Monetary policy is set to remain neutral wherever we look. This should notably be the case for the Fed, which is increasingly expected to communicate it has already delivered what was necessary in terms of the hikes needed to tame inflation. It just needs to hold those rates for longer so that the economic landing can happen.

 

The current environment described by our indicators could be summarised in one sentence: growth will stabilise as disinflation continues its journey along a bumpy road, while central banks will keep rates higher for the foreseeable future. Such a phase has a name in business-cycle analysis: a late cycle.

 

Figure 1: Growth, Inflation and Monetary Policy Nowcasters

Source: Bloomberg, LOIM

 

You’ve seen late cycles already

Situations similar to that of today have been observed in past cycles, although not since 2009. Figure 2 shows three different time series – Fed rates, oil prices and the US manufacturing ISM as a measure of manufacturing activity – with three late-cycle periods, which all occurred before three recessions: the 1990 investment-driven recession, the 2001 tech-bubble recession and, finally, the housing market-driven 2008 recession. During these three occasions, interesting and (more importantly) consistent patterns were observed:

• Central bank rates remained stable or declined over these periods. For instance, in Q4 2008, the Fed had already decided to cut rates, as it did in 1989 or Q1 2001. Late cycles seem to be occasions where rates are kept high and steady or cut rather than hiked.
• Year-over-year variations in oil prices were positive during all three episodes. Annual growth rates were in the region of 40%, marking large and persistent price deviations and were all indicative of rising commodity inflation. And yet, each time, the Fed made the choice to ignore commodity inflation and cut rates.
• Finally, during all three periods, ISM readings remained above 50% (indicating growth) for a portion of that time. Even more striking, the US ISM staged a remarkable recovery that extended from August 1989 until January 1990, even as the US entered into a recession – sound familiar?

 

The three charts in Figure 2 showcase a couple of feats that seem odd at first glance, but could ultimately be indicative of late-cycle periods. Commodity inflation rises but central banks decide to do nothing about it. Manufacturing recoveries occur between a slowdown and a recession. Finally, rates eventually start to decline –  which could be a sign that the central bank’s job is done –  triggering a recession, the worst phase of any cycle for long-only investors. Late-cycles usually lead into a recession and that point, so far, remains undisputed by history.

 

Figure 2: Late-cycle evolutions in Fed rates, commodities and manufacturing ISM

Source: Bloomberg, LOIM

 

The case for a soft landing

Does this mark the end of the soft-landing narrative? Is a genuine recession about to knock at the door in the near future? To use a classic economist’s answer, that depends, specifically on the behaviour of real rates. Since 21st August, 10-year real rates in the US have repeatedly hit the 2% ceiling, and when markets cross that particular rubicon it can indicate something meaningfully different about the type of economic slowdown that awaits us. The basic economic equation here boils down to a comparison between structural real growth (an expectation of the return on investment) and real rates (an expectation of the cost of investing). In its latest World Economic Outlook, the IMF lifted its real growth forecast for 2028 from 1.7% to 2.1% in light of the investment programmes that are currently being deployed in the US. These programmes reflect different elements but are predominantly tied to the management of the environmental transition and the impact of the ongoing regionalisation of the world’s economies. Central banks slow economies by slowing investment: when the cost of investing outpaces the returns on it, CAPEX collapses and plunges an economy into recession. If trend growth in the US has shifted to 2% then, with real rates at that very same level, the Fed places itself in the situation of delivering a soft landing. Should real rates move higher than 2% then the situation would be radically different, as the cost of capital would surpass the return on it.  This situation is illustrated in Figure 3 which compares the performance of US equities and the spread between trend growth and real rates. That relationship is clearly positive and, with the current level of spreads at zero, it shows a consistent message with what the Fed has been communicating. Whereas, the lower left quadrant of the chart is a definitive hint about what could happen should (1) the IMF be wrong (with lower trend growth in the US) or (2) should the 10-year real yield rise above 2%. In its Jackson Hole communication, the Fed showed that it is very much aware that the risk surrounding a soft landing remains elevated.

 

Figure 3: Real growth in excess of real rates vs S&P 500

Source: Bloomberg, LOIM

 

Has the late cycle started yet?

