ETF Systematic Strategies: Introducing Momentum

|
Includes: AMJ, BKLN, BWX, CMBS, CWB, EMLC, HYG, IDV, LQD, MBB, MUB, PBP, PCY, PFF, REM, TIP, TLT, VNQ, VNQI, VYM
by: ADS Analytics
Summary

Systematic trading strategies offer a compelling alternative to the more discretionary or benchmark-based investing.

We introduce the Momentum strategy for liquid income ETFs which outperforms an equally-weighted alternative on an absolute and risk-adjusted basis.

The strategy is currently relatively defensive after the December sell-off - overweighting mortgages, TIPS, and Loans.

The hard-won mantra "cut your losers and let your winners run" is well known in investment circles. Unfortunately, it is rarely followed by the majority of investors who are prone to a number of behavioral biases which limit total returns of their portfolios. These biases create persistent patterns in financial markets which can be monetized with algorithmic trading strategies.

Non-discretionary systematic trading strategies are often called "risk factor" or "risk premia" strategies. The number of these factors is very large but the most commonly known ones are Momentum, Value, Carry, and Quality. There is already a number of so-called "smart beta" funds which attempt to follow these factors, particularly, as applied to the US stock market.

What is needed to implement a risk factor strategy is the following:

  • economic rationale - this ensures that there is a fundamental behavioral or economic reason for the existence of the systematic strategy - that it is not just a fleeting data-mined result.
  • financial assets - can be almost anything tradable - in our case, we are looking at 20 benchmark income ETFs.
  • a set of rules - determine which sub-asset of assets to go long and how often they are rebalanced.

In this article, we introduce one of the more common types of such risk factors - Momentum and apply it to a suite of benchmark income ETFs. Our overall aim in this and later articles is to explore which of these risk factors generate consistent alpha beyond a static, equally-weighted portfolio of income funds. Apart from this general overview, our other aim is to create a number of actionable fund portfolios that income investors who are interested in systematic strategies can use in their asset allocations.

Our approach differs somewhat from the more traditional factor-based investing products in that it is cross-asset, it is focused on income funds and it is fully transparent.

There are many methods investors can use for portfolio construction and while a systematic approach is not suitable for everyone, we think a risk factor systematic investment approach has a lot going for it. It has a history of delivering excess returns, it controls the damaging impact of behavioral issues and it avoids the need to forecast asset prices which we think is mostly luck for even the most experienced investors.

Introducing Momentum

The existence of momentum, and more specifically price momentum, suggests that asset prices exhibit trends rather than being completely random. Momentum, as a trading strategy, takes the view that a trending price means an asset that recently appreciated is likely to continue appreciating and vice-versa.

The efficient market hypothesis runs counter to this view since past returns should not predict future returns. Despite this, multiple studies across many markets and time spans have shown the persistence of Momentum.

What are the causes of Momentum?

  • Herding behavior, "fear and greed" and performance chasing tends to exacerbate both positive and negative price trends. Interestingly, the disposition effect (the tendency of investors to sell winners and hold onto losers) should counterbalance extended trends, but it is clearly not powerful enough to fully offset it.
  • Under an overreaction to news as investors tend to overlook news and extrapolate past trends and market narratives.
  • Positive feedback loops between the market and the broader economy where appreciating asset prices leads to greater consumer and business confidence which leads to further asset price appreciations and vice-versa.

Our Fund Universe

Our starting universe of assets is the following 20 funds. Our aims in selecting these assets are to:

