Delivering high quality alternatives to traditional active management


Delivering high quality alternatives to traditional active management

08 Jun, 2023

In November 2022 we introduced a
Solutions framework for generating alpha.

To recap, the framework revolves around combining individual investment components – or building blocks – into investment solutions. Such solutions have the following advantages:

  1. They can be specifically customised to meet individual investor requirements.
  2. They are flexible, so the component parts can be rescaled or substituted as needs evolve or new objectives arise.
  3. Beta and alpha can be sourced independently. 
  4. Directly sourced beta lowers tracking error.
  5. Independent alpha production enables more consistent alpha and can be accurately assessed.
  6. Investment solutions can benefit from diversification benefits where building blocks are uncorrelated.

Here we expand on our alpha solutions concept introduced previously. We call the solution AlphaPlus, and it combines pure alpha from our Liquid Alternatives Balanced Strategy (LABS) with an investor’s choice of beta exposure. 

AlphaPlus seeks to replicate and potentially improve on the investment experience of investing with an active manager. By combining independently sourced alpha and beta, the solution aims to deliver more consistent alpha over the chosen benchmark. In addition to flexible choice of beta, AlphaPlus can be tailored to meet different alpha targets and/or levels of risk tolerance (tracking error).

AlphaPlus as an alternative to traditional active management

A truism of active management is that even though alpha generation is the objective of active management, it can only ever be achieved as a residual of the active management process. An active manager builds a portfolio of their favourite securities, that they expect to outperform the market. The return on the portfolio is by nature some combination of beta and alpha, but we can only directly observe the sum. Alpha is then calculated as the difference between the portfolio’s return and that of the investment benchmark. It is an ex-post consequence of the portfolio’s performance, but only exists as a residual. 

As a result, it is impossible for active managers to isolate, optimise and calibrate the alpha component in isolation. They can, in some sense, control the overall performance of their portfolio because they choose which securities to include. However, they have precisely zero control over the performance of the benchmark, and by extension the amount of alpha they hope to deliver to investors.

Active managers invoke patience and long-termism, and rightly so because the noise of markets makes it difficult for fundamental security analysis to deliver consistent returns over time, even where the investment process successfully generates long-term value for investors. A corollary of this is that it is hard for investors to adequately judge performance of active managers except over very long timeframes. Good active managers exist, but so do bad ones. The problem for investors is having to wait a long time before they know which they have.

The beauty of our AlphaPlus solution is we can generate alpha and beta independently and directly. Directly sourced beta has no tracking error, so is not a source of active risk. When we generate alpha independently of beta, the investment process can be optimised to deliver more consistent absolute returns, free from many of the constraints facing an active manager (for example tracking error of a pure absolute return process is irrelevant). Equally importantly, uncorrelated returns can be an investment objective for an absolute return fund, which means AlphaPlus solutions can benefit from explicit diversification between beta and alpha sources. 

Best of all, investors are in a much better position to judge performance over shorter timeframes. Clearly, no absolute return strategy can deliver positive excess returns all the time, and as with active management, both good and bad strategies exist. Shorter timeframes in this context still might be counted in years, however, the optimal assessment timeframe should still be much shorter than for a traditional active manager. Done well, the risk of the AlphaPlus solution should also be substantially lower, which lowers the likelihood of highly adverse outcomes.

Measuring AlphaPlus performance

To illustrate the benefits of our AlphaPlus solution framework, we take a hypothetical AlphaPlus MSCI World solution1 and compare it with a selection of active global equity managers2. Remember, AlphaPlus is not a single solution, it’s a flexible solutions framework and the example used here is just one of many possible solutions. However, we believe the example serves to clearly demonstrate the power of the approach.

