Dynamic Asset Allocation Based on Valuation and Momentum

Applied to a Global All-Equity Portfolio

∗
James White & Victor Haghani
Elm Partners
March 2020

Summary
There has been considerable research into dynamic global tactical asset allocation (GTAA) strategies driven by simple
measures of Valuation and Momentum, applied to a baseline balanced portfolio of equities and fixed income (see Blitz and
van Vliet 2008, Wang and Kochard 2011, Gnedenko and Yelnik 2014, and Dewey and Haghani 2016). In sum, this research
has found that historically, these Valuation-and-Momentum-driven investment approaches have produced higher returns
and more attractive risk-adjusted returns than a static-weight approach.

However, the above-cited research has been mostly silent on the question of how such an approach could be implemented
and would have performed in the context of an All-Equity portfolio, where the Valuation and Momentum signals would
drive a dynamic allocation between different regional equity markets. This note explores this question. It finds that such a
dynamic approach would have produced higher absolute returns, and higher risk-adjusted returns, than a static approach
would have produced. However, the improvement in simulated historical returns is less than the improvement that Val-
uation and Momentum deliver when applied to a balanced baseline portfolio of equities and fixed income. For example,
simulated historical returns for the All-Equity strategy described below over the past roughly 40 years had about 50% more
variability but only 20% higher returns than did a similar dynamic Global Balanced strategy.

As was the case with our earlier research on this topic (Dewey and Haghani 2016), we examine a form of the strategy
that can be followed by any investor with a simple brokerage account, using low-cost, liquid and widely available ETFs or
index funds, and without use of leverage or shorting. The implementation we explore in this note trades monthly, with
turnover averaging about 50% per annum. It is based on data that is freely available in the public domain.

Background
The approach we take follows the principle that investment sizing should be proportional to forward-looking expected real
returns. Allocating proportionally to expected real returns has deep roots in Samuelson and Merton’s work on Decision-
Making Under Uncertainty starting in the early 1960s1 , along with commonly-used applications such as the Kelly Criterion.
Estimating asset class prospective expected returns using valuation and momentum as the two primary indicators was
popularized by Asness, Moskowitz and Pedersen in their seminal “Value and Momentum Everywhere” paper, and by our
own research including “A Case Study for Using Value and Momentum at the Asset Class Level,” in the Journal of Portfolio
Management, which traced this market phenomenon back to 1925.2

Some of the reasons a valuation-and-momentum-driven approach to asset allocation might be expected to produce
superior risk-adjusted returns are:
• Simple valuation metrics like the Cyclically-Adjusted Earnings Yield of a broad equity market are robust predictors
of the market’s long-term expected real return,
• Momentum is a good medium-term predictor because investors have a tendency to extrapolate future market returns
from recent historical samples, which results in trends generated by “return chasing” behavior, and
∗ Victor is the founder and CIO of Elm Partners and James is Elm’s CEO. This not is not an offer or solicitation to invest. Past returns

are not indicative of future performance. Thank you to Doug Lucas, Bruce Lafranchi, Larry Hilibrand, Arjun Krishnamachar and Vlad
Ragulin for their helpful comments.
1 See our notes here and here for a deeper discussion on this body of work.
2 A working paper version of “Value and Momentum Everywhere” by Asness, Moskowitz and Pedersen was in circulation from 2009 SSRN

here, and presented at the 2010 AFA meeting, although the paper only appeared in the Journal of Finance in 2013. You can find a copy of our
paper here.

Electronic copy available at: https://ssrn.com/abstract=3559426

1
 • Valuation and momentum signals often yield offsetting recommendations, which makes them more effective when used
together than individually.
Furthermore, this investment approach benefits from being implementable using low-cost, tax-efficient, liquid index funds
and ETFs.

Methodology
We start by constructing an All-Equity Baseline portfolio intended to be more representative of the Global Market Portfolio
than the adjusted-market-capitalization indexes of MSCI and FTSE, with a moderate “home bias” preference for US in-
vestors. The process is the same as applied to constructing our Global Balanced Baseline, except in the All-Equity Baseline,
we put a 0% weight on cash and fixed income assets.

