Join our community of smart investors

Understand your risks and get comfortable

Investors Chronicle's 10-asset portfolio aims to deliver better returns - with less risk - than investing in shares alone, using just a handful of ETFs and cash. So far the system has succeeded, but it might be wise to use a blend of active and passive strategies
March 17, 2017

There is a nerdy enjoyment to back-testing investment systems - the satisfaction in finding a set of rules that delivers superior risk-adjusted performance. Once you switch from analysing the past to tracking in real time, however, it can prove disappointing, especially if there is a lot of trading involved and it feels as though you wasted your money. On the surface, this hasn't been the case with Investors Chronicle's 10-asset portfolio, which uses a blend of exchange traded funds (ETFs) and cash to try to achieve optimal returns while diversifying away more risk than a portfolio containing just shares.

We can certainly take encouragement from the past three years, although, conducting a deeper analysis, there are additional risks to consider before concluding that the strategy is sound.

Tracking from February 2014 and rebalancing annually, the equal-weighted ETF/cash portfolio has achieved a real annualised rate of return of 5.01 per cent which, even were you to subtract charges, is comfortably better than the 3.68 per cent annualised real rate of return from just the iShares FTSE 100 ETF (CUKX) over the same period. Thanks to its broad asset allocation our portfolio is also apparently less risky, with standard deviation of monthly returns of 2.14 per cent versus 2.89 per cent for CUKX alone.

We won't really know how robust the 10-asset system is until there is another episode of prolonged distress in financial markets, but so far it has fared better during periods of volatility. The worst peak-to-trough fall was between February and September 2015 - when concerns about China, US rate tapering and tumbling commodities prices led to a loss of 8.22 per cent. This was still less severe than the 12.25 per cent loss for CUKX over the same months. What's more, CUKX suffered more in the short sell-off at the start of 2016.

 

Can we do better by watching for signals based on the 10-month moving average?

When the article '10 moves to a winning portfolio' first appeared, it suggested that we may be able to avoid some of the really big falls in asset prices by selling out when the total return indices tracked below their 10-month moving average. When the value is next above the 10-month moving average, then it's time to buy back in. Back-testing this methodology on the indices our ETFs seek to replicate, it was clear that such a system would indeed have side-stepped the really big losses of the 2007-09 financial crisis, thanks to shoving over 90 per cent of capital into cash during the darkest of those times. The drawback is that in less unsettling conditions it is possible to move in and out of markets too frequently, often missing upside as the market recovers from minor shocks.

To mitigate this unwelcome consequence, for the moving average version of the 10-asset system, we can modify the rule and sell when the total return value of the ETFs is below 95 per cent of the 10-month moving average. This prevents us whipsawing in and out of assets such as developed market equities, but for more volatile assets such as emerging market equities and commodities, there is still a propensity to make trades that with hindsight look unnecessary. Overall, the annualised real return since February 2014 from an ETF/cash portfolio that uses the moving average methodology is 5.24 per cent, and the worst drawdown was just under 6 per cent, which is better than the portfolio that rebalances annually.

 

ETFs used in the 10-asset portfolios

 

Asset classETF (10% weight in rebalancing portfolio)Current weight in 10-mth moving average portfolio
FTSE 100iShares FTSE 100 (CUKX)9.28
FTSE 250Vanguard FTSE 250 (VMID)*9.51
Global developed equityLyxor SG Global Quality Income (SGQL)11.37
Emerging market equityiShares MSCI Emerging Markets (IEEM)10.7
UK giltsLyxor iBoxx Gilts (GIL5)10.07
UK index-linked giltsiShares £ Index-linked Gilts (INXG)11.84
Global real estateHSBC FTSE EPRA/NAREIT Developed (HPRO)12.38
GoldETF Securities Physical Gold (PHGP)0
Commoditiesdbx-trackers commodities (XDBG)8.49
Cashna16.36

 

Source: Investors Chronicle

*We will be using the VMID ETF to invest in the FTSE 250 for the next year. The portfolio performance referenced to in this article for 2014-17 is based on using our previous choice of FTSE 250 ETF, which was the dbx-trackers FTSE 250 ETF (XMCX).

 

The drawbacks of following the moving average rule are that it can get disheartening for investors regularly buying in and out of some of the volatile assets and missing upside. From a risk-management perspective, however, there is plenty to like. The system has a built-in mechanism for adding risk when momentum is positive and reducing the exposure when momentum is negative.