Chart 1 5y performance of AlphaPlus MSCI World vs benchmark and active manager sample

It is clear from the chart that our AlphaPlus example provides alpha over the benchmark with minimal tracking error. The same can’t be said for some of the active managers. In any given period, there is a subset of active managers which outperform (sometimes substantially), but with very high tracking error (the lines look very different to the benchmark). And, for some of the managers in this sample, a lot of that outperformance is 1) lumpy; and 2) transitory. The “best” managers in the sample as of November 2021 come right back to the pack by the time we get to May 2023, which suggests some of that alpha may have been either random or the result of some equity factor bias (eg a growth bias) which outperforms when markets go up but underperforms when markets go down.

Let’s put some numbers around what we have qualitatively observed from the chart above. The table below outlines some of the performance metrics for our AlphaPlus example and the sample managers. 

Table 1 AlphaPlus MSCI World vs benchmark and active manager sample 5y daily history

Relative to the benchmark, only two of the managers delivered positive alpha over the five-year period. That may well be a function of the subset of managers we selected, so we don’t want to read too much into that. 

More important for us is the tracking error. This quantifies the amount of risk a manager is taking in an effort to deliver alpha. Even for a manager that has historically delivered positive alpha, risk is two-sided. There is a higher probability of a manager with higher tracking error underperforming the benchmark by a given amount than for a manager with lower tracking error. This is important for any investor, but it particularly important for a Super fund subject to the Your Future, Your Super performance test. 

The AlphaPlus example has very low tracking error of 1.2%pa – meaningfully lower than any of the sampled managers. This is a direct consequence of our building blocks framework which sources the beta and alpha directly and independently of the other. The beta building block delivers the investor exactly the benchmark, so no unnecessary tracking error is introduced from there. The only tracking error comes from the volatility of the alpha building block, and we have the flexibility to scale that appropriately to meet either an alpha or tracking error objective.

Another way of looking at it is the worst period of relative performance during the five-year history. This is the worst negative alpha delivered by the manager, shown below in Chart 2. Of the managers sampled, the “least-worst” was -12%. Even a very good manager will sometimes experience devastatingly bad negative alpha, even if only temporarily. 

The AlphaPlus example we show here has an allocation of 20% to LABS. The absolute return on LABS would therefore need to be -60% to deliver a 12% underperformance. We suggest the probability of a -60% return in LABS is vanishingly low, and in any event far lower than the probability of an active manager delivering -12% alpha at some point in time. The upshot is the AlphaPlus approach substantially reduces risk, compared to even the best performing active manager. 

Chart 2 Worst negative alpha any period over past five years

Table 2 Discrete annual alpha

Table 2 shows the alpha for discrete one-year periods. Here we can see the point about the tracking error. Take Manager 3 as an example. In the year to May 2020, Manager 3 outperformed the benchmark by 45%(!) and followed that up in 2021 with a 19.5% outperformance. Undeniably strong numbers. But in the year to May 2022, the manager underperformed by a massive -35%, and referring back to Table 1 you can see that the overall alpha for the full five-year period was essentially zero. That’s a lot a risk to take (and potentially fees to pay) to have nothing to show for it after five years. 
The story is essentially the same across the board, and even those managers that did deliver positive alpha over the full five-year period (Managers 1 and 5) had a pretty bumpy ride along the way. For us, the “lumpy” nature of the some of the manager alphas in Table 2 is an indicator of lower quality alpha, confirmed statistically by the (low) information ratios shown in Table 1. 
We believe high quality active alpha should be:
  1. Positive;
  2. Consistent; and
  3. Commensurate to the risk taken.
Our AlphaPlus framework delivers all three, with the bonus of our alpha-block being uncorrelated (to pretty much everything). Investors gain substantial diversification benefits by building a solution from uncorrelated building blocks. And the best bit is we can work with investors to design customised versions to give them investment solutions that they want/need, not just whichever investment products we happen to be selling.

1 The example simulates the performance of a 20% allocation to LABS combined with an MSCI World (AUD) total return swap funded at BBSW – 10bp, and earns BBSW flat on excess cash.

2 The managers represent a sample taken from the Global Share universe in the Mercer Insights database. 6 of the 8 managers were ranked in the top quartile over the 5y to 31 January 2023.