The difficulty for investors is the assessment of whether we have entered into a late-cycle process or not. Our nowcasting indicators are consistent with this, as already mentioned, but have markets started to price in this risk yet? The short answer is: probably. A late-cycle period tends to match four key features: long rates moving higher, emerging market assets outperforming, commodities rising and a decline in the US dollar. Since the summer, we have increasingly been ticking these boxes: rates and oil prices are notably higher – but this is not enough to be certain. Leaving the question of the dollar aside (see our recent Simply put article), the outperformance of emerging market assets is probably the key point for genuinely pricing a late-cycle period.  An enthusiastic response to the reopening of China at the beginning of the year was later followed by disappointment and this seems to have pushed out the timeline of an emerging market recovery (China is a major piece of this). And yet, our growth indicator is currently stabilising, while the exposure of institutional asset managers to the emerging market theme is progressively building up. Figure 4 shows an increase in the beta of macro hedge funds to equities of late, an increase that has not been obtained by higher exposure to the Nasdaq, quite the opposite: the most “value” part of the equity space has increasingly been revisited by these investors since the start of the summer. In summary, emerging market equities (and credit) today are looking a lot like US stocks did in November last year: under-owned and undervalued, two ingredients that could help them reach higher ground in this late-cycle narrative.

 

Figure 4: Percentiled macro hedge fund exposures to regional equities

Source: Bloomberg, LOIM

 

In a nutshell: Throughout their history, the All Roads funds have added to and cut their exposure to capital markets, as dictated by our Expected Shortfall risk models and our Dynamic Drawdown Management mechanism These have historically helped us meet our risk targets and have also supported our risk-adjusted performance metrics From a cautious stance currently, our strategies could re-risk if rate risks decline or global markets advance

 

Total exposure for our medium-risk profile All Roads fund has historically fluctuated between highs of about 250% and lows of less than 30%. These seemingly drastic shifts are mainly driven by two components within our process, our Expected Shortfall (ES) risk measure and our Dynamic Drawdown Management (DDM) mechanism. They have helped deliver our strategies’ objective of capital protection and improved their risk-adjusted performance. Since the start of 2022, the exposure of the All Roads fund has stayed below 100%, as a result of a significant cash exposure. Having disinvested the portfolio as the drawdown environment materialised and risk levels spiked, the question now is what will trigger a re-risk? This is a topic often discussed with our clients and prospects, and we thought now was an opportune time to dive deeper into our unique process.

 

Risk allocation and risk budgeting have helped over the past 10 years

Risk allocation and risk budgeting are the twin engines driving the core All Roads investment process. Collectively, we may view them as distinct components of our “risk overlay”, but this can be associated with a defensive strategy that negatively weighs on performance over the longer term, similar to an insurance premium. Nevertheless, time and experience have cemented these overlay approaches at the core of our All-Roads strategies, although we have implemented some marginal improvements over the years. Our risk overlay is based on several key rules:

• Risk premia expected shortfall: when our portfolio target risk levels are exceeded, our process mechanically adjusts allocations to remain in line with our objective. The rule is symmetrical and we will also raise exposures when our portfolio is below its target.
• Real carry per risk premia: when risk premia’s carry in excess of short rates is low, our risk model pushes us to disinvest (and invest in the opposite scenario). As a by-product when short rates rise (as is the case today), real carry declines and our market exposure is lower as a consequence. Decreases in short rates produce the opposite effect, with everything else staying equal.
• The performance of our portfolio: when market performance declines, we tend to sell some of our exposure, or re-risk as markets rise. While resembling a momentum strategy, this mechanism is more a function of our risk tolerance and risk-appetite indicators.
• Furthermore, the above risk-based signals may also be compensated or amplified by other components of our process, such as momentum signals or our systematic macro nowcasters.

 

A pay-as-you-go insurance policy

The historical annual performance of our flagship balanced strategy is shown in Figure 1 and has been broken down to show its performance with and without DDM. The chart also shows the DDM’s impact when the strategy has a positive or negative performance. Over the longer run, our analysis shows that our DDM system has cost our balanced strategy about 0.4% per year. In return for that “premium paid”, it mitigated the negative performance during 2015, 2018, 2020 and 2022 – all of which were poor years for multi-asset strategies. What’s more, during these challenging periods the DDM performance contribution was larger than what it detracted from performance during the “good” periods. As shown in figure 1, our DDM added 1.8% per year during negative years for multi-asset strategies but only cost 1.3% during the good years. The cumulative impact is slightly negative, given positive years have outnumbered the negative ones during the period analysed. But we have also experienced rolling periods of 5 years when the cost has been null, if not “negative” (in that it added value). Over the long run, there should be an opportunity cost to using DDM, which investors are well aware of, but importantly for us there is an attractive asymmetry to using it.