Another advantage is that there is no reliance on the continuation of past correlations in asset price movements to mitigate risks. For example, traditionally the volatile capital component of bond returns (if not held to maturity) could be expected to move in an opposite direction to equities, so any fall in the prices of one asset would be expected to be at least partially offset by increases in the price of the other. In the post-quantitative easing (QE) world this relationship has broken down somewhat as bond yields (which have an inverse relationship with prices) are so low that they have given impetus to stocks. This is because equity prices are to a large extent justified on the basis of the expected excess return over the risk-free yield component of government securities. The worry is that a sell-off in bonds and subsequently higher yields would undermine the current valuation case for equities, with the result that both asset classes deliver a negative return concurrently, thanks to price falls from elevated levels.

 

 

Alternative approaches - risk-efficient portfolios

The broad asset allocation of the 10-asset portfolio means we can benefit from a rich opportunity set of investments. A different approach to either spreading risk across a fairly constant mix or buying in and out of ETFs according to the moving average rule is to select a more concentrated portfolio according to the efficiency of past risk and return observations. This carries the inherent risk of past correlations breaking down, but requires less frequent trading to maintain the system.

One theory is that the best portfolio is a combination of assets that deliver the highest Sharpe ratio. This measure (named after William Sharpe, who was instrumental in developing the capital asset pricing model in the 1960s) is the ratio of the portfolio excess returns (the portfolio returns minus the yield that could be achieved from holding assets with no default risk to maturity), to the standard deviation (risk) of the portfolio.

In our example, we use the nominal yield that could have been earned on cash as the risk-free rate, even though for the cash component of the portfolio we do assume some standard deviation in returns due to changes in the Libor rate. Asking the solver function in Excel to find the combination of our 10 assets that achieved the highest Sharpe ratio, the best we could do is a figure of 0.4025 from three holdings. These were UK Mid-cap shares (31 per cent), global large and mid-cap shares (32 per cent) and index-linked gilts (37 per cent).

Before going into detail about the weaknesses in this approach, of which the short timeframe we are looking back over is one, annually rebalancing a portfolio from February 2014 using just three ETFs (XMCX, SGQL and INXG) would have made an annualised real return of 9.84 per cent, with a maximum peak-to-trough drawdown of less than 5 per cent. This simplified portfolio is ostensibly a 60:40 equities to bonds split, but with interesting tilts within the two asset classes.

 

Understand the products and allow for hidden risks we can't quantify

Although the three-year performance is impressive, we don't believe this is a sensible portfolio allocation. That is not to say that the three ETFs we are using here are bad, rather there are factors that are either unknown or not fully quantified over the short timeframe of our analysis. Therefore it is just not prudent to opt for such a concentrated strategy with these three products alone.

The first hidden risk we cannot quantify (which applies to the whole 10-asset system), is how exchange traded funds will behave in a serious financial crash. Again, these are often excellent products, which give investors liquid, transparent and cost-effective access to a multitude of markets and asset classes, but nearly all of the ETFs we have used in our portfolios were launched post-2009. So, despite a few spikes of volatility, the behaviour of these products can only be judged against the backdrop of long-running bull markets in several asset classes, stoked by unprecedented central bank policy decisions. In many instances, ETFs are the best and cheapest way to make tactical asset allocations, but can we really know how, in a crisis, they will track underlying indices in less liquid assets?

In the case of the three-asset system a third of the portfolio is invested in a synthetically replicating ETF for the overseas equity allocation. The SGQL fund does not own the underlying securities that it tracks, instead matching the returns of the index using swap derivatives and placing collateral with a third party to protect investors' capital from counterparty risk. This structure affords added liquidity, replicating the SG Quality Income Index and, over the three years of our analysis, the ETF has achieved an annualised growth rate of 11.45 per cent. The drawback is the unknown risk of how collateral stock will hold its value in another crisis (what would happen in the event of a sell-off in quality bonds?) and whether the liquidity touted as an advantage would be choked off in difficult conditions.

Another concern is that the risk of the long-duration INXG bond ETF is not properly reflected by just looking at the standard deviation of the ETF returns over three years. The underlying bond portfolio has an average duration of 23 years, which is the point at which it effectively matures at the required rate of return. Such a long duration means that the capital value of the ETF is highly sensitive to changes in interest rates. This is less of a concern to income investors with a very long time horizon, but in the short run we would expect the value of the fund to drop dramatically if rates rose significantly. This big underlying risk of interest rates moving higher has been dormant in the period we have studied and any volatility seen in the price of the ETF has been due to general fluctuations in government bond yields. Effectively, for short-term investors this fund is a bet on pension funds and insurers bidding up the price of inflation-protected securities, so it isn't appropriate to form the entirety of our fixed-income allocation on its own.

 

Using the findings of our Sharpe Ratio analysis to inform tilts within a broader asset allocation

In light of the reservations we should have about following the three-asset strategy to the letter, the Sharpe ratio analysis might seem pointless. What it does do, however, is flag up the combination of themes and investing styles that have worked best in the recent past. This is important because it can be our starting point in making an assessment as to the sustainability of risk-adjusted performance.