The next step is to determine, at each portfolio rebalancing date, the desired deviations from the Baseline weight of each
asset bucket using a combination of a valuation metric and a momentum indicator. This is also done largely in the same
way as we do for our Global Balanced program, with several modifications to fit the context of an All-Equity portfolio.
Most significantly, we normalize the target weights so that they always sum to 100%, and we also impose a constraint that
no bucket’s target weight can be more than twice its Baseline weight.3 We measure momentum for each asset class relative
to the return of the Baseline4 and we’ve reduced the intensity of the value signal to counter the amplification which can
result from the target weight normalization process.5

Implementation
Portfolios, at the time or writing, would have weighted average expense ratios arising from the underlying vehicles in the
range of 0.08% to 0.12% per annum. The table below shows a sample of the instruments which can be used to build
All-Equity portfolios, again for US taxable investors.

3 To normalize the target weights to 100% and impose the 2x constraint, we iteratively impose the constraint and re-normalize until both the

normalizing condition and the constraint are fully satisfied.

4 This is also consistent with the treatment of momentum signals in this early paper on implementing value and momentum in an asset

allocation context: Blitz and Van Vliet, “Global Tactical Cross-Asset Allocation: Applying Value and Momentum Across Asset Classes,” Journal
of Portfolio Management (2008).
5 We set the slope of the value signal in the All-Equity program to 0.5, in contrast to 1 in our Global Balanced programs. You can find more

detail on our Baseline construction and value and momentum overlay here.

Electronic copy available at: https://ssrn.com/abstract=3559426

2
Historical Simulation
We stress that historical data on its own is not sufficient to establish that an investment strategy such as the one outlined
in this note is a good one. Past returns are not indicative of future performance. However, history can lead us to conclude
that a strategy is poor, and it is with that perspective that we look to the past.

The chart below shows a simulated back-test from December 31, 19746 to September 30, 2018. The All-Equity program’s
dynamic value and momentum overlay added 1.3% pa of extra return with no material increase in volatility, sampled
monthly, versus the returns of a static weight All-Equity Baseline portfolio. This is in line with what we would have
expected given that the value and momentum overlay applied to the Global Balanced Baseline increased returns by 2.7% a
year with no material increase in volatility over roughly the same historical period, as we reported in our 2016 Journal of
Portfolio Management paper. This makes sense, as the Balanced portfolio has more flexibility to vary its asset allocation
compared to the All-Equity program.

The variability of the difference in returns between the dynamic and static All-Equity portfolios was 1.9% pa, sampled
monthly, suggesting a Sharpe Ratio for the dynamic versus static strategy of 0.7 over the full period. This is consistent
with the finding that the value and momentum overlay resulted in outperformance 60% of the time to a monthly horizon.
However, as we’d expect, to a longer-term horizon of five years the outperformance was more consistent, occurring in 97%
of 466 rolling five-year periods evaluated at the end of each month. This is shown in the chart below.

6 Many of the historical data series we need for this analysis begin December 31, 1974, particularly the MSCI data series for regional equity

market total returns and earnings.

Electronic copy available at: https://ssrn.com/abstract=3559426

3
 The final chart shows the simulated historical desired asset allocation of the dynamic value and momentum asset
allocation. Portfolio turnover over the period averaged 60% per annum. In practice, we would expect lower turnover is
achievable without materially affecting returns, as investors may not want to rebalance every bucket back to its exact
desired weight.