 

Figure 1: Historical contribution of Dynamic Drawdown Management to our strategies

Source: Bloomberg, LOIM. Simulated performance results do not reflect actual trading and have inherent limitations. For illustrative purposes only.

 

Being risk-based investors, the key metrics we scrutinise are Sharpe and Calmar Ratios. Figure 2 highlights what our balanced strategy would have achieved with and without drawdown management. The opportunity cost of the DDM is amply rewarded by the decline in volatility and drawdown: our Sharpe Ratio moves from 0.66 to 0.71 while our maximum drawdown is reduced by 4.35%. Similarly, our Calmar Ratio is improved by 0.05. These realised statistics are, from our perspective, a remarkable validation of how appealing DDM can be for risk-based investors. Given our balanced profile objective is to limit 1-year drawdowns to a maximum of 10%, having such an explicit risk control tool is essential and the opportunity cost is compensated by favourable risk-adjusted returns. For instance, in 2022, the cash weighting formed as a consequence of DDM was one of the key sources of performance over the year. Holding cash may have been a drag during the past decade when rates were essentially zero, but the recent rise in global rates makes it less costly. Yet, it must be stressed that a strategy delivering a higher Sharpe or Calmar Ratio can also be sized up to target a slightly higher risk on average. The bottom line is that we can deliver a higher return target with lower risk, or we can stick with a lower risk tolerance while delivering a higher return. This is a key objective for us in the management of our cross-asset strategies.

 

Figure 2: Sharpe Ratio and drawdown with and without DDM

Source: Bloomberg, LOIM

 

The road to re-risking

2022 was an occasion for us to considerably lower our market exposure as short term rates rose, carry declined, performance turned negative and volatility picked up. This drove us to build a cash pocket in our balanced strategy that hovered around 50%. So, what will it take for our market exposures to normalise? Using our balanced strategy as an example, figure 3 shows the outcome of our calculations, breaking down the simulated progression of our market exposure as a function of the normalisation of each of our risk overlay’s three key components, and adding a fourth component that serves as an interaction element between them. The chart is based on the following hypothesis:

• Volatility: the volatility of all risk premia goes back to its respective historical median
• Short-term rates: rates decline by 2%
• Carry: real carry reverts to its long-term average
• Market performance: performance for bonds and equities corresponds with their 5% monthly percentile (respectively 2.6% for bonds and 4.4% for equities).

 

The chart illustrates an important message: most of the volatility contribution, when it comes to re-risking, should be behind us (although Section 2 highlighted that bonds are still experiencing usually high volatility). The bulk of our re-risking, meaning how our market exposure could significantly increase, should come from (1) a decline in short rates, reflecting a normalisation of rate conditions, and (2) a positive performance by a 50/50 bond/equity allocation of about 3.5%. This re-risking could happen quite quickly, should both of these conditions materialise. Quite naturally, if central banks decide to cut rates in reaction to a deteriorating outlook, then the lower rates effect would be compensated by the likely negative market performance and higher volatility. In short, a soft landing (positive market performance and moderate decline in short rates) would be interpreted by our investment process as a reason to add market exposure. A hard landing would be seen as a reason to maintain our cautious stance or even lower it as a function of volatility and market performance.

 

Figure 3: Re-risking mechanics broken down

Source: Bloomberg, LOIM, as at 20 June 2023. For illustrative purposes only.

 

In a nutshell: Timing factors are a workhorse for the financial industry, be it at a strategy level to mitigate drawdowns, to achieve a better risk/return, or at a portfolio level when selecting relevant factors to suit the market environment For valuation strategies, we analyse whether incorporating asymmetry in timing long and short signals differently would help to mitigate the betting-against-trend effect When combining factors, we investigate if a similarity-based approach would enable the identification of previous coherent periods to help in the selection of factors in similar market configurations

 

Classic systematic strategies have been struggling in recent years. While this apparent drought does not diminish the notion of a premium for such strategies, it does emphasise the necessity to adapt factor allocation and/or construction according to the current market environment. In this section, we investigate how timing elements can be incorporated at a strategic or portfolio level.

