Market Thinking

making sense of the narrative

Model Portfolios – Global Bonds

Print Friendly, PDF & Email
At Market Thinking, we run Model Portfolios for two reasons: Firstly, to help us understand the dynamics operating within markets and second to demonstrate the power of Dynamic Allocation between components of any portfolio in order to generate a superior risk/return trade off. Passive investing is cheap in a bull market, but nobody wants to track a benchmark down. Meanwhile, portfolio insurance also tends to become very expensive the minute you need it. We believe that by Dynamically allocating where we are most confident – including cash if we need it – we can achieve better returns by being Active with Passive.

We are making a series of posts about our various model portfolios, starting with Global Bonds. The Bond Portfolio shown here is key to what drove our earlier views on not owning bonds at the end of last year (see The Real Problem with Bonds). Interesting that now, after the worst first half in living memory for Bond Portfolios, the dynamics are starting to look more positive again in the second half, with many of the components moving back to ‘neutral’ from negative in terms of our confidence scores, including, importantly, high yield debt, the highest risk return within the basket we look at, which is climbing in the confidence indicators. Also interesting that towards the end of August they started to turn back down.

The confidence scores themselves are generated according to rules codified from our long term experience of what we like to refer to as Market Thinking:

Three Tribes – Market Thinking

Market Thinking is based on the idea that, at any given time, market prices reflect the differing risk/return expectations of three different ‘Tribes’ – short term traders, medium term asset allocators and long term investors – and that inefficiencies and opportunities arise as the balance of (buying and selling) power shifts between them. Sometimes they are all moving in the same direction, creating powerful directional markets, while at other times they are pulling in different directions. Understanding the behaviours of the different groups – traders are usually leveraged and focussed on absolute return, while asset allocators are usually unleveraged and focus on relative return, for example – is key to analysing potential market behaviours.

Principles of market thinking

The (Market Thinking based) Global Bond Portfolio shown here thus aims to outperform the benchmark Barclays Global Aggregate Bond Index by dynamically allocating and selecting a blend of US long Bonds, Index Linked, Corporate Investment Grade, High Yield Bonds and ultra short term bonds (cash), as well as the benchmark Global Aggregate Bond index itself. We do this by combining our insights based on behavioural finance generated over more than 30 years of investment strategy advice with a series of indicators about the risk/return stance of the different investor groups to establish a series of proprietary confidence or conviction scores for each index. The idea then, is to weight the overall portfolio according to the conviction, the higher the conviction or confidence, the greater the allocation, with the important caveat that, with the lowest level of conviction there is no weighting at all (the use of cash). If there is little conviction across the whole asset class then the allocation is to cash (in effect ultra short dated bonds). This is relatively rare, but it is certainly where we found ourselves early this year.

As can be seen from the performance in Chart 1, this process appears to work extremely well, especially when, as now, the Bull market for fixed income has flipped and the ability to hold ‘cash’ protects the downside. We concede that ‘nobody has ever met a back test they didn’t like’, but that is usually because the model has been created by back testing ‘what worked’, so it’s something of a self-fulfilling argument. By contrast, what we have here is a form of ‘out of sample’ back test; we have applied rules, based on the long established investment principles of Market Thinking and tested ‘what would have happened’ if we had run a basket of passive index tracking ETFs, on a weekly rebalancing system, on this basis. We have also been running these models ‘live’ since 2019.

Chart 1: Model Portfolio Performance v Benchmark

Benchmark: Bloomberg Global Aggregate Index. Source Bloomberg, Market Thinking

*All performance calculated from back-tested model portfolios, but run ‘live’ since 2019

Thus in Chart 1, we can see that the ability to hold different combinations of risk (eg high yield) allowed for outperformance of the Global Aggregate benchmark both in the last period of the bull market, 2020/21, as well as in the bear phase that followed, where the ability to be largely ‘out of the market’ was key. It is this ability to take so called ‘benchmark risk’, ie not have to track something when it isn’t working, that is the essence of applying Market Thinking to portfolios. In effect, take higher risk only when the risk/reward profile looks favourable and vice versa.