  • select income funds
  • end up with a diversified set of exposures
  • pick benchmark funds within their asset type
Fund Name Sector AUM ($bn) Asset Class
AMJ J.P. Morgan Alerian MLP Index ETN MLP 2.6 Equity
BKLN Invesco Senior Loan ETF Loans 5.4 Debt
BWX SPDR Bloomberg Barclays International Treasury Bond ETF IG - Sovereign Debt Ex-US 1.1 Debt
CMBS iShares CMBS ETF CMBS 0.3 Debt
CWB SPDR Bloomberg Barclays Convertible Securities ETF Convertibles 3.8 Debt
EMLC VanEck Vectors J.P. Morgan EM Local Currency Bond ETF EM Sovereign Local Debt 4.6 Debt
HYG iShares iBoxx USD High Yield Corporate Bond ETF High Yield 12.6 Debt
IDV iShares International Select Dividend ETF Equity - Ex-US Developed Dividend 3.8 Equity
LQD iShares iBoxx USD Investment Grade Corporate Bond ETF High Grade 29.6 Debt
MBB iShares MBS Bond ETF Agency 12.3 Debt
MUB iShares National AMT-Free Muni Bond ETF Munis 11.9 Debt
PBP Invesco S&P 500 BuyWrite ETF Covered Call 0.3 Equity
PCY Invesco Emerging Markets Sovereign Debt ETF EM Sovereign USD Debt 3.6 Debt
PFF iShares U.S. Preferred Stock ETF Preferred 13.3 Equity
REM iShares Mortgage Real Estate ETF MREIT 1.2 Equity
TIP iShares TIPS Bond ETF TIPS 21.3 Debt
TLT iShares 20+ Year Treasury Bond ETF Treasury - Long 8.3 Debt
VNQ Vanguard Real Estate ETF REIT - US 27.7 Equity
VNQI Vanguard Global ex-U.S. Real Estate ETF REIT - exUS 5.2 Equity
VYM Vanguard High Dividend Yield ETF Equity - US Dividend 21.3 Equity

This is a chart of fund total returns, normalized to 100 at the inception of the fund.

Source: ADS Analytics LLC, Bloomberg

Getting Started

Our aim in this article is to build a simple momentum trading strategy that we can investigate and refine further. In this section, we sketch out what a simple strategy looks like. These are the rules we follow:

  • Each monthly rebalancing period, assign a momentum score to each fund in our universe
  • The momentum score is calculated as the gross price return over the past year
  • Split our fund universe into four buckets based on the score with the first bucket having the strongest momentum and the last bucket having the weakest momentum
  • Require at least six traded funds to initiate strategy

Typically, risk factors are defined as long/short baskets in order to generate pure alpha without taking beta exposure. In our definition, however, we simply pick the top basket as our strategy because our expectation is that income investors will want beta exposure. This is because a typical investor portfolio takes beta exposure as another source of risk premium compensation. For example, an equity portfolio takes macroeconomic growth and earnings risk and seeks compensation that risk, a bond portfolio takes inflation, and duration risk for which it seeks a risk premium.

What we are doing with our definition of risk factors is add an additional alpha risk premium on top of an already-existing beta risk factor. We do this by holding fewer funds and rebalancing into those funds having the highest scores based on our risk factor definition.

If we run the strategy, this is what the resulting total returns look like for each of the four buckets.

Source: ADS Analytics LLC, Bloomberg

The first thing to notice is that the strategy starts in 2008 - this is where we have a minimum number of funds required to initiate the strategy.

The second thing to notice is that the first bucket (the bucket holding funds with the highest scores each month) total return is well ahead of the other three buckets. This is a good indication that the strategy does appear to capture some alpha. The table below captures the return statistics of the four buckets sorted by PX Rtn (gross price total return). If sorted by total return, bucket 3 slightly outperforms bucket 2; however, when sorted by SR (Sharpe ratio), the buckets are in the right order which is gratifying and suggests we are capturing a real signal.

These results do not include trading costs; however, given our ETF universe comprises some of the most liquid funds, we do not expect the results to be substantially different.

Source: ADS Analytics LLC, Bloomberg

So, even in the very simple model of momentum that we use in this article, we see that we are able to capture a good amount of alpha.

The key question that remains is whether a momentum strategy outperforms an equally-weighed benchmark portfolio. The chart below shows that it does.

Source: ADS Analytics LLC, Bloomberg

Total return statistics tell us that momentum outperforms by 0.7% per year and marginally outperforms on a Sharpe ratio basis as well. However, the key differentiator to us is the drawdown figure where the worst 1-year drawdown of Momentum is almost half that of the benchmark.

Source: ADS Analytics LLC, Bloomberg

Does Momentum Have a Place in Investor Portfolios?

We naturally cannot make individual recommendations in this article, so our comments will be generic. For a strategy to be investable, we think it needs to address two key areas: strategy-specific and behavioral-specific issues.

The reason we think both of these issues matter is because trading strategies and investment outcomes do not always align. And while it would be great if investment outcomes tended to outperform strategies, it is usually the other way around.