We have already discounted the suitability of relying entirely on the index-linked bond fund for our fixed-income allocation and the tendency of passive funds to move into higher-duration strategies is definitely a sign that we should seriously consider some actively managed funds in this space. Returns from passive bond funds have been caught up on the rising tide of quantitative easing (QE) since 2009, but we need to take a more nuanced approach, which means a skilled manager carefully monitoring the duration and liquidity of bonds in the fund portfolio.

The equity ETFs have a tilt towards mid-caps overall and the SG index is international in scope. It is also a value index, but uses dividend yields, where the payout is covered and sustainable, to screen for companies. We shouldn't write off synthetic ETF structures per se but, rightly or wrongly, the maximum allocation of around 10 per cent in the 10-asset system instinctively feels more prudent than having a third of capital tied up in this product structure. Again, for a simple diversified portfolio with fewer holdings, a managed fund with a similar global dividend-led value strategy is more appropriate.

For the FTSE 250 allocation, by contrast, the physically replicating ETF from the 10-asset universe is completely appropriate if investors believe the UK economy can continue to perform well, as it has over the past three years. Of course, that may not be the case, but it would be manageable to use our modified moving average rule for just one asset class like this (sell when the index falls below 95 per cent of the 10-month moving average and buy back in when it rises above 100 per cent of the 10-month average). For implementing this type of risk-on, risk-off policy as we've seen, the passive ETF is the most cost-effective and efficient product.

 

Funds to use in a low-maintenance portfolio

A mix of active and passive funds should give us the best balance between returns, expenses, asset volatility and product-specific risks (whether these are proven or unproven it is circumspect to imagine some unknown pitfalls exist). We can use the allocations that our ETF Sharpe ratio analysis threw up as a starting point, but applying judgment to try to anticipate any risks that our short-term study does not adequately reflect.

Firstly, in the fixed-income space, we don't think that having 40 per cent of your portfolio in such a long-duration fund is sensible. Instead, the allocation should be split between managed bond funds, the ETF we have used for shorter-dated UK government bonds and, in appreciation of the fact that bond prices look expensive across the board, we cut the overall allocation from 40 to 30 per cent. From the Investors Chronicle Top 100 Funds list, we've gone with the Fidelity Strategic Bond Fund (GB00B469J896), which has a shorter effective duration of 4.50 years, according to MorningStar. The fund mandate does include index-linked bonds and we are happier to rely on the judgment of an expert rather than investing passively. The other managed fund is the MI TwentyFour Dynamic Bond Fund (GB00B57TXN82) - another strategic fund, but with more of a weighting to asset-backed securities, which should provide further diversification.

In the case of our equities allocation, instead of using an ETF for the international exposure, we are going with the Fidelity Global Dividend Fund (GB00B7778087). It isn't totally fixated on income, but yield is an important criteria nonetheless and gives the fund similar quality-value characteristics to the SG Index that the passive ETF tracked. International in scope, the fund gives large-cap exposure to North America, continental Europe and Japan, as well as the UK.

We have a spare 10 per cent, given our watering down of the fixed-income position, so we have chosen First State Global Listed Infrastructure fund (GB00B24HK556) as a play on the expected increase in government investment (nearly 50 per cent of exposure is to the US, where President Trump is expected to spend heavily on infrastructure). The yield is not huge, so this fund is not exactly a bond proxy, and the fund buys shares in companies so there will be positive correlation with stock markets. This said, the holdings in utilities companies or industrials with long government contracts ought to display defensive qualities that will make our now 70:30 equities-to-bonds portfolio more robust.

As discussed, we're happy using the 10-month rule and investing with a passive ETF for UK mid-caps. Also, this tactic means we will be fully invested when the momentum for equities is positive, but will have a disciplined method for taking a sizeable cash position in a down period. Since the 10-asset portfolio was first implemented in 2014, however, Investors Chronicle has reviewed the best ETFs for tracking major asset classes and, in keeping with our 2016 Top 50 ETFs selection, we are shifting to the Vanguard FTSE 250 ETF (VMID) as our product of choice.

 

Staying invested and getting comfortable versus a constant dynamic approach

So, discarding the three-ETF portfolio as having too many uncertain risks, we have three investment methods. Going forward we will continue to follow the annually rebalancing 10-asset ETF portfolio and the more dynamic 10-month average method. For very active investors, the 10-month rules on cash management should continue to be a very effective way to sidestep losses. For those who like to take more of a step back, then the 10-asset ETF portfolio is broader in scope than our mixture of active and passive funds. There may be some added benefit to the concentrated approach some of our active managers take and, within the fixed-income holdings, it is reassuring to have experts managing complex factors. Either way, both of these strategies are simple to implement.