As Charles Ellis pointed out in the 1980s, (Winning the Loser’s Game) whereas in Professional Tennis around 80% of points are ‘won’, in Amateur Tennis, 80% are ‘lost’. The trick then for the amateur is to keep the ball in play and let your opponent make a mistake. His view was that investment is similar; you survive (and thrive) by losing less in the downturn, leaving you with more capital to compound in the upturn. (He is also a big believer in the importance of asset allocation being a dominant driver of returns – as, of course, are we).

The Market Thinking process of Dynamic Allocation is a variation on the old Charley Ellis dictum of Investment being ‘A loser’s game’ – similar to Amateur Tennis.

see Charles Ellis, Winning the Loser’s game.

As the table below demonstrates, the cumulative returns from this strategy look very attractive, especially to a benchmarked investor, with both higher returns and lower volatility than the benchmark. Note that the returns are calculated by looking at the performance not of the indices themselves, but of the passive ETFs that track those indices, giving a better representation of returns available to actual investors. Also note that trades are calculated at mid closing price on day T+1, in other words, we are not trying to claim trading skill, this is about dynamic allocation capturing the bigger moves.

Table 1: Cumulative and Annualised Returns

Cumulative Return (%) to end July 2022

Source Market Thinking, Bloomberg

Annualized Return and Risk (%)

Source Market Thinking, Bloomberg Benchmark: Bloomberg Global Aggregate Index
*All performance calculated from back-tested model portfolios

The Process

The essence is the Dynamic Allocation process driven by the conviction scores. These are calculated daily, but unless something dramatic occurs, we only rebalance weekly. This is about capturing medium/long term trends, not short term trading.

The ability to protect the downside through allocations to cash is an obvious, but key component of protecting returns

Chart 2 illustrates that, for long periods of time, the portfolio will have been fully invested (fully shaded in blue). However, for a few periods, and certainly since the start of the year, the weighting has been very heavily skewed towards cash (ultra short duration bonds) and away from the Bond ETFs as shown by the white areas.

Chart 2:The ability to go ultra short duration has been key for 2022

Source, Market Thinking, Bloomberg. End July

Currently, we note that the ‘cash’ element, while still high, is back down to the lowest level all year, supporting our view that the Bond markets are finding some form of ‘base’. This also has read-across to other markets, since it is the low conviction in Bonds and their increased volatility that has been behind much of the wider market instability so far in 2022 (see August Market Thinking). We will look at the corresponding messages from other markets – and other model portfolios -in subsequent posts, but for now, looking at the monthly returns profile for Bonds in Table 4, we can see not only the monthly drivers of the compounding returns shown in table 1 but also the monthly impact of ‘losing less’, especially during 2022.

Table 2 meanwhile gives some insight into the process, illustrating the trailing monthly average Conviction Score for a series of Bond ETFs since 2020. The scores are calculated daily and used to drive weekly rebalancing of the Model Portfolios – the higher the score, the greater the allocation. Clearly as conviction falls across the board the ‘cash’ element rises.

Table 2: Dynamic Bond Allocation over time – Conviction Scores

Source, Market Thinking

Tables 4 and 5 meanwhile shows the monthly and annual returns for both the Model Portfolio and its Benchmark, while table 6 compares the returns from Tables 3 and 4 and shows the excess return of the Global Bond Allocation Portfolio against the Global Bond Benchmark.

Table 4: Monthly and Annual Returns, Global Bond Allocation Portfolio

Source Market Thinking, Bloomberg

Table 5: Monthly and Annual Returns, Global Bond Benchmark

Source Market Thinking, Bloomberg

Table 6: Bond Allocation Fund, Excess Return

Source Market Thinking, Bloomberg

To Conclude.

By back-testing from long established advisory principles that are embedded in what we call Market Thinking, we believe that we can show that a system of dynamic allocation within a basket of Bond ETFs can not only outperform the benchmark Global Aggregate Bond Index, but can do so with lower volatility and lower monthly drawdowns. Moreover, the simple expedient of being able to take considerable benchmark risk at times of high uncertainty through holding ‘cash’ considerably enhances not only the efficiency but also the effectiveness of the strategy.

Share this article