Back-test Assumptions and Details

In order to simulate returns from 1975, we had to make a number of simplifications to the strategy owing to some data
not being available over the entire history. Going forward, we would update the Baseline weights annually, but in this
back-test we have kept them fixed at today’s Baseline weights. However, we do not think this has a material effect on the
performance of the value and momentum dynamic overlay relative to the Baseline. Also due to limitations of the available
data, the buckets we’ve used for the back-test are not a perfect match of the buckets we will divide the portfolio into going
forward. For example, the back-test does not include some buckets that investors may want to include in the program, such
as low Price/Book and small cap buckets, and on the other hand the back-test has a more granular split than investors may
choose, with Europe split into Europe x-UK and UK and Developed Asia split into Developed Asia x-Japan and Japan.
Return figures include a 0.3% per annum reduction in the dynamic All-Equity strategy for transactions costs, fees and
non-recoverable foreign withholding taxes, 0.2% for the static Baseline and 0.15% for the MSCI All Country World index
to May 31, 2008, and afterwards we use the total return of the iShares ETF ACWI. We assumed rebalancing back to target
each month-end. We used the same Cyclically-Adjusted Earnings Yield centering point of 6% for all regional equity markets.

We stress again that this historical data should not, by itself, be the basis for making an investment decision.

Electronic copy available at: https://ssrn.com/abstract=3559426

4
References and Further Reading
• Asness, C.S., T.J. Moskowitz, and L. Pedersen. “Value and Momentum Everywhere,” Journal of Finance, Vol. 68,
No. 3 (2013), pp. 929-986.
• Blitz, D., and P. van Vliet. “Global Tactical Cross-Asset Allocation: Applying Value and Momentum Across Asset
Classes,” The Journal of Portfolio Management, Vol. 35, No. 1 (2008), pp. 23-38.

• Campbell, J., and R. Shiller. ‘‘The Dividend-Price ratio and Expectations of Future Dividends and Discount Factors.”
Review of Financial Studies, 1 (1988), pp. 195-228.
• Cochrane, J. “The Dog That Did Not Bark: A Defense of Return Predictability.” Review of Financial Studies, Vol.
21, No. 4 (2008), pp. 1533-1575.

• De Grauwe, P., and M. Grimaldi. “Bubbling and Crashing Exchange Rates.” Working Paper, CESifo (Series No.
1045), 2003.
• Dewey, R., and Haghani, V. “A Case Study for Using Value and Momentum at the Asset Class Level.” Journal of
Portfolio Management, volume 42 number 3, (Spring 2016).

• Fama, E.F., and K.R. French. “Business Conditions and Expected Returns on Stocks and Bonds,” Journal of Financial
Economics, 33 (1989), pp. 25-49.
• Fama, E.F., and K.R. French. “Dissecting Anomalies.” Journal of Finance, 63 (2008), p. 1653-1678.
• Ferson, W. E., and C. Harvey. “The Variation of Economic Risk Premiums.” Journal of Political Economy, Vol. 99,
No. 2 (1991), pp. 385-415.

• Gnedenko, B., and I. Yelnik. “Dynamic Risk Allocation with Carry, Value and Momentum.” Working paper, ADG
Capital Management LLP, 2014.

• Jegadeesh, N., and S. Titman. “Returns to Buying Winners and Selling Losers: Implications for Stock Market
Efficiency.” Journal of Finance, Vol. 48, No. 1 (1993), pp. 65-91.
• Kahneman, D., and A. Tversky. “Judgment Under Uncertainty: Heuristics and Biases.” Science, Vol. 185, No. 4157
(1974), pp. 1124-1131.
• Moskowitz, T.J., Y.H. Ooi, and L.H. Pedersen. “Time Series Momentum.” Journal of Financial Economics, Vol. 104,
No. 2 (2012), pp. 228-250.

• Pirrong, C. “Momentum in Futures Markets.” Working paper, University of Houston, 2005.

• Soros, G. The Alchemy of Finance. New York, NY: Simon and Schuster, 1988.
• Wang, P., and L. Kochard. “Using a Z-score Approach to Combine Value and Momentum in Tactical Asset Alloca-
tion.” Working paper, Georgetown University Investment Office, 2011.

Electronic copy available at: https://ssrn.com/abstract=3559426

5