In our last post on Model Portfolios*, we considered looking at Global Equities through the lens of Factors – Momentum, Min Volatility, Value, Quality and Size. This allowed us to achieve diversification, while Dynamic allocation between the factors allowed us to achieve a superior risk return profile – primarily through the ability to use cash to protect the downside in times of low conviction. In this note, we look at how we can construct a different Global Equity Model Portfolio using the same process of Conviction Scores and Dynamic Allocation, but looking instead at a diversified portfolio through the lens of ‘Global Themes’.

The themes we have chosen are largely long term (and thus long duration) growth themes. Our idea is to capture areas that will be subject to long term capital flows, but across a range of themes to allow for (further) diversification. These include:

• Robotics and Automation
• Digitalisation
• Digital Security
• Emerging Market Consumers
• Global Clean Energy
• Healthcare Innovation
• Gold

To these we have subsequently added (this year)

• Global Fintech
• Chinese Internet
• Global Energy
• Energy Infrastructure

As we can see from Chart 1, the Global Theme Portfolio has managed to outperform in down turns – with periods of relative ‘stability’ (q4 2018, q2 2020 and much of this year,) when conviction is low as well as achieving significant outperformance of the MSCI All Country World Index (MSCI ACWI) in upturns, mainly by allocating to performing themes when conviction levels are higher.

Chart 1: Increase Risk as Confidence rises and vice versa.

The process of Dynamic Allocation between the themes is driven by the changes we have in levels of conviction, a process in turn based around our concept of Market Thinking, as described in our previous discussion on Global Equity Factors. How these have changed over time can be seen from the graphic below – the higher the score, the higher the level of conviction – shown darker in the heat map.

Table 1: Conviction Scores

Picking small and esoteric themes without the ability to turn them into an investable product risks reducing the process to an academic exercise only. Accordingly, each of these themes has an externally calculated Global Index, as well as a number of large and liquid ETFs that track that benchmark index. Importantly, we can add themes to the basket even if, as in the example of Global Fintech and Chinese internet, they currently have low conviction scores; they will go into the Model Portfolio as and when the Confidence Score improves. Performance is based on a weekly rebalance at T+1 close. We are not claiming to achieve performance through trading. This is about dynamic allocation between themes.

As with the Global Equity Factors Model Portfolio, our back tests are effectively ‘out of sample’, i.e. we ran simulations of how our Confidence Score based system of Dynamic allocation between the themes would have behaved historically. Here we are limited to the time period the themes have actually existed as not only Indices, but also as tradable ETFs. In addition we only include the 7 themes that we originally selected in 2019, avoiding the hindsight bias of selecting themes – such as energy – after they had already performed well.

Chart 2 thus illustrates the performance of the Portfolio (in Red) against not only the benchmark MSCI All Country World Index (ACWI), but also the sub-components.

Chart 2: Portfolio versus individual Themes

Obviously the standout here is Global Clean Energy, with the blended Portfolio beating all other themes over the 5 year period. An overweight exposure to Clean Energy during 2020 boosted the Portfolio, while an underweight in 2021 helped moderate the downside, driving significantly higher cumulative returns as shown in table 2.

Table 2: Cumulative Returns %

Importantly, as with the Global Equity Factor Model Portfolio, the Global Equity Thematic Portfolio has achieved these returns with considerably lower risk than the benchmark as well, as shown in table 3

Of course the Global Clean Energy Theme was extremely volatile along the way and would not be suitable for any type of ‘Core’ Equity Portfolio on its own. Blended together though, it enabled the Portfolio to deliver greater upside overall than any of the individual themes, while offering considerably lower volatility and downside risk than any of them, indeed, significantly lower risk than the MSCI ACWI benchmark as well.

Table 3: Annualised Return and Risk %

Note that this is not a Market Timing Model per se, rather it is able to utilise its principal active component – the ability to rotate away from low conviction to higher conviction themes – as a form of portfolio protection. When there is low conviction, it has more cash. Note also that, as we saw in 2020, as conviction improved, it was able to go back into certain themes, driving stronger upside.

As with the other model Portfolios, the key to limiting the downside risk is the ability to allocate to cash (ultra short dated bond ETF) at times when the confidence scores are low. When this happens across the board – a form of beta effect – then as chart 3 illustrates, the Model Portfolio can be heavily ‘out of the market’. Obviously this is linked to a ‘risk off’ market, as higher risk premia, in the form of higher discount rates, have the biggest impact on ‘long duration’ growth stocks.

Chart 3: At times of very low conviction, the Portfolio skews heavily to cash.

Tables 4 and 5 show the monthly and annual return profile for the Global Thematic Model Portfolio compared to the benchmark MSCI All Country World Index. Perhaps the main points to note (as also shown in Chart 1) are the stability in 2018 – losing a lot less than the benchmark, and the very sharp outperformance in 2020, reflecting a combination of losing less at the beginning – a relatively early move to higher cash levels – and then getting back into higher conviction areas in the second half.

By contrast 2021 was flat to slightly down for the Model Portfolio, but a better year for the benchmark, but to refer to the earlier tables and Chart, the Model Portfolio still wins over time, by going down less in the downturn and thus over the period from early 2020 has significantly out-performed the benchmark, with – as noted earlier- significantly lower volatility, drawdown and downside risk. To refer back to Charles Ellis and ‘Wining the losers’ game’ – investing is not like professional tennis, where 80% of points are ‘won’, it is far more like amateur tennis, where 80% of points are ‘lost’. Lose less when (not if) you are wrong!

Table 4: Global Thematic Model Portfolio: Monthly Return %

Table 6 simply illustrates the difference between tables 4 and 5 as a form of ‘alpha’. Note, as with the Global Factors Model Portfolio, July was very ‘kind’ to the benchmark, reducing its deficit ytd by almost a third.

Table 6: Monthly and Annual Returns – Alpha (excess return) of Global Thematic Portfolio over benchmark. %

To conclude

We are long terms supporters of both Global Investing and Thematic Investing, but are very aware that the diversification benefits of a Global approach can be lost with too concentrated a portfolio from a Thematic perspective, making it difficult to incorporate Themes into the ‘Core’ part of an Equity Portfolio.

However, we believe that by taking a basket of Themes, albeit all with a long duration growth bias, and dynamically allocating between the themes on a Conviction score basis, we can regain the diversification benefits while maintaining a growth bias. However, we acknowledge that long duration portfolios are traditionally highly sensitive to discount rate and thus vulnerable to ‘risk off’ periods in markets. As such the ‘normal’ 5% (10% if you are lucky) maximum holding of cash offers little protection to underlying investors. Thus by utilising the ability to hold high levels of ‘cash’ when Conviction Scores are low across multiple themes, we believe that the out of sample back tests demonstrate the ability to sensibly manage the greater potential downside risks associated with a growth portfolio. Indeed, by using this approach we appear to have the ability to moderate downside risk without being forced into a structurally low risk/low return portfolio, with outperformance on the upside as well as the downside, combined with lower volatility, drawdowns and downside risk.

(*NB, the terminology used here of Model Portfolios is not to be confused with that used by wealth managers and Independent Financial advisors, i.e. that these are provided as solutions for investors. Market Thinking is not a licensed investment manager and does not make investment recommendations. These are, literally, representations of internal portfolios based upon our models)

‍