Trading strategies do not exist in a vacuum - they need an individual or a team to administer. Since people are liable to behavioral biases or impulses, a strategy that mitigates some of these impulses is more likely to be followed. At the very least when we consider a strategy, we should look at it both from a strategy-specific and an investor or behavioral-specific perspectives. Strategy-specific factors ensure we choose strategies that have sound economic fundamentals, good risk-reward balance, and are robust to varying market environments and behavioral-specific factors mitigate investor behavioral biases and close the gap between strategy returns and actual investment returns.

On the strategy side, the dynamics of the momentum strategy is the following:

  • avoids long drawdowns in individual funds - we can see this in the performance during 2008 where the first bucket completely avoids drawdowns by overweighting government bonds, munis, and agencies. This behavior assumes that some assets (like government bonds in 2008) outperform. If all assets fall in price equally, then the strategy will not outperform an equal-weight benchmark unless it is defined in terms of absolute rather than relative momentum.
  • overweights assets with strong uptrends - this allows the strategy to capture strong performance in individual assets. And while an equally-weighted benchmark will also capture this uptrend, the momentum strategy can overweight outperforming assets and generate higher returns.
  • liable to underperformance in mean-reverting markets - choppy markets are usually not supportive of the strategy as the strategy will buy high and sell low, allocating to outperforming assets that subsequently underperform. The current version of the strategy is particularly vulnerable to this dynamic as the poor performance following December 2018 attests to. In future articles, we will develop alternative versions that mitigate this dynamic.

What are the behavioral implications of the strategy?

  • the momentum strategy dynamic of reallocating away from underperforming funds means that strategy drawdowns are usually more benign than that of an equally-weighted portfolio. This dynamic allows an investor to stay fully invested at market bottoms and mitigate the natural desire to sell at the lows and buy back at higher prices.
  • the strategy mitigates the downside of the disposition effect where investors tend to hang on to losers - the strategy reallocates away from underperforming assets in a controlled way. This also means however that momentum usually trades against valuation considerations - as assets become increasingly cheaper their momentum score reduces and they are more underweight by the strategy. This is not a bad thing necessarily as many assets that fall in price tend to be value traps.

Outside of these considerations specific to Momentum - there are potential downsides common to all strategies:

  • Strategies require additional expenses in the form of transaction costs. In this particular case, we think these costs are minimal given our asset universe comprises highly liquid ETFs.
  • Strategies are more "black-box" than narrative-based or equally-weighted benchmarks. In the case of Momentum - the scores are highly transparent, however, real-life application of this strategy can become highly complex and significantly less transparent. Trading strategies that are less transparent can make it difficult to stick with during long stretches of underperformance.

What's Next?

Starting with this initial foray into systematic strategies we plan to expand on this work across the following dimensions:

  • Improve and develop new factors: in the coming months we plan to improve on this version of Momentum but also develop factors like Value, Low-Volatility, Carry and others
  • Branch out across other assets: in particular, we will expand out factor work to individual closed-end funds as well as closed-end fund sectors
  • Move beyond factors to portfolios: this work will entail combining factors into a single strategy e.g. risk-adjusted carry with a momentum overlay and pulling individual factors together into a single portfolio.

As we continue this work we plan to publish articles on individual assets and factors. These reports will discuss recent performance, current allocations and how individual strategies respond to particular themes or markets.

Conclusion

We think systematic trading strategies offer compelling alternatives to the more common discretionary-based investing or passive benchmarks. A compelling example of this is the Momentum trading strategy which, in the long-term, attempts to mitigate long drawdown periods and overweight outperforming assets. Whether or not such strategies are appropriate to a given investor is not something we can definitely answer - the best we can do is offer our thoughts above and leave each investor to make a decision for themselves. Those who decide that systematic strategies are appropriate for them we encourage to follow us on this journey where we attempt to develop robust way to generate alpha and outperform benchmarks.

Disclaimer: This article is for information purposes only and does not constitute investment advice or an offer or the solicitation of an offer to buy or sell any securities. Past performance is not a guarantee and may not be repeated. Investment strategies are not suitable for everyone and you should always conduct your own research or speak to a financial advisor. Although information in this document has been obtained from sources believed to be reliable, ADS ANALYTICS LLC does not guarantee its accuracy or completeness and accept no liability for any direct or consequential losses arising from its use. ADS ANALYTICS LLC does not provide tax or legal advice. Any such taxpayer should seek advice based on the taxpayer's particular circumstances from an independent tax advisor.

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.