 

Valuation timing and the asymmetric effect of betting-against-trend

In our last quarterly article, we discussed leveraging Hamilton’s regression filter to derive a model-free valuation metric. The valuation metric, being based on the residuals of a stationary linear system between current prices and lagged prices, means their divergence from equilibrium can be interpreted as a mismatch between the current valuation of an asset and its econometric intrinsic value – something most investors would call a valuation gap. A natural cross-asset valuation strategy might therefore take a long position in the assets with negative residuals (meaning they are cheap compared to their intrinsic value) and short positions in the assets with positive residuals (meaning they are expensive compared to their intrinsic value).

Of course, such a strategy ought not to be directly implemented: markets typically trend for as long as their current value is not far enough from the fundamental value of the asset. This demands that valuation strategies should implicitly bet against trend-following markets. While this explains the typically negative correlation between the two factors, it also underlines the difficulty of building a valuation strategy that does not suffer from the negative carry of betting against the trend. Alleviating this effect amounts to finding a good timing signal for when to take positions in cheap or expensive assets in order to benefit from the convexity stemming from mean reversion, but not entering too early and suffering from the trend. We propose that waiting for the residuals to be significantly far from their 0 level – the level at which the asset is fairly priced, means they are out of a range of δ standard deviation (σ), with δ being small enough that the valuation strategy remains reactive (δ tends to infinity the probability of the signal being larger or smaller than δ times σ, which tends to 0). For example, if the standardised residuals are assumed to be normally distributed, setting the threshold δ = 2 means that the strategy would only be active in 5% of the sample. Interestingly, these activation signals may be set asymmetrically, depending on the sign of the signal (whether it is cheap or expensive).

 

Figure 1: Sharpe Ratio heatmap of the cross-asset valuation strategy as a function of expensive/cheap activation thresholds

Sources: Bloomberg, LOIM

 

Figure 1 shows how this asymmetric effect can help obtain a better risk-return profile by setting the threshold for expensive signals higher than the one for cheap signals. This implies that the strategy will wait longer before taking a short position in an overvalued asset, limiting the effect of betting against a trending asset. However, this effect is not symmetrical as the Sharpe Ratio improves by being more reactive when taking a long position in undervalued assets. Figure 1 illustrates that the Sharpe Ratio improves as a function of the expensive threshold, but not to the same extent as a function of the cheap threshold: hence the importance of looking into this asymmetry.

 

Veni vidi mementi

While timing plays an important role at a single strategy level, the benefit of a timing scheme at a portfolio level is the subject of vivid debate in financial literature. At their core definition, systematic factors should not be timed as they provide a premium for bearing a systematic risk in a cross-section of assets. However, recent disappointing performance, as well as the absence of decorrelation in specific market environments of these systematic strategies, has spurred a new interest in timing factors.

As discussed previously, a natural timing exercise could involve trying to identify common regimes in the distribution of factor returns and conditioning the expected returns to the probability of being in a specific regime. However, the results may be highly dependent on the number of regimes chosen and may not be linked to other variables such as macroeconomic data. We propose a different approach based on a similarity analysis between the current macroeconomic environment and the past. The idea is relatively simple: when calculating the expected returns of a factor or an asset, one should weigh past information from a similar market environment more heavily than past information from totally different market conditions. To calculate this déjà-vu indicator, we transformed exogenous macro-linked variables that represent the current state of financial markets: Equity (SP&500 Index), Volatility (VIX Index), Yield (US 3-month), Oil (Crude), Dollar, Credit, Slope (10-year US tenor – 3-month US tenor) by considering their rank relative to past observations. Then, following Czasonis et al (2023)¹, we calculated the distance between the current observations and the past observations of our seven transformed state variables. This metric allows us to locate any set of observations for our seven exogenous variables in this multivariate distribution; observations with short distances to the current data are more pertinent to our model as the market environments were similar.

 

Figure 2: Distance metrics on specific dates

Sources: Bloomberg, LOIM. Reading note: the charts show a ranking of years based on their proximity to the selected dates. The lower the value, the closer the year to the selected dates.

 

Figure 2 identified the years that had the closest market environment to two different dates: the Covid-19 market crash and the end of August 2023. We noticed that the Covid-19 period is dissociated with periods of strong growth conditions such as 2017 and 2005. Whereas the similarity between 2005 and 2006 and the current market environment highlights its consistency with a late-cycle period. By weighing past data inversely to the distance of current observations, we can then define expected return patterns from historical similarities that favour coherent points in the sample. The vector of expected returns can then be used as a portfolio optimisation exercise.